Producer Dynamics: New Evidence from Micro Data (National Bureau of Economic Research Studies in Income and Wealth, Vol 68)

Producer Dynamics Studies in Income and Wealth Volume 68 National Bureau of Economic Research Conference on Research...

Author: Timothy Dunne | J. Bradford Jensen | Mark J. Roberts

57 downloads 735 Views 3MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Producer Dynamics

Studies in Income and Wealth Volume 68

National Bureau of Economic Research Conference on Research in Income and Wealth

Producer Dynamics New Evidence from Micro Data

Edited by

Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts

The University of Chicago Press Chicago and London

TIMOTHY DUNNE is a senior economic advisor in the Research Department at the Federal Reserve Bank of Cleveland. J. BRADFORD JENSEN is a senior fellow at the Peterson Institute for International Economics and an associate professor at the McDonough School of Business at Georgetown University and a research associate of the NBER. MARK J. ROBERTS is professor of economics at Pennsylvania State University and a research associate of the NBER.

The University of Chicago Press, Chicago 60637 The University of Chicago Press, Ltd., London © 2009 by the National Bureau of Economic Research All rights reserved. Published 2009 Printed in the United States of America 18 17 16 15 14 13 12 11 10 09 1 2 3 4 5 ISBN-13: 978-0-226-17256-9 (cloth) ISBN-10: 0-226-17256-2 (cloth)

Library of Congress Cataloging-in-Publication Data Producer dynamics : new evidence from micro data / edited by Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts p. cm. — (Studies in income and wealth ; v. 68) Includes index. “This volume contains revised versions of most of the papers presented at the Conference on Research in Income and Wealth entitled ‘Producer Dynamics: New Evidence from Micro Data,’ held in Bethesda, Maryland, on April 8–9, 2005”—Preface. ISBN-13: 978-0-226-17256-9 (cloth : alk. paper) ISBN-10: 0-226-17256-2 (cloth : alk. paper) 1. Commerce— Econometric models—Congresses. 2. Industrial productivity— Econometric models—Congresses. I. Dunne, Timothy. II. Jensen, J. Bradford. III. Roberts, Mark J. HF1008.P77 2009 338.501'5195—dc22 2008009629

o The paper used in this publication meets the minimum requirements of the American National Standard for Information Sciences— Permanence of Paper for Printed Library Materials, ANSI Z39.48-1992.

National Bureau of Economic Research Oﬃcers Elizabeth E. Bailey, chairman John S. Clarkeson, vice-chairman Martin Feldstein, president and chief executive oﬃcer Susan Colligan, vice president for administration and budget and corporate secretary

Robert Mednick, treasurer Kelly Horak, controller and assistant corporate secretary Gerardine Johnson, assistant corporate secretary

Directors at Large Peter C. Aldrich Elizabeth E. Bailey Richard B. Berner John H. Biggs John S. Clarkeson Don R. Conlan Kathleen B. Cooper Charles H. Dallara

George C. Eads Jessica P. Einhorn Martin Feldstein Roger W. Ferguson, Jr. Jacob A. Frenkel Judith M. Gueron Robert S. Hamada Karen N. Horn

John Lipsky Laurence H. Meyer Michael H. Moskow Alicia H. Munnell Rudolph A. Oswald Robert T. Parry Marina v. N. Whitman Martin B. Zimmerman

Directors by University Appointment George Akerlof, California, Berkeley Jagdish Bhagwati, Columbia Glen G. Cain, Wisconsin Ray C. Fair, Yale Franklin Fisher, Massachusetts Institute of Technology Mark Grinblatt, California, Los Angeles Saul H. Hymans, Michigan Marjorie B. McElroy, Duke

Joel Mokyr, Northwestern Andrew Postlewaite, Pennsylvania Uwe E. Reinhardt, Princeton Nathan Rosenberg, Stanford Craig Swan, Minnesota David B. Yoﬃe, Harvard Arnold Zellner (Director Emeritus), Chicago

Directors by Appointment of Other Organizations Jean-Paul Chavas, American Agricultural Economics Association Gail D. Fosler, The Conference Board Martin Gruber, American Finance Association Timothy W. Guinnane, Economic History Association Arthur B. Kennickell, American Statistical Association Thea Lee, American Federation of Labor and Congress of Industrial Organizations

William W. Lewis, Committee for Economic Development Robert Mednick, American Institute of Certified Public Accountants Angelo Melino, Canadian Economics Association Harvey Rosenblum, National Association for Business Economics John J. Siegfried, American Economic Association

Directors Emeriti Andrew Brimmer Carl F. Christ George Hatsopoulos Lawrence R. Klein

Franklin A. Lindsay Paul W. McCracken Peter G. Peterson Richard N. Rosett

Eli Shapiro Arnold Zellner

Relation of the Directors to the Work and Publications of the National Bureau of Economic Research 1. The object of the NBER is to ascertain and present to the economics profession, and to the public more generally, important economic facts and their interpretation in a scientific manner without policy recommendations. The Board of Directors is charged with the responsibility of ensuring that the work of the NBER is carried on in strict conformity with this object. 2. The President shall establish an internal review process to ensure that book manuscripts proposed for publication DO NOT contain policy recommendations. This shall apply both to the proceedings of conferences and to manuscripts by a single author or by one or more coauthors but shall not apply to authors of comments at NBER conferences who are not NBER aﬃliates. 3. No book manuscript reporting research shall be published by the NBER until the President has sent to each member of the Board a notice that a manuscript is recommended for publication and that in the President’s opinion it is suitable for publication in accordance with the above principles of the NBER. Such notification will include a table of contents and an abstract or summary of the manuscript’s content, a list of contributors if applicable, and a response form for use by Directors who desire a copy of the manuscript for review. Each manuscript shall contain a summary drawing attention to the nature and treatment of the problem studied and the main conclusions reached. 4. No volume shall be published until forty-five days have elapsed from the above notification of intention to publish it. During this period a copy shall be sent to any Director requesting it, and if any Director objects to publication on the grounds that the manuscript contains policy recommendations, the objection will be presented to the author(s) or editor(s). In case of dispute, all members of the Board shall be notified, and the President shall appoint an ad hoc committee of the Board to decide the matter; thirty days additional shall be granted for this purpose. 5. The President shall present annually to the Board a report describing the internal manuscript review process, any objections made by Directors before publication or by anyone after publication, any disputes about such matters, and how they were handled. 6. Publications of the NBER issued for informational purposes concerning the work of the Bureau, or issued to inform the public of the activities at the Bureau, including but not limited to the NBER Digest and Reporter, shall be consistent with the object stated in paragraph 1. They shall contain a specific disclaimer noting that they have not passed through the review procedures required in this resolution. The Executive Committee of the Board is charged with the review of all such publications from time to time. 7. NBER working papers and manuscripts distributed on the Bureau’s web site are not deemed to be publications for the purpose of this resolution, but they shall be consistent with the object stated in paragraph 1. Working papers shall contain a specific disclaimer noting that they have not passed through the review procedures required in this resolution. The NBER’s web site shall contain a similar disclaimer. The President shall establish an internal review process to ensure that the working papers and the web site do not contain policy recommendations, and shall report annually to the Board on this process and any concerns raised in connection with it. 8. Unless otherwise determined by the Board or exempted by the terms of paragraphs 6 and 7, a copy of this resolution shall be printed in each NBER publication as described in paragraph 2 above.

Contents

Prefatory Note

xi

Introduction Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts

1

I. CROSS-COUNTRY COMPARISON OF PRODUCER DYNAMICS 1. Measuring and Analyzing Cross-Country Diﬀerences in Firm Dynamics Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta Comment: Timothy Dunne

15

II. EMPLOYMENT DYNAMICS 2. Studying the Labor Market with the Job Openings and Labor Turnover Survey R. Jason Faberman 3. What Can We Learn About Firm Recruitment from the Job Openings and Labor Turnover Survey? Éva Nagypál 4. Business Employment Dynamics Richard L. Clayton and James R. Spletzer

83

109 125

vii

viii

Contents

5. The LEHD Infrastructure Files and the Creation of the Quarterly Workforce Indicators John M. Abowd, Bryce E. Stephens, Lars Vilhuber, Fredrik Andersson, Kevin L. McKinney, Marc Roemer, and Simon Woodcock Comment: Katharine G. Abraham

149

III. SECTOR STUDIES OF PRODUCER TURNOVER 6. The Role of Retail Chains: National, Regional, and Industry Results Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda Comment: Jeﬀrey R. Campbell

237

7. Entry, Exit, and Labor Productivity in U.K. Retailing: Evidence from Micro Data Jonathan Haskel and Raﬀaella Sadun

271

8. The Dynamics of Market Structure and Market Size in Two Health Services Industries Timothy Dunne, Shawn D. Klimek, Mark J. Roberts, and Daniel Yi Xu

303

9. Measuring the Dynamics of Young and Small Businesses: Integrating the Employer and Nonemployer Universes Steven J. Davis, John Haltiwanger, Ronald S. Jarmin, C. J. Krizan, Javier Miranda, Alfred Nucci, and Kristin Sandusky Comment: Thomas J. Holmes 10. Producer Dynamics in Agriculture: Empirical Evidence Mary Clare Ahearn, Penni Korb, and Jet Yee Comment: Spiro E. Stefanou

329

369

IV. EMPLOYER-EMPLOYEE DYNAMICS 11. Ownership Change, Productivity, and Human Capital: New Evidence from Matched Employer-Employee Data Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom Comment: Judith K. Hellerstein

397

Contents

12. The Link between Human Capital, Mass Layoﬀs, and Firm Deaths John M. Abowd, Kevin L. McKinney, and Lars Vilhuber 13. The Role of Fringe Benefits in Employer and Workforce Dynamics Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson Comment: Dan A. Black

ix

447

473

V. PRODUCER DYNAMICS IN INTERNATIONAL MARKETS 14. Importers, Exporters, and Multinationals: A Portrait of Firms in the U.S. that Trade Goods Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott Comment: James Harrigan 15. The Impact of Trade on Plant Scale, Production-Run Length, and Diversification John Baldwin and Wulong Gu Comment: James Tybout Contributors Author Index Subject Index

513

557

597 601 607

Prefatory Note

This volume contains revised versions of most of the papers presented at the Conference on Research in Income and Wealth entitled “Producer Dynamics: New Evidence from Micro Data,” held in Bethesda, Maryland, on April 8–9, 2005. Funds for the Conference on Research in Income and Wealth are supplied by the Bureau of Economic Analysis, the Bureau of Labor Statistics, the Census Bureau, the Federal Reserve Board, Statistics of Income/IRS, and Statistics Canada. We are indebted to these organizations for their support. We thank Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts, who served as conference organizers and editors of the volume. We also thank the staﬀ of the NBER for their assistance in organizing the conference and preparing this volume. Executive Committee, January 2007 John M. Abowd Susanto Basu Ernst R. Berndt Carol A. Corrado Robert C. Feenstra John Greenlees John C. Haltiwanger Michael J. Harper Charles R. Hulten, chair

Ronald Jarmin J. Bradford Jensen Lawrence F. Katz J. Steven Landefeld Brent R. Moulton Thomas B. Petska Mark J. Roberts Matthew Shapiro David W. Wilcox

xi

Introduction Producer Dynamics Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts

The process of firm entry, growth, and exit has always been an integral part of the mechanism of resource reallocation in a market economy. Spurred by developments in micro data construction by government statistical agencies and access to these data by researchers, the empirical analysis of producer dynamics has become a major focus of economic research over the last fifteen years. The crucial input that has made the empirical study of producer dynamics possible is comprehensive longitudinal micro data that allow researchers to track new firms over their lifetimes. Using these data for a large number of countries, researchers have identified links between the characteristics of firms and their subsequent success or failure that provide a better understanding of the sources of firm and worker dynamics and their implications for the long-run growth and performance of a market economy. In recognition of its importance to public policymaking, the primary U.S. statistical agencies—the Census Bureau and the Bureau of Labor Statistics (BLS)—have recently begun to produce oﬃcial statistics that measure the dynamic movements of firms in and out of business and workers in and out of jobs. The development of new data resources and empirical facts on producer dynamics has impacted many research fields in economics including industrial organization, labor, growth, macro, and international trade. Since the initial measurement studies of Dunne, Roberts, and Samuelson (1988, 1989), the longitudinal data sets have been exploited by industrial organiTimothy Dunne is a senior economic advisor in the Research Department of the Federal Reserve Bank of Cleveland. J. Bradford Jensen is an associate professor in the McDonough School of Business, Georgetown University, and a research associate of the National Bureau of Economic Research. Mark J. Roberts is a professor of economics at Pennsylvania State University, and a research associate of the National Bureau of Economic Research.

1

2

Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts

zation economists to study the competitive eﬀects of producer turnover and why the process diﬀers across industries and time periods. Building on the work of Davis, Haltiwanger, and Schuh (1996), an enormous literature, both empirical and theoretical, has developed in labor and macro economics to measure and explain the gross employment flows due to job creation and job destruction by firms. Bailey, Hulten, and Campbell (1992) and Griliches and Regev (1995) show how sectoral and industry productivity gains can be traced back to productivity diﬀerences that exist at the micro level combined with the exit of low productivity firms and the entry and growth of higher productivity firms. The intertemporal pattern of lumpy plant-level investment present in the micro data (Doms and Dunne 1998) have been analyzed by macro economists as a source of aggregate investment fluctuations (Caballero, Engel, and Haltiwanger 1995). International economists have also studied how trade flows are shaped by both the growth of existing exporters and importers and the flows of producers in and out of international markets (Bernard and Jensen 1995; Bernard, Eaton, Jensen, and Kortum 2003; Das, Roberts and Tybout 2007). None of these lines of research could have developed without the use of firm- and plant-level longitudinal surveys and censuses conducted by government statistical agencies. This volume is the result of a two-day conference in April 2005 devoted to the measurement and explanation of producer dynamics. The meeting was sponsored by the Conference on Research in Income and Wealth (CRIW) and had as its primary goal, as do all CRIW conferences, encouraging interaction between the statistical agencies that are developing the longitudinal firm-level data series, and data users from academics, government, and the private sector. The timing was motivated by the development of several new micro-data sets that provide much more comprehensive coverage of U.S. firms and plants than has been previously available. These include: the Longitudinal Business Database (LBD), constructed at the Center for Economic Studies of the Census Bureau; the Business Employment Dynamics (BED) and Job Openings and Labor Turnover Survey (JOLTS) programs at the Bureau of Labor Statistics; and the matched worker-employer database under construction as part of the Longitudinal Employer-Household Dynamics (LEHD) program at the Census Bureau. These data sets are also the major source of new government statistics on producer and employment dynamics. The BLS produces quarterly statistics on gross job gains and gross job losses for private sector employers through its BED program. The BED is constructed from state unemployment insurance records and provides job creation and job destruction statistics by industry, state, and firm size. Complementing the BED job flows data is worker flow data from the relatively new JOLTS program at BLS. Job Openings and Labor Turnover Survey (JOLTS) is a monthly survey of roughly 16,000 nonfarm establishments that measures job vacancies, new

Introduction

3

hires, and separations. It provides higher frequency data that is more timely than the BED but has less geographic detail. The Census Bureau has also institutionalized a program to construct Quarterly Workforce Indicators (QWI) that summarize employment dynamics in local labor markets and are based on the data from the LEHD project. The QWI reports information on both job and worker flows down to the county level and for detailed industries. These are some of the first government statistics to summarize the dynamic patterns of producer-level adjustment in the U.S. economy. These programs are discussed in chapters in this volume. Each of these data sets is a significant new resource, and together they are going to be a major source of our knowledge of producer dynamics in the U.S. economy for at least the next decade. The chapters also include analysis of longitudinal micro data sets from Canada, the OECD countries, Sweden, and the United Kingdom, which provide useful sources for comparison. Other chapters in this volume are designed to disseminate information on these data sources within the research community, provide a reference source for future users of the data, and present new empirical results that extend the measurement and analysis of producer dynamics to sectors of the economy beyond manufacturing, to a broader range of countries, to firm transitions in international markets, and to linkages between firm and worker turnover. All of these are areas where empirical research on producer dynamics is in its infancy. Cross-Country Comparison of Producer Dynamics The volume is divided into five sections based on the type of data that is used in each chapter. The first section reports the results of a project undertaken by Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta to develop comparable cross-country data on firm entry, exit, and turnover. Over the last decade there has been tremendous eﬀort to develop statistics on producer dynamics in many countries, but the eﬀorts are largely independent and reflect idiosyncracies in each countries’ data collection process. The usual problems of comparability that exist when analyzing data from diﬀerent countries’ national accounts are compounded when cross-country comparisons on firm turnover are attempted. The unit of analysis (establishment, firm, line of business), the population of firms under study, the definitions of entry and exit, and the variables used to measure entry and exit (producer counts, employment, sales) often diﬀer across countries. In this chapter, the authors report results from a large research project bringing together researchers from twenty-four countries to standardize data definitions and construct comparable statistics on producer dynamics and productivity. Even with the extensive coordination in the construction of the individual country data, measurement diﬀerences still exist, but the authors are

4

Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts

able to draw some broad comparisons. Looking across countries, annual entry and exit rates are substantial in most cases, averaging between 5 percent and 10 percent of the business population. Somewhat surprisingly, countries that are often believed to have rigidities that impede the development of new businesses have relatively high entry and exit rates. For example, France has entry and exit rates quite similar to the United States and Canada. Eastern European countries, in general, are found to have extensive restructuring in their business populations with very high entry rates of new businesses. Less than one-half of new firms survive through their seventh year in most countries studied. Bartelsman, Haltiwanger, and Scarpetta also document the micro-level sources of productivity growth through a set of productivity decompositions. The goal is to identify the relative contributions to labor productivity growth of entering and exiting firms, within firm productivity changes, and between firm reallocation in shares. The findings show that the within-firm changes in productivity and net entry are the major sources of labor productivity growth in most countries. Employment Dynamics A second significant line of data construction and research over the last decade has focused on the patterns of employment dynamics—the movement of workers in and out of jobs and the creation and destruction of employment positions. The Bureau of Labor Statistics and Census Bureau have made considerable progress in developing new data surveys and augmenting existing data programs to produce information on employment dynamics. The second section of this volume contains four chapters that discuss and utilize these new data series. The first two chapters, by Jason Faberman and Éva Nagypál, report on a new BLS survey: Job Openings and Labor Turnover Survey (JOLTS). This survey provides information on labor force dynamics by surveying establishments monthly about vacancies, hiring, and separations. Faberman presents an overview of the JOLTS program and an analysis of establishment-level vacancies and employment flows. A particular strength of the JOLTS program is that it produces a new series on job vacancies that is much less idiosyncratic than the help-wanted indices used in previous studies. The micro data also show a much more complex adjustment process than that observed in the aggregate series. Cyclical variation in separations is driven more by shifts in the distribution of growth rates of establishments than by changes in the average separation rates across the distribution of establishments. Establishments that are contracting or expanding have greater hiring and separation rates than stable establishments. While these patterns in labor turnover are related systematically to local unemployment conditions, diﬀerences in state unemployment rates explain little of the overall variation in establishment-level employment flows.

Introduction

5

Nagypál raises a number of important measurement issues in the JOLTS data. First, she identifies the large discrepancy in employment growth over the period 2000 to 2004 between JOLTS and other BLS establishment surveys. It is due primarily to the understatement of separations in the earliest JOLTS surveys. Over time, BLS has made improvements to the survey to reduce the problem, though Nagypál reports that at the industry-level large discrepancies remain. Nagypál also discusses a number of measurement issues with regard to vacancies. Job Openings and Labor Turnover Survey (JOLTS) only measures vacancies that are to be filled within a thirty-day period. Hiring environments where vacancy posting substantially precede the actual hiring date are excluded from the data. Job Openings and Labor Turnover Survey (JOLTS) also measures vacancies as a stock of positions and misses short-duration vacancies. The magnitude of the measurement error will be larger in sectors and time periods with high arrival rates of job candidates. Each of these issues will cause a systematic understatement of vacancies in the data. The final step of the author’s analysis estimates a simple matching function from the JOLTS data, and she finds that the matching function diﬀers markedly across industries. The third chapter in this section, by Richard Clayton and James Spletzer, provides an overview of the Business Employment Dynamics (BED) database at the BLS. This database has been constructed from state unemployment insurance records through the Quarterly Census of Employment and Wages (QCEW) program. The BED contains data on virtually all private business establishments in the United States from 1992 onwards and produces statistics on quarterly job creation and destruction due to plant openings, expansions, contractions, and closings. Clayton and Spletzer provide a detailed analysis of job creation and destruction in the 2001 recession and the subsequent years. Job destruction initially rose sharply but then fell back to prerecession levels quickly. Alternatively, the drop in job creation persisted. To better understand the sources of job flows during the 2001 recessions, the authors examine the underlying micro changes and find that most of the decline in employment is due to concentrated increases in job creation and destruction in a relatively small number of establishments. The final chapter in this section, by John Abowd, Bryce Stephens, Lars Vilhuber, Frederik Andersson, Kevin McKinney, Marc Roemer, and Simon Woodcock, presents detailed documentation of the LEHD data sources and the methods used to construct the QWI. The QWI represents a major new statistical initiative by the Census Bureau to construct job flow statistics for county and MSA-level labor markets. The data underlying the QWI are drawn from the LEHD database, which combines employer and employee information. The QWI reports statistics on job creation, new hires, separations, and earnings for all employees and new hires disaggregated by industry, geography, and worker characteristics such as age and

6

Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts

gender. This level of detail is far greater than any other currently available statistics on employment flows. In addition to the creation of the underlying micro data set, the QWI project has invested heavily in the development of disclosure techniques that preserve the confidentiality of the data but allow for the release of very disaggregated summary statistics. Overall, the chapter provides a valuable reference source for users of the QWI and the LEHD. Sector Studies of Producer Turnover The earliest studies of producer dynamics focus on the manufacturing sector because this tends to be the sector that is most consistently surveyed and has the best micro data on producers. In addition, almost all studies of producer dynamics use data on firms with paid employees and ignore nonemployer firms. The third section of this volume contains chapters that look beyond the traditional data sources, focusing on producer dynamics in retailing, service industries, and agriculture, and extending the measurement of producer dynamics to the nonemployer segment of the business universe. The chapter by Ron Jarmin, Shawn Klimek, and Javier Miranda documents the entry and exit of establishments and firms in the U.S. retail sector based on analysis of the Census Bureau’s newly developed Longitudinal Business Database (LBD). The LBD covers all establishments with at least one paid employee and all industrial sectors of the economy for the period 1976 through 2005. While the LBD contains limited information on the establishment’s characteristics and activities, it can be linked with other Census Bureau establishment data, which considerably enhances the scope and depth of the available information. This new data has the potential to enhance our understanding of such topics as job creation and destruction, firm turnover, the life cycle of establishments, and changes in the industrial structure of the U.S. economy. In their chapter, Jarmin, Klimek, and Miranda document the overall changes in employment and the number of establishments in the retail sector focusing on diﬀerences between chain stores and individually-owned establishments. Over the last several decades there has been a fundamental shift in the organizational structure of the industry, with a significant expansion of stores owned by multi-store firms and a decline of individually-owned stores. The chapter shows that firm turnover has declined over time in most retail industries but diﬀers systematically by market size and ownership structure. Metro areas have the highest producer turnover while rural areas have the lowest turnover. Independently-owned stores experience higher turnover compared to chain stores, but there is little diﬀerence in turnover across diﬀerent types of chain stores—local, regional, or national chains. Continuing with the analysis of the retail sector, Jonathan Haskel and

Introduction

7

Raﬀaella Sadun document producer dynamics and productivity growth in U.K. retailing. Store entry and exit rates are quite high in the United Kingdom, averaging 10 to 15 percent per year over the period 1998 to 2003. These rates are similar across most retail industries with the exception of pharmacy stores, which has much lower rates. The chapter decomposes changes in sectoral productivity between 1998 and 2003 and finds that entry and exit play an important role in accounting for the productivity growth in U.K. retailing. These findings suggest that producer turnover in U.K. retailing enhances productivity by replacing lower productivity exiting firms with higher productivity entering firms. One complication in the U.K. micro data is that the surveys collect information from diﬀerent reporting levels, making it diﬃcult to combine data on firm-level productivity with store-level entry and exit measures. The chapter by Timothy Dunne, Shawn Klimek, Mark Roberts, and Daniel Yi Xu models the entry and exit flows in two medical services industries—dentists and chiropractors—using data for small geographic markets in the United States. They provide some of the first evidence on producer dynamics in healthcare industries using the U.S. Census Bureau’s Census of Services. In the industrial organization literature, researchers have used models of entry to explain diﬀerences in the number of firms across markets of diﬀerent size. While useful for understanding long-run market structure, the models do not explain diﬀerences in entry and exit flows across markets. The authors use a dynamic model that recognizes the diﬀerent costs faced by incumbent producers and potential entrants, and specify entry and exit flow regressions consistent with the dynamic framework. They find an important role for past market structure and the number of potential entrants as determinants of the level of producer turnover; this supports the dynamic framework. A common theme of virtually all papers on producer turnover is that they focus on firms or establishments with paid employees. In the United States in 2000, almost 75 percent of all firms (15.4 million out of 20.8 million) had no employees. The chapter by Steven Davis, John Haltiwanger, Ron Jarmin, C. J. Krizan, Javier Miranda, Al Nucci, and Kristin Sandusky represents the first eﬀort to measure producer dynamics for this segment of the business population. A key contribution of the project is that it not only documents producer turnover in the nonemployer segment but also identifies transitions between nonemployer and employer firms. Of the 2.3 million employer businesses in their industry sample in 2000, 11 percent can be linked to a nonemployer business that existed between 1992 and 2000. However, it is rare for a nonemployer firm to become an employer firm. Of the almost 7.4 million nonemployer firms in the industries under study in 1994, only 3 percent became an employer firm by 1997. This data source provides enormous potential for a better understanding of the evolution of young and small firms. For example, the study shows that fluctuations in

8

Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts

nonemployer size, measured in terms of revenue, from year to year are much larger for nonemployer firms than employer firms; but as nonemployer firms age and grow, the volatility of their revenue stream declines. This latter finding is similar to age and volatility patterns observed in the employer data. The final chapter on sectoral patterns of producer turnover provides the first statistics on producer dynamics in the agriculture sector. Mary Ahearn, Penni Korb, and Jet Yee utilize data from the U.S. Census of Agricultures from 1978 to 1997 to provide new statistics on the entry, exit, and growth of farms. Entry and exit rates are measured by the number of farms, the volume of sales, and the acreage of land under cultivation. The main patterns show considerable turnover of farms over the entire period. Average annual entry and exit rates appear higher than those reported for other sectors of the U.S. economy, especially when one considers weighted measures such as sales or acreage share of entering and exiting farms. In their data, entry and exit include the sale and purchase of farmland; thus, entry and exit statistics can reflect sales or leases of an existing farm and thus does not directly correspond to the movement of land in or out of agricultural production. The authors document patterns of producer dynamics that diﬀer from those found in many manufacturing sectors. Older cohorts have relatively low shares of sales and land and there is only a slight increase in the average size of farms as a cohort ages. Within a cohort, small continuing farms actually tend to shrink over time while larger farms have higher growth. This is opposite the patterns one sees in manufacturing, where there is a strong inverse relationship between growth and size conditioning on a firm remaining in business. Employer-Employee Dynamics A broader view of labor market dynamics integrates producer decisions to enter and exit production and expand and contract the employment positions within a firm, with the worker’s decisions to move in and out of existing employment positions. Both are a potentially important source of labor market flows, but the data requirements to measure these separate sources are demanding. Section four of this volume includes chapters that use linked employer and employee data to present a more detailed picture of worker turnover and the human capital present at a workplace. The first chapter, by Don Siegel, Ken Simon, and Tomas Lindstrom, uses matched employer-employee data from the Swedish manufacturing sector to study how corporate ownership changes aﬀect the performance of the firm and the composition of the firm’s workforce. They find that plants undergoing ownership changes have lower labor and total factor productivity prior to the ownership change but that productivity rises to industry norms after the ownership change. The composition of the plant’s

Introduction

9

workforce also changes. Average age, worker experience, and the percent of college-educated employees rise in the plant after a change in ownership, while the share of women falls slightly. Overall, it appears that in the downsizing of these operations, plants shed workers with short job tenures and these are more likely to be younger and female workers. The two remaining chapters in this section use data from the LEHD. John Abowd, Kevin McKinney, and Lars Vilhuber utilize the LEHD to measure the human capital embodied in a firm’s workforce and relate it to the performance of the firm. The authors construct an index of human capital for each worker in a plant by decomposing the employee’s wage into a firm component and a worker component. For each employer, they construct the distribution of human capital for the workforce and examine if this is correlated with the probability a firm undergoes a mass layoﬀ or closes. They find that mass layoﬀs and firm failure are much more likely in firms with a large proportion of low human capital workers. Finally, firms that do not fail generally upgrade the human capital of their workforce. Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson leverage the richness of the LEHD data set by augmenting the LEHD with publicly-available data on employee benefits (collected in IRS Form 5500) oﬀered by diﬀerent companies. This allows them to combine measures of worker characteristics and employer characteristics with information on nonwage compensation, including health plans and defined benefit and defined contribution pension plans. The authors show that the level of benefits oﬀered by the firm is negatively correlated with employee turnover, but this largely reflects underlying diﬀerences in the human capital of the workforce. Firms that oﬀer benefits have higher-skilled workers and these skilled workers have lower turnover rates. Moreover, firms oﬀering benefits have higher labor productivity and are more likely to survive, even after controlling for worker and firm characteristics and wage compensation. Producer Dynamics in International Markets Research in international trade has recognized the importance of firm heterogeneity in productivity and profitability as factors that aﬀect the decision to participate in international markets. The final section of this volume contains two chapters that use micro data to study transitions of firms into and out of import and export markets. Andy Bernard, Brad Jensen, and Peter Schott develop a new data set on import and export activity of U.S. firms, and provide a set of stylized facts on participation patterns. The authors combine transaction-level records of imports collected by U.S. Customs with firm-level exports collected by the U.S. Census Bureau for the period 1993 to 2000. They link these observations with the LBD, which will allow researchers to incorporate a large set of firm characteristics from the LBD into the analysis of micro trade flows. An important feature of

10

Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts

these data is that it allows the authors to distinguish between related party and arms-length transactions. Thus, one can measure the flow of crossborder goods within multinational firms. The chapter documents a number of striking patterns. The fraction of firms engaged in trade is small but growing—two to three percent of the total number of firms in the United States. However, these importing/exporting firms are large, accounting for approximately 40 percent of private sector employment in the United States. Ninety percent of import and export activity involves multinational firms and related-party transactions make up approximately one-half of imports and one-third of exports. The authors also analyze the employment dynamics of firms involved in trade. Firms that export had higher employment growth than nonexporters, and firms that entered the export or import market between 1993 and 2000 experienced very high employment growth rates. Alternatively, firms that stopped exporting and/or importing suﬀered declines in employment. The final chapter, by John Baldwin and Wulong Gu, explores the impact of trade liberalization resulting from the 1989 Canada-U.S. Free Trade Agreement on the decision of Canadian manufacturers to enter or exit the export market. Using a theoretical framework in which producers diﬀer in their productivity, they characterize the determinants of a firm’s decision to enter the export market. Firms export depending on their relative cost advantage—the most eﬃcient firms produce for the domestic and export markets, less eﬃcient firms produce only for the domestic market, and the least eﬃcient firms close. Trade liberalization increases the size of the market and results in greater firm specialization. Exporting firms withdraw from some product markets and expand the volume of output in their remaining products. Nonexporting firms, however, do not benefit from this increase in market size and instead face increased competition and, on average, become smaller. Using micro data for Canadian manufacturing plants, the authors test the predictions of the model using tariﬀ rate changes as a measure of trade liberalization. They find that nonexporting firms reduce the number of product lines and decrease plant size in response to a lowering of tariﬀs. They find that exporting firms become more specialized and larger but these changes are not strongly correlated with industry-specific tariﬀ reductions. Conclusion Producer dynamics can be viewed from many perspectives, including movements of firms or plants in or out of production, transitions between diﬀerent geographic or product markets, or shifts of an entrepreneur from self-employment to employer status. Regardless of the focus, the decisions of firms to change the nature of their production is an important mechanism contributing to the reallocation of resources. The chapters in this vol-

Introduction

11

ume have developed and utilized a number of important micro data sets that provide a window on this diverse set of producer transitions. A recent report by the National Research Council, “Understanding Business Dynamics: An Integrated Data System for America’s Future” (Haltiwanger, Lynch, and Mackie 2007), presents a blueprint for further development of the U.S. data system to allow more accurate and timely measurement of the dynamic forces at work in the economy. Among the recommendations in the report is one to encourage the interaction of the statistical agencies that create the producer micro data and the researchers from academia, business, and government that analyze it. The chapters in this volume provide ample evidence of the knowledge that can be gained by researchers working with the statistical agencies to document and analyze the dynamic process of firm entry, growth, and exit. In many areas, particularly the service sector, the nonemployer universe, and the international arena, measurement issues have only recently begun to be addressed and much work remains. The recent eﬀorts of the U.S. statistical agencies to produce new statistics that document the flows of workers and firms in a timely and consistent way is another important avenue through which knowledge of the process of producer dynamics is expanding. The Business Employment Dynamics (BED) program at the BLS and the Quarterly Workforce Indicators (QWI) program at the Census Bureau are providing timely information on dynamic aspects of the U.S. economy that complement the traditional focus on aggregate statistics at a point in time. Still, the series are relatively new and a better understanding of the economic forces that drive the dynamic patterns in the producer data is needed.

References Bailey, M. N., C. Hulten, and D. Campbell. 1992. Productivity dynamics in manufacturing plants. Brookings Papers on Economic Activity: Microeconomics: 187–249. Bernard, A., J. Eaton, J. B. Jensen, and S. S. Kortum. 2003. Plants and productivity in international trade. American Economic Review 93 (4): 1268–90. Bernard, A., and J. B. Jensen. 1995. Exporters, jobs, and wages in U.S. manufacturing, 1976–1987. Brookings Papers on Economic Activity: Microeconomics: 67–119. Caballero, R., E. M. R. A. Engel, and J. Haltiwanger. 1995. Plant-level adjustment and aggregate investment dynamics. Brookings Papers on Economic Activity 1955 (2): 1–54. Das, S., M. J. Roberts, and J. R. Tybout. 2007. Market entry costs, producer heterogeneity, and export dynamics. Econometrica 75 (3): 837–73. Davis, S. J., J. Haltiwanger, and S. Schuh. 1996. Job creation and destruction. Cambridge, MA: The MIT Press. Doms, M., and T. Dunne. 1998. Capital adjustment patterns in manufacturing plants. Review of Economic Dynamics (1): 409–29.

12

Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts

Dunne, T., M. J. Roberts, and L. Samuelson. 1988. Patterns of firm entry and exit in U.S. manufacturing industries. The RAND Journal of Economics 19 (4): 495– 515. ———. 1989. The growth and failure of U.S. manufacturing plants. The Quarterly Journal of Economics 104 (4): 671–98. Griliches, Z., and H. Regev. 1995. Firm productivity in Israeli industry 1979–1988. Journal of Econometrics 65:175–203. Haltiwanger, J., L. M. Lynch, and C. Mackie, eds. Understanding Business Dynamics: An Integrated Data System for America’s Future. 2007. Report to the Committee on National Statistics, The National Research Council. Washington, D.C.: The National Academies Press.

I

Cross-Country Comparison of Producer Dynamics

1 Measuring and Analyzing Cross-Country Diﬀerences in Firm Dynamics Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

1.1 Introduction Cross-country comparisons and analysis of firm dynamics are inherently interesting, but also inherently diﬃcult. Such comparisons are important because they provide insights into the eﬃciency with which resources are allocated in the economy and its eﬀects on output, productivity, and employment. Empirical evidence for developed economies shows that healthy market economies typically exhibit a high pace of churning of outputs and inputs across businesses.1 Moreover, the evidence shows that this churning is productivity enhancing as outputs and inputs are being reallocated from less productive to more productive businesses. These findings raise the question as to whether diﬀerences in economic performance across countries can be accounted for by diﬀerences in the eﬃciency of the churning process across countries. A closely related question is whether certain regulations and institutions in diﬀerent markets aﬀect the churning Eric Bartelsman is a professor of economics at the VU Amsterdam, and a research fellow of the Tinbergen Institute. John Haltiwanger is a professor of economics at the University of Maryland, and a research associate of the National Bureau of Economic Research. Stefano Scarpetta is Head of the Country Studies Division III in the Economics Department of the Organization for Economic Cooperation and Development (OECD), and a Research Fellow and Deputy Program Director at IZA. We would like to thank our discussant Timothy Dunne and the participants to the 2005 NBER Conference on Research on Income and Wealth on “Producer Dynamics: New Evidence from Micro Data” (April, Washington, D.C.) for insightful and useful comments. We are grateful to the World Bank for financial support of this project and to Karin Bouwmeester, Helena Schweiger, and Victor Sulla for excellent research assistance. The views expressed in this chapter are those of the authors and should not be held to represent those of the institutions of aﬃliation. 1. See Caves (1998), Bartelsman and Doms (2000); Ahn (2000); and Foster, Haltiwanger, and Krizan (2001) for surveys.

15

16

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

process in a manner that slows the reallocation of resources towards more productive uses. In this chapter, we adopt the working hypothesis that policy and institutions aﬀecting the business climate (broadly defined) may have important implications for the magnitude but also the eﬀectiveness of firm dynamics and resource reallocation. While individual country studies can provide important insights into this issue by looking at within-country variation in performance of sectors or individual firms, another way to test the hypothesis is to link firm performance across countries that diﬀer in their regulatory and policy settings. This strategy, however, involves an ongoing measurement and research agenda to develop comparable measures of firm dynamics across countries that can be directly related to business climate conditions. The interest in this type of analyses is rapidly spreading beyond the industrial countries and involves many developing and emerging economies that are struggling with regulatory reforms to stimulate private investment and productivity growth.2 In principle, using firm-level data to assess cross-country diﬀerences in economic performance is attractive. It avoids some of the problems typically aﬀecting macro analyses. For example, interpreting the observed persistent diﬀerences in income per capita across countries or even growth rates of GDP and productivity has been a challenge for a long time. This is not because of the lack of candidate explanations, but rather because of the overwhelming number of possible factors. As such, the finding of a statistically significant correlation between cross-country diﬀerences in economic performance and any possible policy, institutional, or structural variable is fraught with problems of interpretation given the (many) omitted variables.3 It is misleading to argue that the firm dynamics approach overcomes the omitted variable and associated unobserved heterogeneity problems that aﬄict macro analyses. But the firm-level approach potentially oﬀers a tighter theoretical link between specific institutional measures and relevant outcomes. For example, indicators of firm dynamics allow testing whether regulatory distortions that impinge on entry costs indeed aﬀect the pace and nature of firm entry. In practice, cross-country comparisons of measures of firm dynamics suﬀer from significant definitional and measurement problems. Changes at the firm level take diﬀerent forms, and no single indicator is likely to capture this complexity in a way that can be related to all regulatory or insti2. A number of works have recently explored the role of firm dynamics for productivity and growth in developing and emerging economies. They include Eslava et al. (2004); Roberts and Tybout (1997); Aw, Chung, and Roberts (2003); and Brown and Earle (2004). 3. This explains the diﬃculty in obtaining robust empirical results from macro growth regressions (e.g., Barro and Sala-i-Martin [1995] and Doppelhofer, Miller, and Sala-i-Martin [2004]). See Scarpetta (2004) for recent attempts at estimating macro growth regressions for the OECD countries.

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

17

tutional issues in a meaningful way. This conceptual problem is often confounded by measurement problems induced by cross-country diﬀerences in coverage, unit of observation, classification of activity, and data quality. The combination of conceptual and measurement problems can be illustrated by considering the most basic measures of firm dynamics—the rates of firm entry and exit—and comparing them across countries with indicators of economic performance. Figure 1.1 shows the rank ordering of countries according to gross firm turnover (entry plus exit rates) and GDP per capita levels and growth rates.4 We consider these rank orderings for a set of countries for which we have harmonized statistics on firm turnover rates. The rank ordering of GDP per capita levels and growth rates are quite plausible. But while the rough order of magnitude reported in figure 1.1 for firm turnover is also reasonable, the rank ordering across countries of the firm turnover rates is more diﬃcult to interpret. Relatively high firm turnover rates are observed both in countries with high income levels and/or high growth rates as well as in poorer and/or slow-growth countries (and vice versa).5 We argue in the chapter that this is because it is not clear whether there is an unequivocal relationship between firm turnover and economic performance, but also because there could be measurement problems that aﬀect the cross-country comparisons of firm turnover. In this chapter, we review the measurement and analytical challenges of handling firm-level data so as to provide a user’s guide on how to construct and how to compare measures of firm dynamics across countries. In broad terms, we have three basic messages. First, it is very important to make every attempt to harmonize the indicators of firm dynamics by imposing the same metadata requirements and aggregation methods on the raw firmlevel data. Second, while harmonization is necessary, it is far from suﬃcient. As illustrated in figure 1.1, some core cross-country comparisons will not only be problematic because of remaining possible measurement problems, but also because some firm-level indicators cannot be unequivocally linked to better or worse economic performance. However, the third message is that there are ways to overcome at least some of the measurement problems. While the details diﬀer depending on the type of measure and question of interest, we show that by using measurement or analytic methods that amount to some form of diﬀerence-in-diﬀerence approach, the problems we identify can be significantly reduced. The chapter proceeds as follows. In section 1.2, we describe our distrib4. This chapter draws on firm-level indicators for a sample of countries that participated in the distributed micro-data analysis. We made every attempt to harmonize the statistics by providing detailed protocols and programs to researchers with access to the confidential micro-level data sets in their countries. The indicators in our database are built up from these (confidential) micro-level sources. 5. Note that the correlation between firm turnover and the GDP/capita measures is low (– 0.22 using GDP per capita levels; and 0.18 using GDP per capita growth).

18

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

A

B

Fig. 1.1 Comparisons of GDP per capita growth and firm turnover: A, GDP per capita and firm turnover, 1996; B, GDP per capita growth and firm turnover, 1990–2003 Note: For transition economies (Estonia, Latvia, Hungary, Romania, and Slovenia) 1996–2003

uted micro-data analysis that we advocate and have used in our crosscountry comparison project. As we make clear, the problems illustrated in figure 1.1 are much worse if there is not an attempt at harmonization. In section 1.3, we describe the data collected in the World Bank and Organization for Economic Cooperation and Development (OECD) firm-level projects. In section 1.4, we provide a canonical representation of the possible sources of measurement problems in using firm-level statistics for comparative purposes. We use this representation to help us think through what types of comparisons are likely to be robust and what types of comparisons will not be robust to measurement error of diﬀerent types. Sections 1.5 and 1.6 explore cross-country comparisons that can be made us-

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

19

ing our harmonized data. We present basic facts from these data, which are of interest in their own right, but discuss them in light of the measurement challenges we have described. In section 1.5, we first present the distribution of firms by size; we then document the magnitude and key features of firm dynamics (entry and exit of firms) and, finally, we study post-entry performance of diﬀerent cohorts of new firms. In section 1.6 we analyze the eﬀectiveness of creative destruction for productivity growth. We distinguish between the productivity contribution coming from the process of creative destruction (entry and exit of firms) to that stemming from withinfirm eﬃciency improvements and reallocation of resources across incumbents. In the last section, 1.7, we draw conclusions and discuss next steps for this approach for cross-country comparisons of firm dynamics. In this discussion, we present some ideas of the dos and don’ts of working with firm-level data for purposes of constructing and analyzing cross-country measures of firm dynamics. 1.2 Distributed Micro-Data Analysis The indicators used for cross-country comparisons in this chapter have been collected by a network of researchers with access to (confidential) micro data. The construction of the indicators in each country followed a common methodology and led to a cross-country harmonized metadata. This collection method is an attempt at the generation of comparable cross-country statistics. It is part of a long tradition of statistical harmonization that has resulted in a wide variety of cross-country sources of economic data, ranging from national accounts information to internationally harmonized surveys. Over the past decades, much institutional eﬀort has been devoted to harmonize national accounts data across countries in order to allow meaningful cross-country comparisons. While the nominal and real indicators of GDP available in each country’s national accounts are generally comparable over time, divergence between exchange rates and purchasing power have often clouded cross-country comparisons. Several sources (including the OECD for its member countries and the World Bank for a larger set of countries) now provide Purchasing Power Parity indicators (PPPs) to convert various expenditure components of GDP into internationally comparable units. Significant eﬀorts have also been made to produce comparable statistics at the sectoral level (e.g., the OECD Structural Analysis database—STAN— the United Nations Industrial Development Organization [UNIDO], and more recently, the EUKLEM databases). While the main underlying sources of these data are sectoral disaggregations from national accounts, other sources such as labor accounts and production statistics are generally used to fill holes. Essentially, these data sets are top down, in that sectoral output and compensation add up to national accounts totals, up to various adjustments (such as owner-occupied housing, etc.). These adjust-

20

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

ments are often not well known, and applied researchers using these data (Bernard and Jones 1996a; Griﬃth, Redding, and Van Reenen 2000; Nicoletti and Scarpetta 2003) generally take these sectoral data as given. Comparable micro-level data sets are even less frequent, and comparability issues are generally more severe. However, several attempts have been made to harmonize household panel surveys and labor force surveys to improve cross-country comparability. The Luxembourg Income Study, the European Community Household Panel (ECHP), or the Integrated Public Use Microdata Series (IPUMS) data sets are all examples of this eﬀort to compile and use comparable micro data sets. Standardized Labor Force Surveys, following International Labour Organization (ILO) definitions, are also available for a large set of countries. At the firm-level, no comprehensive survey exists with data for multiple countries, nor are there international data sets that contain micro-level data for comprehensive samples of firms.6 The EU Statistical Oﬃce (EUROSTAT) has recently made a major eﬀort in assembling a data set on firm demographics for a number of EU member countries, using common definitions and classifications.7 The data collection is based on existing data sources and some idiosyncrasies in the data cannot be eliminated. At the same time, the World Bank has been collecting data on relatively small samples of firms in more than fifty developing and emerging economies worldwide (World Bank 2004).8 These data are often limited to a few industries and do not allow tracking firm dynamics. 1.2.1 How to Collect and Compare Firm-Level Data A data set consisting of stacked micro-level data sets from multiple countries will contain the necessary information lacking from either singlecountry micro data sets or multiple-country sectoral data sets. Unfortunately, owing to the legal requirement of maintaining confidentiality of firms’ responses in many countries, micro data sets from individual countries cannot be stacked for analysis. Creating public use data from the underlying sources is a possible workaround for disseminating otherwise confidential data. For firm-level data, a public-use data set made through 6. Commercially published data sets such as Compustat or Amadeus provide panel data on financial information of publicly traded corporations. 7. See EUROSTAT (2004). The Eurostat data focus on eleven European countries over the period 1997–2000, and considers all firms, including those with zero employees. 8. This data collection is based on Investment Climate Assessment (ICA) surveys, including information on firm characteristics and performance as well as perceptions of managers about the regulatory and political environment in which they operate. A discussion of the advantages and disadvantages of the alternative approaches as well as the relationship on key findings from the ICA data set versus the type of firm-level data used here is provided in Haltiwanger and Schweiger (2004). Recent works that have used the ICA data to study firm performance include Bastos and Nasir (2004), Dollar, Hallward-Driemeier, and Mengistae (2003), Hallward-Driemeier, Wallsten, and Xu (2003).

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

21

randomization or micro-aggregation is often not feasible without the loss of necessary information. Another possible work-around is to create a data set consisting of results from single-country studies that become the input for a meta-analysis. For example, a collection of results from single-country studies on the link between Information and Communication Technology (ICT) and growth at the firm-level were presented in a recent volume of the OECD (2004). However, the combination of results of analyses from single-country studies will not provide a solution if the focus of the analysis is not identical or if methodologies diﬀer significantly. In the World Bank and OECD firm-level projects, a hybrid approach was followed that mitigates many of the discussed problems. Given the impossibility of stacking together firm-level data for diﬀerent countries, a common protocol was used to extract from the raw country data set of detailed indicators. The protocol was designed after face-to-face meetings with country experts and collection of metadata describing each country’s data sets.9 The protocol was then run on micro-level data sets in each country separately by experienced researchers. The decentralized output was combined and provided the information necessary for the cross-country analysis. This approach was first developed for the OECD firm-level growth project and is known as distributed micro-data analysis (Bartelsman 2004). It requires tighter coordination and less flexibility in research design in each country than for meta-analysis, where the methodology and output may vary across samples.10 The method of distributed micro-data analysis maintains the advantages of multicountry studies with aggregated data because the output provided by each country consists of indicators aggregated to a prespecified level of detail that passes disclosure in all countries. The method also maintains information on behavior of agents residing in micro data because the computed indicators on the ( joint) distribution of variable(s) are designed to capture hypothesized behavior. While not allowing the full flexibility of re9. In addition to the authors of this chapter, the researchers involved in the distributed micro-data analysis network for the various projects are: John Baldwin (Canada); Tor Erickson (Denmark); Seppo Laaksonen, Mika Maliranta, and Satu Nurmi (Finland); Bruno Crépon and Richard Duhautois (France); Thorsten Schank (Germany); Fabiano Schivardi (Italy); Karin Bouwmeester, Ellen Hoogenboom, and Robert Sparrow (the Netherlands); Pedro Portugal Dias (Portugal); Ylva Heden (Sweden); Jonathan Haskel, Matthew Barnes, and Ralf Martin (United Kingdom); Ron Jarmin and Javier Miranda (United States); Gabriel Sánchez (Argentina); Marc Muendler and Adriana Schor (Brazil); Andrea Repetto (Chile); Maurice Kugler (Colombia and Venezuela); David Kaplan (Mexico); John Earle (Hungary and Romania); Mihails Hazans (Latvia); Raul Eamets and Jaan Maaso (Estonia); Mark Roberts (Korea, Indonesia, and Taiwan [China]); Milan Vodopivec (Slovenia). 10. The methodology for the International Wage Flexibility Project (Dickens and Groshen 2003), evolved over time from meta-analysis to a more coordinated system with centralized research protocols, distributed computation, and centralized analysis, and now is very similar to distributed micro-data analysis.

22

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

search design available with multicountry stacked micro data, distributed micro-data analysis provides a skilled researcher the ability to use crosscountry variation to identify behavioral relationships. 1.3 Description of the Data The firm-level project organized by the World Bank involves fourteen countries (Estonia, Hungary, Latvia, Romania, Slovenia, Argentina, Brazil, Chile, Colombia, Mexico, Venezuela, Indonesia, South Korea, and Taiwan [China]) This project complements a previous OECD study that collected—along the same procedure—firm-level data for ten industrial countries: Canada, Denmark, Germany, Finland, France, Italy, the Netherlands, Portugal, United Kingdom, and the United States. Both projects use a common analytical framework that involves the harmonization, to the extent possible, of key concepts (e.g., entry, exit, or the definition of the unit of measurement) and the definition of common methods to compute the indicators. The distributed micro-data analysis was conducted for two separate themes. The first theme focused on firm demographics, and collected indicators such as entry and exit, job flows, size distribution, and firm survival. The second theme gathered indicators of productivity distributions and correlates of productivity. In particular, information was collected on the distribution of labor and/or total factor productivity by industry and year, and on the decomposition of productivity growth into within-firm and reallocation components. Further, information was collected on the averages of firm-level variables by productivity quartile, industry, and year. The key features of the micro-data underlying the analysis are as follows: Unit of observation: Data used tend to conform to the following definition (EUROSTAT 1998): “an organizational unit producing goods or services which benefits from a certain degree of autonomy in decisionmaking, especially for the allocation of its current resources.” Generally, this will be above the establishment level. However, firms that have operating units in multiple countries will have at least one unit counted in each country. Of course, it may well be that the national boundaries that generate a statistical split-up of a firm in fact split a firm in a real sense as well. Also related to the unit of analysis is the issue of mergers and acquisitions. Only in some countries does the business register keep close track of such organizational changes within and between firms. In addition, ownership structures themselves may vary across countries because of tax considerations or other factors that influence how business activities are organized within the structure of defined legal entities. Size threshold: While some registers include even single-person businesses (firms without employees), others omit firms smaller than a certain size,

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

23

usually in terms of the number of employees (businesses without employees), but sometimes in terms of other measures such as sales (as is the case in the data for France). Data used in this study exclude singleperson businesses.11 However, because smaller firms tend to have more volatile firm dynamics, remaining diﬀerences in the threshold across diﬀerent country data sets should be taken into account in the international comparison.12 Period of analysis: Firm-level data are on an annual basis, with varying time spans covered. Sectoral coverage: Special eﬀorts have been made to organize the data along a common industry classification (ISIC Rev.3) that matches the OECD-STAN database. In the panel data sets constructed to generate the tabulations, firms were allocated to one STAN sector that most closely fit their operations over the complete time span. In countries where the data collection by the statistical agency varied across major sectors (e.g., construction, industry, services), a firm that switched between major sectors could not be tracked as a continuing firm but ended up creating an exit in one sector and an entry in another. For industrial and transition economies, the data cover the entire nonagricultural business sector, while for most of Latin America and East Asia data cover the manufacturing sector only. Unresolved data problems: An unresolved problem relates to the artificiality of national boundaries to a business unit. As an example, say that the optimal size of a local activity unit is reached when it serves an area with ten million inhabitants. In smaller nations, one activity unit must be supported by the administrative activities of a business unit. If the EU boundaries were to disappear, the business unit could potentially serve twenty-seven activity units. This geographic consideration may contribute to explain why we observe a larger average firm size in a country like the United States in our sample, although this is not the case in another large country, Brazil. From a policy perspective, this diﬀerence may point towards aligning regulations in a manner that would allow busi11. The share of firms without employees is large in most countries for which data are available (see EUROSTAT 2004). Their inclusion in the analysis of firm demographics is problematic for a number of reasons, however. Zero employee firms may include part-time activities and formally self-employed people who work regular hours on a long-term basis for a sole client, thus appearing more like dependent employees for most purposes. To the extent that people involved in this false self-employment have little intention to expand their business or innovate, they are of limited interest for studies investigating the role of the entrepreneurial process for technological change, employment growth, and economic performance. In some countries/sectors, the amount of false self-employment may be quite sizeable, and possibly depends on diﬀerent regulations aﬀecting hiring and firing costs as well as taxes on labor use. 12. The productivity data are collected at diﬀerent levels of aggregation in diﬀerent countries and very few are able to work at more than one level. A sensitivity analysis of the productivity decompositions suggests, however, that this issue does not significantly aﬀect the results.

24

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

ness units to enjoy transnational scale economies in meeting administrative requirements. Also related to the unit of analysis is the issue of mergers and acquisitions: only in some countries business registers have been keeping track of such organizational changes within and between firms in the most recent years. 1.3.1 The Source of the Data: Firm Demographics The analysis of firm demographics is based on business registers (Canada, Denmark, Finland, Netherlands, United Kingdom, United States, Estonia, Latvia, Romania, and Slovenia), social security databases (Argentina, Germany, Italy, and Mexico), or corporate tax roles (France, Hungary) (table 1.1). Enterprise census data were used for Brazil, Korea, and Taiwan (China), while annual industry surveys—albeit generally not the best source for firm demographics, owing to sampling and reporting issues, were used for Chile, Colombia, and Venezuela. Data for Portugal are drawn from an employment-based register containing information on both establishments and firms, while data for the three East Asian countries are from census of manufacturing firms. All these databases allow firms to be tracked through time because addition or removal of firms from the registers (at least in principle) reflects the actual entry and exit of firms. However, the three to five year frequency of manufacturing census in East Asia precludes computing many of the demographics indicators. 1.3.2 The Source of the Data: Productivity Decompositions The productivity analysis requires information on output, employment, and possibly other productive inputs such as intermediate materials and capital services. For this reason, enterprise surveys were used for most countries. Using these source data, indicators are calculated on labor and/ or total factor productivity disaggregated by STAN industry and year, and on the decomposition of productivity growth into within-firm and reallocation components. The underlying source data and availability of the indicators are provided in table 1.2. Indicators Collected Depending on the availability of output and input measures, we have calculated diﬀerent indicators of labor and total factor productivity. A number of issues emerged in the calculation of labor and total factor productivity, including: • Labor input was generally based on the number of employees with no correction for hours worked. • Sales and gross output data do not include correction for inventory accumulation. • Capital stock, in countries where available, is based on book values.

Social security Social security Business register Employment-based register Business register Business register Register, based on Integrated System of Pensions Census Annual Industry Survey (ENIA) Annual Manufacturing survey (EAM) Business Register Fiscal register (APEH) Manufacturing survey Census Business register Social security Business register Business register Census Annual Industrial Survey

Germany (West) Italy Netherlands Portugal U.K. U.S. Argentina

Estonia Hungary Indonesia Korea Latvia Mexico Romania Slovenia Taiwan (China) Venezuela

Brazil Chile Colombia

Business register Business register Business register Fiscal database

Source Period

All but civil service, self-employed All All All but public administration Manufacturing Private businesses All Manufacturing Manufacturing Manufacturing All All Manufacturing Manufacturing All All All All Manufacturing Manufacturing

1995–2002 1996–2001 1979–1999 1982–1998 1995–2001 1992–2001 1990–1995 1983–1993 (3 years) 1996–2002 1985–2001 1992–2001 1992–2001 1986–1991 (2 years) 1995–2000

All Economy All All All

Sectors

1977–1999 1986–1994 1987–1997 1983–1998 1980–1998 1988–1997

1984–1998 1981–1994 1988–1998 1989–1997

Data sources used for firm demographics

Canada Denmark Finland France

Country

Table 1.1

Yes Yes Yes No No Yes Yes Yes Yes No No

Yes Yes Yes

Yes Yes Yes Yes Yes Yes

No No Yes Yes

Availability of survival data

Emp ≥ 10 Emp ≥ 1 Emp ≥ 1 Emp ≥ 10 Emp ≥ 5 Emp ≥ 1 Emp ≥ 1 Emp ≥ 1 Emp ≥ 1 Emp ≥ 1 Emp ≥ 1; sample for 1–15

Emp ≥ 1 Emp ≥ 1 Emp ≥ 10

Emp ≥ 1 Emp ≥ 1 Emp ≥ 1 Turnover: Man: Euro 0.58m Serv: Euro 0.17m Emp ≥ 1 Emp ≥ 1 None Emp ≥ 1 Emp ≥ 1 Emp ≥ 1

Threshold

1993–1998 1987–1992 1992–1997 1996–2001

1986–1991 1980–1985 1987–1992 1990–1995 1997–2001 1980–1985 1994–1999 1982–1986 1994–1998 1995–2000 1996–2001 1992–1996 1997–2001 1990–1995 1988–1993 1996–2001 1997–2002 1995–1998 1996–1999 1992–1997 1997–2001 1986–1991 1991–1996 1995–1999 1996–2000

Employment-based register Survey

Census Annual Industrial Survey (INDEC)

Annual Industrial Survey

Annual Industry Survey (ENIA) Annual Manufacturing survey (EAM) Business Register Fiscal register (APEH) Manufacturing survey Census Business register Business register Business register Census Annual Industrial Survey

Portugal U.K.

U.S. Argentina

Brazil

Chile Colombia Estonia Hungary Indonesia Korea (Rep.) Latvia Romania Slovenia Taiwan (China) Venezuela ✓ ✓ ✓

✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓

✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓

✓

✓

✓ ✓ ✓ ✓

✓

✓ ✓

✓ ✓

✓ ✓ ✓ ✓

✓ ✓ Some

✓ ✓ ✓ ✓

✓

✓

✓ ✓ ✓ ✓

✓

✗ ✓

✓ ✗ ✓ ✓

✓

TFP

Productivity LPV, LPQ ✓

Serv

✓

Mfg

Coverage

✓ ✓

✓ ✓ ✓

✓

✓

✓

✓

MFP

Plant Estab Firm Plant Firm Firm Firm Firm Firm Firm Firm

Estab

Estab Estab

Firm Estab

Firm Plant Firm Firm

Firm

Unit

Turnover €0.58m Emp 1 Turnover €5m Emp > 20, emp < 20 → Sample Emp 1 Emp 100, emp 100 → Sample Emp 1 Emp ≥ 9 and $2m threshold Emp ≥ 30 sample of 10–29 Emp ≥ 10 Emp ≥ 10 Emp ≥ 1 Emp ≥ 1 Emp ≥ 10 Emp ≥ 5 Emp ≥ 1 Emp ≥ 1 Emp ≥ 1 Emp ≥ 1 Emp ≥ 1; sample for 1–15

Emp > 5

Threshold

Note: Mfg manufacturing; Serv business services; LPV labor productivity based on value added; LPQ labor productivity based on gross output; Emp employment.

1990–1995 1993–1998 1993–1998 1992–1997

1985–1990 1992–1997 1982–1987 1983–1988

Germany (W) Italy Netherlands

1989–1994

1975–1980

Census Fiscal database with additional information from enterprise surveys Survey Survey Survey

Last

Finland France

First

Source

Periods

Summary of the data used for productivity decompositions

Country

Table 1.2

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

27

• Total Factor Productivity (TFP) at the firm level is the log of deflated output (measured as value added) minus the weighted log of labor plus capital, where the weights are industry-specific and the same for all countries. The weights were calculated using the expenditure shares of inputs for an industry using the cross-country average from the OECDSTAN database. In the World Bank project, TFP was also computed using country and industry-specific average expenditures shares of firms. • Multifactor Productivity (MFP) calculations use expenditure shares for labor, capital, and materials. • Labor productivity estimates are based either on deflated growth output (LPQ) or on deflated value added (LPV). Similarly, MFP estimates are based on deflated gross output and TFP estimates are based on deflated value added. • Deflators for output, value added, and materials are at the two to three digit industry level, usually based on National Accounts sources. Using common factor shares across countries for a particular industry allows, in principle, for cross-country comparisons of productivity levels. However, diﬀerent measurement units for the inputs, notably capital, make cross-country comparisons of TFP or MFP levels problematic. To benchmark the levels of TFP and MFP, the measured units of capital are adjusted with a multiplicative factor, such that value added minus payroll (or gross output minus payroll and materials expenditures) represents a return to capital of eight percent.13 1.4 A Canonical Representation of the Measurement Problems As discussed in the previous section, despite all eﬀorts to harmonize the data, measurement issues remain that can aﬀect cross-country comparisons. In reviewing such measurement issues we use the following simple notation: the indicator I is some aggregate of a (vector of) variables X, with aggregation taking place across units (firms or establishments) f that are element of the (sub)population : (1)

I A[Xf | f ∈ ].

For simplicity, we drop all subscripts (i.e., for countries as well as for disaggregated groupings), such as industry or size-class. These disaggregations are dealt with by adding an appropriate subscript to I and X, and by aggregating over individual firms in an appropriately defined subset of . With this notation framework, we assess measurement problems for a host of indicators. In particular, we consider various aggregator functions, A[..], 13. This adjustment is similar to the arbitrary adjustments to TFP made by Bernard and Jones (1996b) in order to compare apples and oranges.

28

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

such as sums, means, variances, covariances, or statistical analyses yielding reduced-form or structural coeﬃcients (e.g., the aggregator function could be the ordinary least squares (OLS) estimator from a multivariate regression).14 The variable itself may be an aggregation of a function of one or more micro-level variables, such as a ratio (e.g., output per unit of labor) or a transformation using firm-level observations from multiple periods, such as a first diﬀerence. Alternatively, the indicator may be a function of aggregated variables (e.g., aggregate productivity as ratio of aggregate output to labor). Finally, the indicators may vary by the (sub)set of firms over which the aggregation takes place. For example, the typical productivity decompositions (see following equations) focus on the contribution to aggregate productivity growth of diﬀerent sets of firms (e.g., continuing, exiting, and entering firms). Measurement errors can be analysed in a typical errors-in-variable framework, such as: (2)

X X∗ ε

where the observed value, X, is equal to the actual value, X ∗, plus an error term. For the computed indicators, a necessary extension to the framework is that the observed and the actual set of firms, ∗, may diﬀer as well: (3)

∗

where is a general form of disturbance to the correct or actual set of firms in ∗. The disturbance takes away—or adds—units to the actual set. A simple example is when the focus of the analysis is on firms in a given industry, but some firms are erroneously classified in this industry even if they largely operate in another industry. Similarly, the actual set of continuing firms needed for decompositions of productivity growth is given by the intersection of the actual sets of firms at time t and firms at time t – s. Through errors applied to the actual sets at t or t – s the observed set of continuers may deviate from the actual, as will the complementary sets of observed exiting firms and entrants. As an added complication, it may be that the observed set diﬀers from the actual set, but that the actual set is a statistical sample drawn from the actual universe. Or it may be that the observed set is a statistical sample drawn from the observed universe, which itself is a noisy version of the actual universe. We abstract from this by taking the sampling scheme and the errors in classification to both be represented by , regardless of the order in which the sampling process and the errors drive a wedge between the actual universe and the observed set of firms. 14. The latter possibility includes treating estimated parameters from studies of individual countries as aggregate indicators.

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

29

Once a diﬀerentiation is made between the location of the errors, namely in the measurement of the variable(s) at the micro level or in the sampling or registration of the micro-units over which aggregation is made, the eﬀects of the various measurement problems can be traced and diﬀerent forms of errors may be compared. It should be stressed that while these two types of errors aﬀect the analysis of firm-level data in each individual country, diﬀerences in characteristics of these errors across countries influence even more cross-country comparisons. In the remainder of this subsection, we explore some examples of how measurement error in both the measures of economic activity and measurement error due to sample selection can impact the measures of the distribution of output, employment, productivity levels, and productivity dynamics drawn from firm-level data. 1.4.1 Mean or Sum Both measurement errors discussed above aﬀect aggregation indicators such as the mean or sum of firm-level data. We first discuss the case when generates random errors in obtaining the observed set of firms from the actual set. When the indicator of interest is the mean employment per firm, we get a consistent estimate by taking a normal average.15 Without measurement error of the firm-level variable, the variance of this estimator of the first moment is negligible, given the generally large size of available samples (often 90 percent or more of total employment is in the sample). With classical measurement error in employment at the firm level, ε increases the standard deviation of the (unbiased) estimate of the first moment. The estimate of mean firm-size across industries is unbiased, as the extra firms allocated to one industry represent a loss in another and, on average, the eﬀect will be zero. With measurement error ε proportional to size, for example because of weighted sampling by size strata, sample weights are needed to get consistent estimate of first moment of the firm-size distribution. When the indicator of interest is the diﬀerence between the mean (or the sum) of two diﬀerent level measures (e.g., labor productivity can be viewed as the diﬀerence of the [log] of aggregate output and employment), the previous remarks apply. The diﬀerencing does not solve the problem, or even creates further problems if the expected value of the measurement error of both measures is zero. But the variance of the estimated mean is the sum of the two classical measurement error variances, so, in this example, we have a noisier estimate of mean productivity. We need to take this into account when comparing productivity levels across countries. But having an estimate of the variance of ε would help to assess whether diﬀerences in mean productivity across country are significant. 15. It should be stressed that, given data availability, we define labor input as the number of employees and do not control for hours worked.

30

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

1.4.2 Mean or Sum; Endogenous (Sub)samples The measures of mean firm-size and number of firms by size class fall into this category, and will be noisy owing to misclassification. The sizeclass criterion used to split the subsample is not independent of ε: firms with positive noise are more likely to be above a threshold, firms with negative noise more likely below. This is a typical problem (e.g., the well known result of nonclassical measurement errors of dichotomous 0–1 indicator variables built from continuous variables with classical noise). A typical solution for this type of problem is to base the classification of firm-size on average employment in two periods. However, using only firms observed in two periods may, depending on the indicator of interest, introduce a selection bias. This problem of interaction between ε and characteristic used to make the (sub)samples in aggregation shows up for means by quartiles, for job flows, and for other such splits with endogenous classification. The problem is exacerbated if sampling errors () vary systematically with the same characteristics. In principle, weighted results can overcome this problem, but in many cases the at-risk population for the analysis is above a minimum size threshold. 1.4.3 Mean or Sum; Longitudinal Linkages and Measures of Change If aggregations are to be made over subsamples that are based on longitudinal linkages over time, such as entry/exit/continuer status, the sampling noise becomes quite important. For example, if we consider the employment of entering, exiting, and continuing firms, the measurement error in firm-level employment is coupled with possible mismeasurement of the status variables due to poor longitudinal IDs. In addition to measurement in the firm-level indicator and status variables, sample selection can play a large role here since under-sampled groups may exhibit very diﬀerent firm dynamics.16 1.4.4 Higher Moments (Variances) In computing the variance of the distribution of our firm-level variables (e.g., employment) we start by assuming no sampling errors. The estimated variance of the variable will be true variance in the universe plus the variance of ε. Without knowing the distribution of micro-level measurement errors, higher moments cannot be compared directly across countries. One practical solution is to compute the variance of the distribution of employment averaged over two periods (e.g., the decomposition of productivity by Griliches and Regev [1995]). The diﬀerence between estimate of the variance and (the average of) the variances estimated from the two an16. Martin (2005) provides details on how sample weights should be used for computing productivity contributions from exit and entry.

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

31

nual samples equals half the variance of ε. In other words, if the underlying true variance of our variable does not change over the two periods, the reduction in variance moving from the standard to the two-period average variance is a consistent estimate of 0.5 ∗ var(ε). However, this approach only works for calculating the variance of the cross-sectional distribution of the firm-level variable for continuing firms. We would also need to find out how the exit or entrant subsamples aﬀect the variance of the full annual distribution of firm employment. No correction can be made for measurement error of employment for these firms. A closely related alternative is available if the distribution of the measurement error is common to all firms in a country. In this case, disaggregating the data by, for example, industry and then using a diﬀerence-in-diﬀerence comparison of the relative cross-industry variances for diﬀerent countries can be made. Next consider a divergence between the observed and actual sample. If the sampling errors vary systematically by firm-size, we need to do appropriate weighting. If the sample varies, not because of sampling rules, but owing to error, this only matters if the errors are correlated with employment. If they are correlated, no consistent estimate can be made of higher level moments of the employment distribution. 1.4.5 Higher Moments (Covariances and Correlations) All of the previously mentioned problems apply to covariances, correlations, and, by association, estimates from regressions or other related multivariate statistical procedures. The problem with covariances is more complex since we must now deal with the covariance between the measurement error of two variables (either the same variable at diﬀerent points in time or diﬀerent variables at the same unit of time). Classical measurement error will bias any given correlation, but in many cases the measurement error may be systematic in complex ways. While the general intuition is that the classical measurement error implies lower covariances and correlations, in this setting the measurement error may yield diﬀerent results. For example, one key question with firm-level data is whether more productive businesses have higher market shares. Classical measurement error in output measures will yield spuriously high covariances between the output share of a business and its measure of productivity, while classical measurement error in labor input will result in spuriously low covariances between employment share of a business and its measure of productivity. The previously mentioned issue needs to be addressed in particular for the indicator of the gap between weighted and unweighted productivity. The gap is proportional to the covariance between labor productivity and firm employment. If output and labor input are both measured with (classical) error, the gap will be underestimated, with the underestimation dependent on the variance of the measurement error in labor input. In this case, an estimate of the variance of the measurement error of the firm level

32

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

variable will be useful to know how to adjust cross-country diﬀerences in the estimated gap. If instead, the statistical agency uses labor productivity as an analytical ratio to edit the underlying micro data, then the measurement error of productivity and labor may be uncorrelated, so that the gap measure will be unbiased. In either case, computing the covariance between the cross-section of the time average of productivity and the crosssection of the time average of employment will produce a gap estimate with a lower bias, because the mean measurement error goes to zero as more periods are added. Further, diﬀerence-in-diﬀerences approaches, for example looking at relative movements between gaps in diﬀerent industries, and comparing this across countries or over time, will provide robust estimates if the measurement error process of the firm-level variable does not change over time or across industries. 1.5 Assessing the Process of Creative Destruction We start our review by looking at the distribution of firms by size, for the total business sector and the subsectors. We then turn to the analysis of firm demographics—the entry and exit of firms and their impact on employment. Finally, we look at the evolution of cohorts of new firms over the initial years of their life. In all cases, our objectives are to present some of the basic facts that emerge from the newly developed cross-country data and also to evaluate the measurement and inference problems that emerge from such comparisons. In all our analysis we look at simple cross-country comparisons, but also at within-country variations along diﬀerent dimensions (size, industry). We claim that the diﬀerence-in-diﬀerence approach is essential to extract valuable information from our distributed micro-data analysis for at least two reasons: • First, despite our eﬀorts to harmonize the data across countries, there remain some diﬀerences in key dimensions: size or output thresholds that exclude micro-units, diﬀerences in the sectoral coverage and in some cases as well as diﬀerences in the definition of the unit of observation. These diﬀerences may all contribute to limit simple crosscountry comparisons using single indicators of the creative destruction process. • Second, and probably more importantly, simple cross-country comparisons on specific dimensions of the process of creative destruction may be misleading or inadequate. Diﬀerences in market structures and in institutions may lead to diﬀerences in the nature of creative destruction rather than in its absolute magnitude. For example, high barriers to entry may not reduce the overall magnitude of firm turnover

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

33

but rather the composition of entrant and exiting firms. Facing high entry costs, new firms may choose to either enter very small and avoid the bite of regulations (especially in developing countries), or enter with a large size and smooth the entry costs over a larger capital investment. This may lead to bimodal distributions of firm entry by size but not a lower total entry rate. Likewise, in countries with high barriers to entry (and in turn high implied survival probabilities of marginal incumbents), the average productivity of entrants will rise while the average productivity of incumbents and exiting businesses will fall. Similar predictions apply to policies that subsidize incumbents and/or restrict exit in some fashion. These institutional distortions might yield a larger gap in productivity between entering and exiting businesses, but this gap is not by itself suﬃcient to gauge the contribution or eﬃciency of the creative destruction process. In the empirical analysis presented in the remainder of this section and in the next section we focus on: • The period from 1989 onward, and use period averages instead of data for individual years to minimize business cycle eﬀects and possible measurement problems.17 • Twenty-three aggregate industries that cover the entire business sector while maximizing country coverage from the forty-two three-digit (ISIC Rev. 3) industries that are available in some databases.18 1.5.1 Indicators Collected The use of annual data on firm dynamics implies a significant volatility in the resulting indicators. In order to limit the possible impact of measurement problems, it was decided to use definitions of continuing, entering, and exiting firms on the basis of three (rather than the usual two) time periods. Thus, the tabulations of firm demographics is based on the following variables: Entryi,s,t: The number of firms entering in industry i, in the size class s and in year t. Also tabulated, if available, was the number of employees in entering firms. Entrant firms (and their employees) were those observed as (out, in, in) in the register at time (t – 1, t, t 1). Exiti,s,t: The number of firms—and related employees—that leave the register. Exiting firms were those observed as (in, in, out) the register in time (t – 1, t, t 1). 17. For Finland, we use the sample 1992–1998 because in the first years of the 1990s a number of large firms changed legal form in Finland, thus obtaining a diﬀerent firm code in the business register. This reregistration would inflate firm turnover rates for large firms and distort the assessment of firm characteristics among entrants and exiting firms. 18. These twenty-three industries also correspond to the sectoral disaggregation of the OECD Structural Analysis (STAN) database. See www.oecd.org/data/stan.htm

34

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

One-year firmsi,s,t: The number of firms and employees in those firms that were present in the register for only one year. These firms were those observed as (out, in, out) the register in time (t – 1, t, t 1). Continuing firmsi,s,t: The number of firms and employees that were in the register in a given year, as well as in the previous and subsequent year. These firms were observed as in the register in time (t – 1, t, t 1). In practice, a number of complications arise in constructing and interpreting data that conform to the definitions of continuing, entering, and exiting firms described above. In particular, the one-year category, in principle, represents short-lived firms that are observed in time t but not in adjacent time periods and could therefore be treated as an additional piece of information in evaluating firm demographics. However, in some databases this category also includes measurement errors and possibly ill-defined data. Thus, the total number of firms in our analysis excludes these oneyear firms. Given the method of defining continuing, entering, and exiting firms, a change in the stock of continuing firms (C ) relates to entry (E ) and exit (X ) in the following way: (4)

Ct Ct1 Et1 Xt.

This has implications for the appropriate measure of firm turnover. Given that continuing, entering, exiting and one-year firms (O) all exist in time t then the total number of firms (T ) is: (5)

Tt Ct Et Xt Ot.

From this, the change in the total number of firms between two years, taking into account equation 4, can be written as: (6)

Tt Tt1 Et Xt1 Ot Ot1.

Assuming that the one-year firms are measured with random noise, the diﬀerence of these firms in year t and t – 1 is expected to be equal to zero. Thus, a turnover measure that is consistent with the contribution of net entry to changes in the total number of firms should be based on the sum of contemporaneous entry with lagged exit. The above indicators were split into eight firm-size classes, including the class of firms without employees.19 The data thus allow detailed comparisons of firm-size distributions between industries and countries.20 Further, 19. The eight size classes are as follows: no-employees, 1–9 employees, 10–19, 20–49, 50– 99, 100–249, 250–499; 500. For the OECD countries there are only six size groups, with the two groups between one and twenty combined and the groups between 100 and 500 combined. 20. Available data also allow the calculation of total job turnover and the fraction of it due to the entry and exit of firms.

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

35

the collected data allow for survival analysis for a selection of countries over varying time periods. 1.5.2 The Distribution of Firms by Size Firm size is an important dimension in our analysis for several reasons. The empirical literature suggests that small firms tend to be aﬀected by greater churning, but also have greater potential for expansion.21 Thus, a distribution of firms skewed towards small units may imply higher entry and exit, but also greater post-entry growth of successful firms. Alternatively, it may point to a sectoral specialization of the given country towards newer industries, where churning tends to be larger and more firms experiment with diﬀerent technologies. Another factor relevant here is that small businesses may not be subject to the same regulations as large businesses, because they may be exempted to certain laws or regulations (e.g., labor regulations) or because they can more easily avoid them in countries with weak enforcement. In addition, the distribution of firm by size is likely to be influenced by the overall dimension of the internal market—especially for firms in nontradable sectors—as well as the business environment in which firms operate that can discourage firm expansion. The analysis of firm size raises clear problems for cross-country comparability related to sample selection problems. For most of the countries in our sample, the data cover all firms with at least one employee, but the cutoﬀ size is five employees in South Korea,22 ten employees in Chile, Colombia, and Indonesia. And for France and Italy, the data exclude firms with sales below a certain threshold. Second, even amongst the countries for which data cover all firms with at least one employee, the unit of reference is the plant instead of the firm in some countries, and the definition of both may vary across countries. Finally, from a sectoral perspective, community services and utilities are more diﬃcult to compare, given the important role of the public sector, whose coverage changes from country to country, and of regulation in these sectors. Table 1.3 presents the share of firms—and associated employment—in the first two classes of our size distribution: firms with fewer than twenty employees (panel A) and firms with twenty to forty-nine employees (panel B). The table suggests that in all countries the population of firms is dominated by micro and small units. Micro units (fewer than twenty employees) account for at least 80 percent of the total firm population. Their share in total employment is much lower and ranges from less than 15 percent in some transition economies (e.g., Romania)—which still reflects the 21. See Sutton (1997) for a review of the literature. 22. The annual enterprise survey in Venezuela is representative of all firms with at least fifteen employees, and only includes a random sample of firms below this threshold. In our analysis, we have used the data for Venezuela with reference to firms with twenty or more employees, given the lack of coverage for the lower size classes.

Total economy

Nonagriculture business sectora Manufacturing

Firms

Small firms across broad sectors and countries, 1990s

Total business services Total economy

Nonagriculture business sectora

Manufacturing

Employment Total business services

Panel A: The share of micro firms in the total population of firms and in total employment (firms with fewer than 20 employees as a percentage of total) Industrial countries Denmark 92.3 89.5 35.0 76.6 31.1 91.3 17.6 32.7 France 82.0 82.3 13.6 77.9 16.0 82.1 19.9 15.9 Italy 96.0 93.8 36.4 88.6 39.6 93.8 31.3 35.9 Netherlands 97.1 96.5 32.9 88.3 36.8 96.3 18.3 31.8 Finland 95.3 92.7 39.1 85.4 32.7 93.6 13.5 29.5 West Germany 85.8 83.3 23.8 89.6 16.6 25.8 Portugal 93.8 88.9 42.9 75.3 31.4 89.2 18.9 32.2 U.K. 81.3 12.4 U.S. 88.7 88.0 19.9 72.6 19.3 88.0 6.7 18.4 Latin America Brazil 82.4 17.7 Mexico 92.2 90.0 28.5 82.8 24.5 90.1 13.9 23.2 Argentina 91.2 89.4 27.7 82.1 27.7 90.0 21.3 27.7 Transition economies Slovenia 93.1 88.0 26.0 71.6 13.5 87.7 5.1 13.4 Hungary 90.8 85.5 23.6 71.1 16.4 84.4 8.8 16.0 Estonia 87.1 81.3 34.2 64.6 22.6 80.6 11.5 22.8 Latvia 87.6 87.7 24.2 87.8 24.8 87.7 26.9 24.7 Romania 95.6 91.5 31.6 77.1 12.8 90.9 4.2 12.9 East Asia Koreab 57.0 11.1 Taiwan (China) 82.5 26.6

Table 1.3

This aggregates excludes agriculture (ISIC 1–5) and community services (ISIC3: 75–79). In Korea, data cover firms with 5 or more employees.

b

a

Panel B: The share of small firms in the total population of firms and in total employment (Firms with 20–49 employees as percentage of 20) Industrial Countries 69.7 66.9 60.4 67.6 Denmark 22.9 19.5 22.5 49.9 53.3 63.0 53.2 France 12.9 20.0 12.9 65.5 69.4 67.0 67.3 Italy 22.8 23.0 20.0 58.5 62.9 53.9 58.8 Netherlands 18.6 14.2 15.3 65.2 62.0 54.3 61.0 Finland 18.9 10.3 16.3 60.7 54.0 17.7 12.8 17.2 West Germany 59.0 69.2 63.5 59.1 64.0 Portugal 22.0 21.5 22.6 U.K. 51.2 11.4 63.1 65.0 55.0 62.7 U.S. 13.5 12.2 7.3 Latin America Chile 51.4 15.3 Colombia 49.0 13.9 62.9 58.9 51.2 59.0 Mexico 16.0 11.5 15.1 Brazil 58.7 15.0 Venezuela 24.9 4.5 60.6 61.7 59.8 61.1 Argentina 18.6 19.0 18.4 Transition economies 49.8 38.4 29.2 38.5 Slovenia 7.2 4.6 7.4 61.9 56.2 48.3 54.6 Hungary 12.5 10.1 12.9 55.1 46.2 39.2 45.3 Romania 5.5 3.3 5.7 66.9 62.6 55.5 62.4 Estonia 21.3 17.0 22.1 58.0 57.9 60.1 58.1 Latvia 17.9 20.1 17.8 East Asia Korea 59.4 17.3 Indonesia 49.6 7.3 Taiwan (China) 65.8 25.2 12.4 14.3 11.2 27.1 17.5

16.8

17.1

12.7

22.9

22.0 11.2 15.6 13.9 19.0

38

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

presence of large (formerly or still) state-owned firms inherited from the central plan period—to less than 20 percent in the United States and around 30 percent or more in some small European economies. To check the robustness of these results, we also look at the incidence of small firms (i.e., the population twenty to forty-nine over the total population of firms with twenty or more employees). This allows for a larger country sample and greater comparability as it is not aﬀected by diﬀerences in the threshold of micro units. Small firms account for about 50 percent of the total population of firms with twenty or more employees, again with the exception of the transition economies (e.g., Romania and Slovenia) still dominated by large firms. It is also important to notice that the rank ordering of countries obtained by focusing on the share of micro units (fewer than twenty) is only loosely correlated with the rank order of the same countries based on the share of small firms (twenty to forty-nine).23 Cross-country diﬀerences in firm size may reflect specialization towards industries with a small eﬃcient scale. To assess the role of sectoral specialization versus within sector diﬀerences, we first look at the average firm size across industries in table 1.4. The first column of the table presents the cross-country average size for each industry and the other columns present the country/industry average relative to the industry cross-country average. If technological factors were predominant in determining firm size across countries, we should find that the values in the country columns to be concentrated around one. If, on the contrary, the size diﬀerences were explained mainly by country-specific factors inducing a consistent bias within industries, then we would expect the countries with an overall value above (below) the average (i.e., in the “Total” category) to be characterized by values generally above (below) one in the subsectors. Among industrial countries, the United States has a very high proportion of industries with an above-average firm size, both in manufacturing and in business services. The Western European countries tend to have smaller firms in most industries, with several exceptions in heavy industries (e.g., Germany and Portugal), high-tech industries (e.g., Finland and, to a lesser extent, France and Italy), or some of the low-tech industries (e.g., United Kingdom) or in basic services (e.g., France and Portugal). Thus, it is not possible to map diﬀerences in firm size across countries according to either the overall size of the country (apart from the United States), the underlying technological level of the industry, or its degree of maturity. Another way to shed light on country-specific factors versus industryspecific technological factors is to use a shift-and-share decomposition. The decomposition identifies the component due to cross-country diﬀerences in firm size within each sector, the component due to diﬀerences in the sectoral composition across countries, and a cross term that can be 23. The country rank correlation is only 0.3.

1.20 1.09

1.14 0.99

0.79 0.82 0.92 0.89 1.04 0.87 0.82 0.71 0.76 0.68 1.71 0.82 1.00 0.95

115

81

125

870 218 107

125 272

86 153

140

202

204

148

1.04

0.52

1.17 1.32

1.20 1.24

1.10 0.96 1.13

1.25

1.00

1.01

143

1.49 1.40 1.14

1.20

Other countries

0.67 0.76 0.87

0.87

Industrial

77 173 137

118

Crosscountry average

0.77

0.65 0.26

0.10 0.58 0.81

0.66

0.71

0.55

0.93

0.44 0.27 0.69

0.76

Denmark

0.62

0.82

0.92 0.72

0.53

1.27

0.69 0.49

0.61 0.52

0.34 0.80 0.69

0.71

0.58

0.50

0.66

0.84 0.44 0.63

0.84

Italy

1.08

0.79

0.84

0.78 0.65 0.81

0.59

0.68

0.68

0.57

1.02 0.48 0.72

1.11

France

0.17

0.06

0.14

0.02

0.27 0.18

0.29 0.32

0.42 0.25

0.29

0.25

0.22

0.33

0.21 0.13 0.34

0.38

Netherlands

0.95

1.39

0.57

1.99

0.78 0.84

0.97 1.45

0.92 1.01 0.96

2.02

1.73

0.78

1.15

0.78 1.12

1.01

Finland

0.82

0.33 1.54 1.20

0.86

0.91

0.89

0.62

0.54 1.59 0.94

0.88

West Germany

Within-industry average firm size, firms with 20 or more employees (as a share of cross-country sectoral average)

Agriculture, hunting, forestry and fishing Mining and quarrying Total manufacturing Food products, beverages, and tobacco Textiles, textile products, leather, and footwear Wood and products of wood and cork Publishing, printing, and reproduction of recorded media Coke, refined petroleum products, and nuclear fuel Chemicals and chemical products Rubber and plastics products Other non-metallic mineral products Basic metals Fabricated metal products, except machinery and equipment Machinery and equipment, N.E.C. Office, accounting, and computing machinery Electrical machinery and apparatus, nec Radio, television, and communication equipment Medical, precision, and optical instruments

Total economy

Country

Table 1.4

0.90

1.58

0.98

0.46

0.74 0.43

0.66 0.37

2.52 0.52 0.64

0.58

0.70

0.76

0.61

0.74 0.39 0.62

0.72

Portugal

1.12

1.41

4.52

1.21 1.11

1.12 1.29

1.54 2.34 1.56

1.60

1.00

1.58

2.99

1.00 1.14 1.71

1.32

U.S.

2.13 (continued )

0.85 0.82

1.31 0.59

0.84

0.86

1.07

1.53

1.03

U.K.

(continued)

Agriculture, hunting, forestry, and fishing Mining and quarrying Total manufacturing Food products, beverages, and tobacco Textiles, textile products, leather, and footwear Wood and products of wood and and cork

Total economy

Country

Motor vehicles, trailers, and semi-trailers Other transport equipment Manufacturing nec; recycling Electricity, gas, and water supply Construction Services Market services Wholesale and retail trade; restaurants and hotels Transport, and storage, and communication Finance, insurance, Real Estate, and business services Community, social, and personal services

Country

Table 1.4

1.00 1.07 0.94

250

135

123

0.81 0.88 0.99 0.83

0.76

0.75

0.85

1.19

Colombia

0.93

79

Chile

0.94 1.21 0.87 1.24 0.79 0.95 0.94

Industrial

316 305 92 505 75 111 107

Crosscountry average

1.01

0.98

0.99

1.56 0.75 1.01

1.00

Mexico

1.06

0.91

1.00

1.09

1.04 0.85 1.13 0.62 1.32 1.07 1.08

1.72

2.50

1.35

2.05 3.14 1.49

1.40

0.95

1.24

1.18

1.20 1.23 1.08

1.11

1.21

1.58

0.71

0.76

1.60 0.85 0.60 1.55 0.76 1.13 0.98

Italy

3.43

2.82

0.95

1.49

Indonesia

1.04

1.29

0.38

1.45

0.92 0.89 0.95 0.15 0.93 1.30 1.36

France

Hungary

0.92

0.98

0.41

0.83

0.77 0.37 0.75 0.84 0.82

Denmark

Slovenia

Other countries

0.77

0.91

0.77

0.79

Korea

0.70

0.71

1.10

0.57

Taiwan (China)

0.52

0.37

0.21

0.36

0.10 0.08 0.88 0.29 0.32 0.43 0.37

Netherlands

0.70

0.97

0.83

0.83 2.92 0.73

0.72

Estonia

0.85

0.75

1.21

0.30 0.68 0.73 0.29 0.98 0.90 1.15

Finland

0.84

0.88

1.28

0.50 0.86

Brazil

0.41

0.88 0.36 0.74

1.06

0.81

0.97

0.42 0.65

0.83

3.00

1.77

2.20

1.55

1.23

0.86

1.36

2.45 4.16 1.39 0.97 0.83 1.35 1.24

U.S

0.71

0.73

0.90

0.71 0.71 0.70

0.85

Argentina

0.97

U.K.

Romania

0.58

1.52

3.64

0.76

0.53 0.58 0.56 5.38 0.97 0.80 0.88

Portugal

Latvia

West Germany

Publishing, printing, and reproduction of recorded media Coke, refined petroleum products, and nuclear fuel Chemicals and chemical products Rubber and plastics products Other non-metallic mineral products Basic metals Fabricated metal products, except machinery and equipment Machinery and equipment, N.E.C. Office, accounting, and computing machinery Electrical machinery and apparatus, nec Radio, television, and communication equipment Medical, precision, and optical instruments Motor vehicles, trailers, and semi-trailers Other transport equipment Manufacturing nec; recycling Electricity, gas, and water supply Construction Services Market services Wholesale and retail trade; restaurants and hotels Transport, storage, and communication Finance, insurance, Real Estate, and business services Community, social, and personal services 1.48 0.80

0.98

0.94 0.52 0.35 0.58 0.70 0.50 0.38 0.39 0.68

0.98 0.62

0.18

0.52

0.43

0.43

0.30 0.62 0.80 0.67 1.10 1.32 0.35 1.89 1.13 1.20 1.47 1.26 0.56 0.77

1.41 0.57 1.06 0.48 0.86 0.99 0.90 0.95 0.46 1.15 1.48

1.04

0.27

1.79 1.23

0.44 1.37 1.82 1.45 1.79

0.08 0.68 0.88 0.98 0.69

0.31 0.63 0.90 1.05 0.67

1.62

0.16 0.56 0.82 0.83 1.03

0.83

0.85

0.95

0.75

0.80

1.77

1.12

1.01 0.58 1.04 0.90 0.87 1.19 1.27

0.82

1.14

1.40

1.22

1.02 0.72

7.32 1.42 0.91 1.33 1.04

0.84

0.71 0.68 1.86

2.64

2.37

1.15

1.81 1.05

0.12 0.86 1.98 0.80 0.94

1.39

0.61 1.11 0.91

0.67

0.88

0.52

0.91

0.91 0.53

0.18 0.56 0.70 0.76 0.52

0.64

0.31 0.30 0.67

0.46

0.41

0.90

0.56 0.35

0.52 0.62 0.51 0.25

0.51

0.50

0.43

0.62

0.65

0.52 0.74 1.37 0.37 0.75 0.65 0.69

0.74

0.88

0.89

0.25

0.69 0.59

1.70 0.58 0.48 0.79 0.21

0.54

0.83 0.31 0.77

0.73

0.85

0.75

0.92

0.92 0.79

0.39 0.68 0.84 0.60 0.66

0.99

0.87

0.73

0.65

1.16

0.10 0.15 0.56 0.17 1.35 0.90 0.93

1.14

0.82

0.22

0.73

0.71 0.84

0.03 0.28 0.49 0.79 0.10

0.67

1.45

1.18

1.72

1.14

4.52 3.94 2.91 1.65 2.41 1.40 1.44

2.78

1.93

4.10

0.81

2.27 6.47

2.08 3.01 2.66 3.44 5.19

2.61

0.93

1.24

0.58

0.88

0.43 0.24 0.52 0.39 0.99 0.99 1.00

0.39

0.93

0.38

0.32

0.65 0.43

0.38 0.56 0.60 0.80 0.50

0.74

42

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

Table 1.5

Shift and share analysis of the determinants of firm size Contribution coming from differences in:

Country Denmark France Italy Netherlands Finland Portugal U.K. U.S. Canada Brazil Mexico Argentina Slovenia Hungary Estonia Latvia Romania Korea Taiwan (China)

Sectoral composition

Average size of firms

Interaction between sectoral comp. and size

Total

0.14 0.08 –0.02 0.01 –0.02 –0.05 –0.01 0.00 0.01 0.00 0.06 0.04 0.01 0.01 –0.03 –0.03 0.08 0.04 0.03

–0.03 –0.05 –0.17 –0.13 –0.05 –0.04 –0.02 0.42 0.03 –0.08 –0.06 –0.14 0.30 0.14 0.07 –0.20 0.97 0.12 –0.14

–0.09 –0.05 –0.01 –0.04 –0.02 0.02 –0.03 –0.07 –0.02 –0.01 –0.02 –0.02 –0.07 –0.02 0.02 0.04 –0.36 0.02 –0.03

0.01 –0.02 –0.20 –0.16 –0.09 –0.07 –0.06 0.34 0.01 –0.09 –0.02 –0.12 0.24 0.12 0.06 –0.20 0.68 0.18 –0.14

Note: The Total represents the percentage deviation of average size from the cross-country average; the other columns decompose the total into subcomponents.

interpreted loosely as an indicator of covariance: if it is positive, size and sectoral compositions deviate from the benchmark in the same direction.24 The decomposition (table 1.5) suggests that within-sector diﬀerences generally play the most important role in explaining diﬀerences in overall size across countries: this component is much larger (in absolute terms) that the sectoral composition component in many countries.25 The withinindustry size component is particularly large in the United States, con24. The decomposition is as follows: sj – s ∑ ij sij – ∑ i si ∑( ij – i )si ∑(sij – si) ii ∑(sij – si )( ij – i ) i i i i i s s where sj is the average firm size in country j, sij is the average firm size in subsector i, and ij is the share of firms in subsector i with respect to the total number of firms; s is the overall mean across countries and i is the share of overall number of firms in subsector j. 25. In a sensitivity analysis, we have also replicated the decomposition for the sample of OECD countries and the non-OECD countries (including also Hungary and Mexico) separately. The results are broadly unchanged in the two subsamples. Moreover, we have replicated the decomposition at a finer level of sectoral disaggregation and again the results are broadly unchanged.

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

43

firming the idea that a larger internal market tends to promote larger firms, but also in some transition economies (e.g., Romania). However, the sectoral composition also plays an important role in some small European countries such as Denmark and Portugal, but also in a relatively larger country such as France and an emerging economy like Mexico.26 All in all, diﬀerences in average firm size seem to be largely driven by within-sector diﬀerences, although in some countries sectoral specialization also plays a significant role. Smaller countries tend to have a size distribution skewed towards smaller firms, but the average size of firms does not map precisely with the overall dimension of the domestic market. 1.5.3 Gross and Net Firm Flows The second step in our analysis is to look at the magnitude and characteristics of firm creation and destruction. We present entry and exit rates for all firms with more than one employee, and for those firms with twenty and more employees, to avoid comparability problems related to size cutoﬀs in some country data. As discussed in the previous section, we focus on time averages (1989 onwards) rather than annual data to minimize possible measurement problems. Figure 1.2 shows entry and exit rates for the business sector and for manufacturing. The results point to a high degree of turbulence in all countries (and confirm one of the regularities pointed out by Geroski [1995] for industrial economies). Many firms enter and exit most markets every year. Limiting the tabulations to firms with at least twenty employees to maximize the country coverage, total firm turnover (entry plus exit rates)27 is between 3 and 8 percent in most industrial countries and more than 10 percent in some of the transition economies. If we extend the analysis to include micro units (one to nineteen employees), we observe total firm turnover rates between one-fifth and one-fourth of all firms. These data also confirm previous findings that in all countries net entry (entry minus exit) is far less important than the gross flows of entry and exit that generate it. This suggests that the entry of new firms in the market is largely driven by a search process rather than augmenting the number of competitors in the market (a point also highlighted by Audretsch [1995]). There are also interesting diﬀerences across countries. The Latin American region shows a wide variety of experiences; for example, while Mexico 26. The decomposition also suggests that the two elements of the decomposition are not highly correlated; the interaction term is negative in most cases, and the sign of the two elements of the decomposition also tend to diﬀer in most cases. In other words, there is no clear link between size structure and sectoral specialization tilted towards productions naturally characterized by large firms (see Davis and Henrekson [1999] for a discussion). 27. The entry rate is defined as the number of new firms divided by the total number of incumbent and entrants firms producing in a given year; the exit rate is defined as the number of firms exiting the market in a given year divided by the population of origin (i.e., the incumbents in the previous year).

A

B

C

Fig. 1.2 Firm turnover rates in broad sectors, 1990s: A, manufacturing, firms with 20 or more employees; B, manufacturing, firms with at least 1 employee; C, total business sector, firms with 20 or more employees; D, total business sector, firms with at least 1 employee

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

45

D

Fig. 1.2

(cont.)

and the manufacturing sector of Brazil show vigorous firm turnover, Argentina shows less turbulence, closer to the values observed in some continental European countries. The transition economies of Central and Eastern Europe provide other interesting features. In most of these countries, firm entry largely outpaced firm exit, while more balanced patterns are found in other countries. Obviously this is related to the process of transition to a market economy, and is not sustainable over the longer run. Still, it points to the fact that new firms not only displaced obsolete incumbents in the transition phase but also filled in new markets that were either nonexistent or poorly populated in the past. As stressed in the previous section, diﬀerences in sample selection and measurement error in longitudinal linkages can yield spurious diﬀerences in measures of firm turnover. It is very diﬃcult without detailed information about the statistical processing in each of these countries—as well as within country validation studies—to assess this problem. Instead, our approach is to consider related measures of firm dynamics that, in some fashion, attempt to overcome these measurement concerns. We begin our inquiry into the validity of the turnover data by first weighting firm turnover by employment and then comparing the size of entrant firms with that of the average incumbent. If we focus on the entire population of firms with at least one employee, we see that less than 10 percent of employment is, on average, involved in firm creation and destruction. The diﬀerence between unweighted and employment-weighted firm turnover rates arises from the fact that both entrants and exiting firms are generally smaller than incumbents. For most countries, new firms are only 20 to 60 percent the average size of incumbents. But the small size of entrants relative to the average incumbents is driven by diﬀerent factors across countries. In particular, we observe that entrant firms are relatively smaller in the

46

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

United States than in most of the other industrial countries. This is in part due to the larger market of the United States that leads to larger average size of incumbents.28 But the wider gap between entry size and the minimum eﬃcient size in the United States may also reflect economic and institutional factors (e.g., the relatively low entry and exit costs may increase incentives to start up relatively small businesses). In the transition economies, new firms are substantially diﬀerent from most of the existing firms that were drawn from the centrally-planned period. Indeed, the net entry of firms (entry rate minus exit rate) is particularly large among micro units (twenty or fewer employees); during the centrally planned system there were relatively few of these micro firms, which exploded during the transition in most of business service activities, however. Unfortunately, the observed diﬀerences in the relative size of entrants across countries may still reflect longitudinal linkages problems. If, in some countries spurious entry is more prevalent and the continuing businesses that are spuriously labeled entrants are larger than true entrants, then this will increase the relative size of entrants in the country. An alternative approach to overcoming measurement problems in firm turnover measures is to disaggregate by some key business characteristic and compare within-country variations in firm turnover. One interesting characteristic in this context is obviously the business size. Figure 1.3 presents entry rates by diﬀerent size classes in manufacturing. In most countries, entry rates tend to decline with firm size, consistent with the view that firms tend to enter small, test the market, and, if successful, expand to reach the minimum eﬃciency scale. But in some European countries, we observe a flattening of the entry rate for firms greater than twenty employees, or even a U-shaped relation whereby entry rates tend to increase for larger firms compared with small firms.29 It is interesting to notice that those countries where we observe the flattening of the entry rates are those generally characterized by relatively high administrative costs to set up a business.30 The latter may stimulate firms to enter very small—and thus partly avoid some of the entry costs that kick in at a given size—or enter at a larger size and thus spread these fix entry costs over a larger investment plan. This is only a working hypothesis, which is however corroborated by more detailed econometric analysis (see Scarpetta et al. 2002). Of course, 28. Geographical considerations are also likely to aﬀect the average size of firms; firms with plants spreading into diﬀerent U.S. states are recorded as single units, while establishments belonging to the same firm but located in diﬀerent EU states are recorded as separate units. 29. Focusing on the total business sector suggests a more monotonic relationship between entry rate and size classes; however, the steepness of the downward relations is less marked in those countries where we observe a flattening or even a U-shaped relation in manufacturing. 30. For example, France (3.5), Italy (4.6), the Netherlands (1.6), and Finland (1.8) all have indicators of the administrative costs of setting up a business (least regulated 0, most regulated 6) largely above the United States (0.7) or the United Kingdom (0.8). See Nicoletti, Scarpetta, and Boylaud (1999) for details on these indicators.

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

Fig. 1.3

47

Entry rates by firm size, manufacturing, 1990s

Note: Data for Finland are from 1992 to 1998.

the specific diﬀerence-in-diﬀerence approach works only if the measurement error in firm turnover does not vary systematically by size class. Longitudinal linkage problems interacting with sample selection problems that vary by size may be a problem in some countries. Another dimension that can be used for this diﬀerence-in-diﬀerence approach is clearly the industry. Sectoral variation within and between coun-

48

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

tries may reflect a rich mix of the technological, cost, and demand factors driving firm dynamics, as well as market structure and institutions in the country. Table 1.6 presents sectoral gross firm turnover rates (entry plus exit rates weighted by employment) normalized by the overall crosscountry industry average. As before, if technological and cost factors were predominant in determining the heterogeneity of firm dynamics across countries, we should find that the values in the country columns of table 1.6 are concentrated around one. The first element to report is that the variability of turnover rates for the same industry across countries is comparable in magnitude to that across industry in each country. Turnover rates (especially if weighted by employment) are somewhat higher in the service sector (especially in trade) than in manufacturing.31 However, in most countries, some high-tech industries with rapid technological changes and market experimentation had relatively high entry rates in the 1990s (e.g., oﬃce; computing and equipments; and radio, TV, and communication). Transition countries, as well as Mexico, tend to have greater firm churning than industrial countries, on average. The finding of important industry eﬀects that hold across countries suggests a possible future avenue for the diﬀerence-in-diﬀerence approach to shed light on the role of institutions in shaping firm dynamics. Taking the U.S. firm dynamics as a benchmark for the underlying churning that is needed by technological and costs factors, it is possible to compare the cross-industry variation in the United States with that of other countries with stricter business regulations. If these regulations were indeed constraining firm dynamics, we should observe smaller variance in countries with stricter regulations. Some recent studies have indeed found some preliminary evidence that this is indeed the case.32 1.5.4 The Post-Entry Performance of Firms Another useful metric to characterize firm dynamics is to examine postentry performance of firms. Understanding the post-entry performance sheds light on the market selection process that separates successful entrant firms that survive and prosper from others that stagnate and eventually exit. In addition, post-entry performance is a measure that exploits variation that may be less subject to measurement error. Conditional on

31. In Italy, however, there appears to be only small diﬀerences in churning between manufacturing and services. This is particularly evident for the employment-weighted turnover and likely reflects the small diﬀerences in average size of firms between manufacturing and services. The lower turnover rate in the French service sector compared with that in manufacturing is likely to depend on the existence of a size threshold in the French data, which tends to be more binding in the service sector than in manufacturing. As an indication, the French data also suggest a higher average size of firms in the service sector than in manufacturing, in contrast with all other countries. 32. See Micco and Pages (2006); Klapper, Laeven, and Rajan (2006).

1.27 1.21

1.19 1.30 1.19 1.09

1.05 1.00

0.94 0.85 0.87 0.75 0.94 0.87 0.89 0.76 0.83 0.94 0.96 0.97 1.00 0.97

22.2

19.8

19.6

19.7 15.5 16.8

18.1 18.0

19.1 17.5

24.1

17.6

19.5

17.1 1.04

1.05

1.38 1.09 1.19

1.11

1.19

0.89

17.6

0.97 1.17 1.18

1.16

1.02 0.90 0.89

0.90

21.6

Other countries

19.8 18.9 19.0

Industrial

Crosscountry average

0.98

0.92 0.98

1.17 0.88

0.95

0.93

1.02

1.06

0.98 0.79 0.97

0.91

Denmark

1.12

0.95

1.08

1.06

0.97

1.18 1.17 0.94

0.99

0.80

0.88

1.06

1.08 0.85 1.06

0.89

France

1.00

0.80

0.81

1.02

0.73 0.74

0.67 0.59

0.51 0.83 0.78

0.71

0.69

0.79

0.83

0.78 0.61 0.77

0.74

Italy

0.67

0.95

0.94

0.86

0.70 0.63

0.76 0.81

0.63 0.93 0.72

0.77

0.62

0.80

0.54

0.66 1.21 0.72

0.77

Netherlands

0.74

0.93

0.88

0.94

0.87 1.08

0.94 0.80

0.94 0.86 0.93

0.84

0.92

0.90

0.95

0.94 0.91

0.96

Finland

Gross firm turnover across countries and sectors (as a ratio of cross-country industry average)

Agriculture, hunting, forestry, and fishing Mining and quarrying Total manufacturing Food products, beverages, and tobacco Textiles, textile products, leather, and footwear Wood and products of wood and cork Publishing, printing, and reproduction of recorded media Coke, refined petroleum products, and nuclear fuel Chemicals and chemical products Rubber and plastics products Other non-metallic mineral products Basic metals Fabricated metal products, except machinery and equipment Machinery and equipment, N.E.C. Office, accounting, and computing machinery Electrical machinery and apparatus, Nec Radio, television, and communication equipment Medical, precision, and optical instruments

Total economy

Table 1.6

0.53

0.53 0.67 0.67

0.66

0.49

0.73

0.57

1.14 0.47 0.61

0.76

West Germany

1.01

1.17

1.03

2.49

1.01

0.95 0.86

0.88 0.89

0.87

1.13

1.03

1.36 1.02 0.96

1.00

Portugal

1.57

1.37

1.39

1.36

1.04 1.37

1.48 0.74

1.12

1.12

1.13

1.43

1.24

U.K.

0.90

0.86

0.48

0.69 0.83

0.88 0.55

0.69 1.01 0.95

0.91

0.91

1.02

0.89

1.03 1.12 0.91

1.05

Canada

0.82 (continued )

0.93

0.83

0.98

0.77 0.82

0.83 0.81

0.82 0.95 0.98

0.91

1.05

1.19

0.98

1.09 1.07 0.93

0.93

U.S.

(continued)

Agriculture, hunting, forestry, and fishing Mining and quarrying Total manufacturing Food products, beverages, and tobacco Textiles, textile products, leather, and footwear Wood and products of wood and cork

Total economy

Motor vehicles, trailers, and semi-trailers Other transport equipment Manufacturing Nec; recycling Electricity, gas, and water supply Construction Services Bus sector services Wholesale and retail trade; restaurants and hotels Transport, and storage, and communication Finance, insurance, Real Estate, and business services Community, social and personal services

Table 1.6

0.82

21.7 1.20

1.08

0.92 1.27 1.29 1.23 1.30 1.48

19.8 18.9 19.0

17.6

22.2

19.8

1.22

0.94

23.9

1.08

21.6

0.95

24.0

1.16

Mexico

0.88

22.9

1.10 1.08 1.18 1.00 1.23 1.14 1.13

Other countries

Crosscountry average

0.92 0.94 0.88 1.00 0.85 0.89 0.90

Industrial

17.5 20.7 20.4 13.5 23.3 22.7 23.3

Crosscountry average

1.13

1.02

1.06

1.27 1.00 1.12

1.21

Slovenia

0.63

1.03

0.80

0.95

0.97 0.89 0.82 0.89 0.95

Denmark

0.56

0.74

0.69

0.76

0.81 0.76 0.73 0.58 0.83 0.70 0.75

Italy

1.39

1.03

1.32

1.00 1.05 1.17

1.21

Hungary

0.85

0.91

0.76

0.78

0.95 0.91 0.88 1.47 0.81 0.84 0.83

France

0.96

0.65

0.78

0.77 0.35 0.81

0.81

Estonia

0.63

0.92

0.70

0.66

0.64 0.72 0.81 1.80 0.57 0.76 0.78

Netherlands

1.29

1.31

1.51

1.16 1.31

Brazil

0.90

0.94

0.88

0.93 0.97 0.85 0.54 1.04 0.92 0.90

Finland

1.42

1.31

1.54

2.15 1.48

1.30

Latvia

0.84

0.61 0.44 0.62

West Germany

0.97

1.13

1.74

0.93

0.69 0.71 0.64 1.85 1.10 0.97 0.95

1.51 1.39 1.48

U.K.

1.19

0.98

1.26

Romania

Portugal

0.67

0.90

1.00

0.95

0.94 0.92 1.00 0.54 0.97 0.87 0.94

U.S.

0.79

0.86

0.86

0.56 0.85 0.78

0.75

Argentina

1.31

0.92

0.96

0.99

0.83 0.99 0.93 1.39 0.94 1.04 0.97

Canada

Publishing, printing, and reproduction of recorded media Coke, refined petroleum products, and nuclear fuel Chemicals and chemical products Rubber and plastics products Other non-metallic mineral products Basic metals Fabricated metal products, except machinery and equipment Machinery and equipment, N.E.C. Office, accounting, and computing machinery Electrical machinery and apparatus, Nec Radio, television, and communication equipment Medical, precision, and optical instruments Motor vehicles, trailers, and semi-trailers Other transport equipment Manufacturing Nec; recycling Electricity, gas, and water supply Construction Services Bus sector services Wholesale and retail trade; restaurants and hotels Transport and storage and communication Finance, insurance, Real Estate, and business services Community, social, and personal services 0.92 1.28 1.12 1.11 0.60 1.11 1.23 1.19

1.26 1.10 1.41 0.69 1.76 1.16 1.16 1.18 1.10 1.11 1.03

17.1

17.5 20.7 20.4 13.5 23.3 22.7 23.3

22.9

24.0

23.9

21.7

1.60

1.04

1.18

1.27

0.93

19.5

1.27 1.12

1.06

1.32

19.1 17.5

0.98 1.11

17.6

1.33 1.25

18.1 18.0

1.81 0.96 1.39

0.73

1.07 1.11 1.22

19.7 15.5 16.8

1.30

24.1

1.25

19.6

1.39

1.18

1.09

1.21

1.04 1.32 1.24 1.43 1.10 1.23 1.20

1.09

1.00

1.14

0.94

1.20 1.03

1.32 1.36

1.16 1.12 1.25

1.26

0.83

0.79

0.71

0.82

0.77 0.69 0.86 0.90 0.74 0.82 0.80

0.73

0.96

0.86

0.69

0.70 0.62

0.86 2.78

5.07 0.57 1.02

0.76

1.31 1.21 1.22

1.38

1.38

1.27

1.22

1.20 1.19

1.18 1.25

0.54 1.42 1.22

1.20

1.32

1.17

1.12

1.22

1.27 1.31 1.41 2.01 1.22 1.23 1.20

1.54

1.27

1.09

1.36

1.56 1.78

1.72 2.32

2.49 1.44 1.29

1.52

1.32

1.28

1.14

1.26

0.92 0.92 1.03 1.23 0.95 1.27 1.24

1.05

0.95

1.14

1.33

1.08 0.96

1.10 0.93

0.55 1.32 1.12

1.23

0.57

0.73

0.92

0.82

0.63 0.77 0.77 0.48 1.03 0.75 0.82

0.60

0.74

0.69

1.16

0.71 0.76

0.74 0.68

0.86 0.73 0.76

0.75

52

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

the sample or register capturing an entrant firm, there is a reasonable chance that the sample or register will be able to follow the firm over time. Figure 1.4 presents nonparametric (graphic) estimates of survivor rates. The survivor rate specifies the proportion of firms from a cohort of entrants that still exist at a given age. In the figure, the survival rates are averaged over diﬀerent entry cohorts (those that entered the market in the late 1980s and 1990s) to minimize possible business cycle eﬀects and possible measurement problems. Looking at cross-country diﬀerences in survivor rates, about 10 percent (Slovenia) to more than 30 percent (in Mexico) of entering firms leave the market within the first two years (fig. 1.4). Conditional on overcoming the initial years, the prospect of firms improves in the subsequent period; firms that remain in the business after the first two years have a 40 to 80 percent chance of surviving for five more years. Nevertheless, only about 30 to 50

A

B

Fig. 1.4 Firm survival at diﬀerent lifetimes, 1990s: A, manufacturing; B, total business sector

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

53

percent of total entering firms in a given year survive beyond the seventh year in industrial and Latin American countries, while higher survival rates are found in transition economies.33 For most countries, the rank ordering of survival is similar whether using a two-year, four-year, or seven-year horizon, suggesting that there is an important country eﬀect that impacts the survival function. However, there are a few interesting exceptions. The United States has relatively low survival rates at the two-year horizon but relatively higher survival rates at the seven-year horizon. This pattern might reflect the relatively rapid cleansing of poorly performing firms in the United States. Table 1.7 provides details on the survival rates at four years of age across industries and countries. The structure of the table is similar to those presented previously. Notably, the variation across countries is more systematic than that across industries. Across industries, between 60 and 80 percent of firms survive after four years, while for example, the survival rate in oﬃce and computing equipment deviates across countries from 40 percent below to 40 percent above the cross-country average of 70 percent. The total employment in each given cohort tends to increase in the initial years because failures are highly concentrated among its smallest units and because of the significant growth of survivors. These facts are best presented by looking at gains in average firm size amongst surviving firms. Given diﬀerences in data collection, the reference average size of entrants is that at duration one for industrial countries and duration zero for other countries, but excluding firms with zero employment. The choice for the industrial countries is dictated by the fact that entrant firms include zero-employee firms. For example, in the United States, the time when the firm is registered and when its employment is recorded diﬀer, giving rise to the possibility that firms are recorded as having zero employees in the entry year and positive employment in the second year.34 This, however, may represent an overcorrection as it eliminates employment growth in firms with positive employment at registration. Figure 1.5 shows the evolution in average firm size of survivors as they age, corrected for possible changes in entry size of the actual survivors by age. In the figure, the average size of survivors at diﬀerent duration is compared with that at entry. The diﬀerence in post-entry behavior of firms in the United States35 compared with the western European countries is partially 33. Survivor rates for firms with twenty or more employees at age one are similar to those observed in the newly compiled EUROSTAT firm-level database (EUROSTAT 2004). 34. However, recent work by the U.S. Census Bureau shows that even after correcting for the zero-employee problem, the size expansion of entrant firms in the United States exceeds that in other industrial countries by a wide margin. The growth in firm size in the ensuing years shows that the United States continues to perform much better than other OECD countries. 35. The results for the United States are consistent with the evidence in Audretsch (1995). He found that the four-year employment growth among surviving firms was about 90 percent. See also Dunne, Roberts, and Samuelson (1988, 1989).

0.97 1.01

0.96 1.04 0.98 1.05 1.02 0.98 1.02 0.99 1.01 1.01 0.88 0.93 0.92 0.96 0.99 0.98

0.59

0.64

0.69

0.73 0.69 0.73 0.68 0.69

0.69 0.73

0.70

0.74

0.71

0.77

0.70 0.65

1.01 1.01

1.04

1.08

1.06

1.10

0.99 0.99

0.96 0.99 1.01 0.98 1.01

1.03

0.98

1.02

0.69

0.94 1.00

Other countries

1.05 1.00

Industrial

0.69 0.67

Crosscountry average

0.87 0.78

1.03

0.99

0.90

0.92

0.95 0.96

1.05 0.89 0.91 0.96 0.94

0.97

0.95

0.95

0.92

1.07 0.94

Finland

1.03 1.00

0.88

0.86

1.01

1.03

1.05 0.96 0.97 1.08

0.92

1.10

0.98

1.10

0.91 1.02

France

0.72 0.77

0.70

0.73

0.71

0.61

0.90 0.70

0.67 0.88 0.96 0.76 0.85

0.82

0.86

0.75

0.69

0.79

U.K.

1.20 1.09 1.00 1.11

1.04

1.12

0.99

1.10

1.14 1.10

West Germany

1.08 1.05

0.92

1.00

1.00

1.05

1.05 1.00

1.23 1.11 1.00 1.10 1.08

1.03

1.09

1.04

1.08

1.10 1.04

Italy

Survival rate (4 years of age) across countries and industries (as a ratio to cross-country sectoral average)

Mining and quarrying Total manufacturing Food products, beverages, and tobacco Textiles, textile products, leather, and footwear Wood and products of wood and cork Publishing, printing, and reproduction of recorded media Coke, refined petroleum products, and nuclear fuel Chemicals and chemical products Rubber and plastics products Other non-metallic mineral products Basic metals Fabricated metal products, except machinery and equipment Machinery and equipment, N.E.C. Office, accounting, and computing machinery Electrical machinery and apparatus, Nec Radio, television, and communication equipment Medical, precision, and optical instruments Motor vehicles, trailers, and semi-trailers Other transport equipment

Table 1.7

1.05 1.14

1.08

0.91

1.00

1.03

1.08 1.09

1.13 1.04 1.06 1.08 1.15

1.03

1.19

1.02

1.01

1.15 1.07

Netherlands

1.29 1.25

1.15

1.00

0.99

1.13

1.12 1.29

1.37 1.14 1.06 1.16 1.01

1.13

1.04

1.14

1.35

1.11 1.12

Portugal

0.92 0.95

0.95

0.95

0.91

0.80

1.00 0.99

0.79 1.01 0.90 0.97 0.93

0.94

0.99

0.81

0.91

0.85 0.95

U.S.

1.01

0.70

0.65

Total nonagricultural business sector

Mining and quarrying Total manufacturing Food products, beverages, and tobacco Textiles, textile products, leather, and footwear Wood and products of wood and cork Publishing, printing, and reproduction of recorded media Coke, refined petroleum products, and nuclear fuel Chemicals and chemical products Rubber and plastics products Other non-metallic mineral products Basic metals

0.98

0.66

1.05 1.00 1.02 0.96 1.04 0.98 1.05 1.02 0.98 1.02 0.99

0.69

0.59

0.64

0.69

0.73 0.69 0.73

0.68 0.69

Industrial

0.69 0.67

Crosscountry average

1.02

0.64

1.02

1.02 1.01 1.07 1.02

0.66 0.82 0.64 0.66

Manufacturing Nec; Recycling Electricity, gas, and water supply Construction Market services Wholesale and retail trade; restaurants and hotels Transport, storage, and communication Finance, insurance, Real Estate, and business services

0.98 1.01

1.11

0.95 1.14

0.95

1.01 0.96 0.99 1.01

1.01

1.19

1.02

0.49 1.05

Estonia

0.97

1.03

0.98

0.94 1.00

Other countries

0.99

0.99

1.02

0.98

0.98 0.99 0.94 0.98

1.04 0.97

0.97 1.04 1.10

1.04

1.08

1.21

1.03

1.11 1.10

Hungary

1.00

1.01

1.22

0.91

0.93 1.14 1.00 0.99

1.17 1.35

1.14 1.07 1.12

1.08

1.04

1.30

1.09

0.98 1.11

Latvia

0.99

0.85

1.05

1.01

0.99 0.98 1.00 0.96

1.09 1.03

1.37 1.09 1.05

1.06

1.09

Romania

0.82

0.78

1.22 1.32

1.37 0.95 1.20

1.23

1.26

1.20

1.15

1.40 1.22

Slovenia

1.05

1.00

1.00

1.02

1.14 1.01 1.10 1.01

0.89 0.90

0.83 1.02 0.94

0.93

0.83

0.91

0.86

0.84 0.89

Argentina

1.04

1.01

1.04

1.03

1.04 1.00 1.03 1.02

0.98 1.13

0.93 1.00 1.02

1.09

1.13

1.08

1.03

1.04

Chile

1.16

1.16

1.07

1.07

1.11 0.99 1.18 1.14

0.83 0.92

1.11 1.00 0.90

1.02

0.77

0.87

0.95

0.87

0.92 0.86 0.81

0.77

0.69

0.80

0.80

0.69 0.76

Mexico

0.97

0.95

0.94

0.96

0.92 0.95 0.98 0.96

0.74 0.78 (continued )

Colombia

1.13

1.10

0.45

1.12

1.29 1.01 1.18 1.09

1.08

0.92 0.96 0.99 0.98 1.02 1.01 1.07 1.02 1.02 0.98 1.01

0.71

0.77

0.70 0.65 0.66 0.82 0.64 0.66

0.64

0.66

0.70

0.65

Total nonagricultural business sector 1.02

1.06

0.93

0.74

0.99

0.99

1.02

0.98

1.01 1.01 0.98 0.99 0.94 0.98

1.04

1.10

0.88

0.99 0.99

Other countries

0.70

Industrial

1.01 1.01

Crosscountry average

0.69 0.73

(continued)

Fabricated metal products, except machinery and equipment Machinery and equipment, N.E.C. Office, accounting, and computing machinery Electrical machinery and apparatus, Nec Radio, television, and communication equipment Medical, precision, and optical instruments Motor vehicles, trailers, and semi-trailers Other transport equipment Manufacturing Nec; recycling Electricity, gas, and water supply Construction Market services Wholesale and retail trade; restaurants and hotels Transport, storage, and communication Finance, insurance, Real Estate, and business services

Table 1.7

1.09

1.06

1.15

1.06

1.07 1.37 1.05 0.95 1.16 1.07

1.30

0.95

1.02

1.42

1.12 1.01

Estonia

1.10

1.06

1.11

1.07

1.14 1.13 1.11 0.98 1.16 1.06

1.07

1.07

1.06

1.16

1.12 1.09

Hungary

1.15

1.13

1.22

1.13

1.43 1.43 1.20 1.12 1.21 1.12

1.15

1.27

1.05

1.10

1.21 0.96

Latvia

1.00

1.00

1.04

0.98

1.14 1.21 1.11 1.05 1.17 0.96

1.01

1.07

1.10

1.02

1.09 1.03

Romania

1.23

1.20

1.14

1.20

1.16 1.06 1.17 1.06 1.31 1.19

1.12

1.22

1.13

1.22

1.27 1.20

Slovenia

0.88

0.91

0.98

0.87

0.95 0.83 0.89 0.95 0.66 0.89

0.99

0.86

0.93

0.60

0.85 0.86

Argentina

1.07

0.96 0.88 1.07

1.04

1.06

1.14

1.42

1.00 0.97

Chile

0.90

0.83 0.88 0.78

0.81

1.04

0.98

1.42

0.82 0.75

Colombia

0.67

0.75

0.78

0.74

0.81 0.76 0.70 0.88 0.32 0.73

0.70

Mexico

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

57

A

B

Fig. 1.5 Average firm size relative to entry, by age: A, manufacturing; B, total business sector

due to the larger gap between the size at entry and the average firm size of incumbents (i.e., there is a greater scope for expansion among young ventures in the U.S. markets than in Europe). In turn, the smaller relative size of entrants can be taken to indicate a greater degree of experimentation, with firms starting small and, if successful, expanding rapidly to approach the minimum eﬃcient scale.36 Latin American countries also oﬀer a wide range of post-entry performance of firms. Argentina has very limited post-entry expansion of successful firms in manufacturing, while in Mexico selection of small firms is stronger than in all other countries. However, post-entry growth of successful firms is also very strong, pointing to vigorous market selection process but also to sizeable rewards for successful new firms. 36. This greater experimentation of small firms in the U.S. market may also contribute to explain the evidence of a lower than average productivity at entry.

58

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

Transition economies also show a diﬀerent behavior from most other countries on firm survival. They tend to show higher survivor rates and large post-entry growth of successful firms, which confirm the hypothesis that new firms enjoyed a period of relatively low market contestability, especially in new low populated markets. Romania is obviously an outlier among transition economies; not only are failure rates higher than in the other countries, but even successful entrants have more limited opportunities of expanding. 1.6 The Eﬀects of Creative Destruction on Productivity 1.6.1 Reallocation and Productivity: Growth versus Level Comparisons In the previous two sections we have presented evidence of significant cross-country diﬀerences in firm characteristics, their market dynamics, and post-entry performance, which cannot be fully explained by diﬀerences in sectoral composition of the economy but rather points to salient diﬀerences in market characteristics and in business environment. The next obvious question is, do these diﬀerences matter for aggregate performance? We address this question in a number of ways. First, we examine the connection between productivity growth and the reallocation dynamics that we have documented in the prior sections. We are particularly interested in the contribution of entering and exiting businesses as well as the contribution of the reallocation of activity among continuing businesses. However, this analysis of dynamic eﬃciency, while inherently interesting, is fraught with interpretational and measurement diﬃculties. We attempt to overcome some aspects of these diﬃculties by exploiting sectoral variation within countries and then, in turn, comparing such sectoral diﬀerences across countries. In addition, we explore static eﬃciency by viewing a cross-sectional decomposition of productivity. The latter turns out to be simpler and more robust in terms of theoretical predictions and measurement problems. The approach taken in much of the empirical literature is to use accounting decompositions that decompose aggregate growth into components that reflect the contributions of productivity growth within continuing firms, the firm turnover process, and the reallocation of resources across continuing firms. The decompositions are correct in an accounting sense but interpreting the results is, as noted, fraught with challenges. Part of the problem here is to develop tight links between theoretical models of productivity enhancing reallocation and these empirical decompositions. One way to think about the empirical decompositions is that they provide a set of moments that models should match. Lentz and Mortensen (2005) take this approach by using a model of reallocation where the key frictions are in the labor market (via search frictions). Levinsohn and Petrin (2005)

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

59

provide a related useful benchmark by showing that in a model without frictions and without entry and exit, aggregate productivity growth is given by the weighted average of the productivity growth of continuing firms. In other words, without frictions there is no contribution of reallocation to aggregate productivity growth. These two studies remind us that the role of reallocation in productivity growth is inherently related to underlying frictions in the markets. For example, for net entry to be important it must be the case that it is costly for firms to enter and exit. If it is costless in terms of time and resources for firms to enter or exit, we would not observe any diﬀerence in the productivity at the margin between entering and exiting businesses. While frictions are at the core of this connection between productivity and reallocation, precisely how these frictions interact with the connection between productivity and reallocation is complicated both conceptually and in the evidence that emerges from cross-country evidence. We turn to these issues and findings now. 1.6.2 Reallocation and Productivity Growth Let’s define the sector-wide productivity level in year t, Pt as: Pt ∑it pit

(7)

i

where i is the input share of firm i and Pt and pit are a productivity measure.37 In this chapter we focus on labor productivity based on gross output data, although other measures are available for a subset of countries/sectors. We also use a decomposition suggested by Baily, Hulten, and Campbell (BHC henceforth, 1992) and in turn modified by Foster, Haltiwanger, and Krizan (FHK henceforth, 2001). BHC and FHK decompose aggregate (or industry-level) productivity growth into five components, commonly called the within eﬀect, between eﬀect, cross eﬀect, entry eﬀect, and exit eﬀect, as shown in order below: (8)

Pt ∑itk pit ∑ it ( pitk Ptk) ∑ it pit i∈C

i∈C

i∈C

∑it ( pit Ptk) ∑itk ( pitk Ptk) i∈N

i∈X

where means changes over the k-years’ interval between the first year (t – k) and the last year (t); it is as before; C, N, and X are sets of continuing, entering, and exiting firms, respectively; and Pt–k is the aggregate (i.e., weighted average) productivity level of the sector as of the first year (t – k).38 The FHK method uses the first year’s values for a continuing firm’s share 37. A variety of measures of productivity have been used in the literature including labor productivity, measures of total factor productivity that vary from estimated residuals from production functions to divisia index approaches to multilateral index number approaches. 38. The shares are usually based on employment in decompositions of labor productivity and on output in decompositions of total factor productivity.

60

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

(it–k), its productivity level (Pit–k), and the sector-wide average productivity level (Pt–k). One potential problem with this method is that, in the presence of measurement error in assessing market shares and relative productivity levels in the base year, the correlation between changes in productivity and changes in market share could be spurious, aﬀecting the within- and between-firm eﬀects. To tackle these potential problems, we have also used the approach proposed by Griliches and Regev (1995), which uses the time averages of the first and last years for them (i, pi, and P ). As a result the cross-eﬀect, or covariance term, disappears from the decomposition. The averaging of market shares reduces the influence of possible measurement errors, but the interpretation of the diﬀerent terms of the decomposition is less clear-cut as the time averaging makes the within-eﬀect term aﬀected by changes in the firms’ shares over time, and the between-eﬀect term aﬀected by changes in productivity over time. The results obtained using this method are qualitatively similar to those obtained using the FHK and are not presented in the chapter. As a final sensitivity analysis, we also use the method proposed by Baldwin and Gu (BG henceforth, 2002) that uses, as a reference for the calculations of the relative productivity of the diﬀerent groups, the average productivity of exiting firms. With this method, the contribution from exiting firms disappears and the entry component is positive if, on average, their productivity is higher than those of firms they are supposed to replace (the exiting firms). In all of these decompositions, the baseline analysis is based on five-year rolling windows for all periods and industries for which data are available. We also present results for three-year rolling windows and test the hypothesis that the contribution from entry changes with the time horizon considered. However, care has to be taken in interpreting the entry and exit components as they do not always reflect a comparison between productivity levels at the same point in time. For example, in the version of the FHK decomposition used here, the entry component comprises the diﬀerence between average productivity among entrants at the end of the threeto five-year period with overall productivity at the beginning. Therefore, it is obvious that a positive entry component does not necessarily mean that productivity among entering firms is above average in relation to their contemporaries. Before discussing the results of these decompositions, it is important to notice that their interpretation is not always straightforward from a theoretical, as well as measurement, point of view. The working hypothesis that poor market structure and institutions will distort the contribution of the creative destruction process has complex implications when using these basic accounting decompositions. The reason is that distortions may aﬀect the reallocation dynamics on diﬀerent margins in a variety of ways. For example, artificially high barriers to entry will lead to reduced firm turnover

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

61

and to a less eﬃcient allocation of resources. But given the high barrier to entry (and in turn the implied ability of marginal incumbents to increase survival probabilities), the average productivity of entrants will rise while the average productivity of incumbents and exiting businesses will fall. Similar predictions apply to policies that subsidize incumbents and/or restrict exit in some fashion. The point is that institutional distortions might yield a larger gap in productivity between entering and exiting businesses, which will contribute to larger net entry term in the previous decompositions. Alternatively, some types of distortions in market structure and institutions might make the entry and exit process less rational (i.e., less driven by market fundamentals but more by random factors). Such randomness may be associated with either a higher or lower pace of churning. Pure randomness would, in principle, increase the pace of churning, but the random factors might be correlated with other factors (e.g., firm size) and thus the impact would be to distort the relationship between churning and such factors with less clear predictions on the overall pace of churning. In any event, such randomness would imply less systematic diﬀerences between entering, exiting, and incumbent businesses—in the extreme when all entry and exit is random there should be no diﬀerences between entering, exiting, and incumbent businesses.39 Another related problem is that a business climate that encourages more market experimentation might have a larger long-run contribution but a smaller short-run contribution from the creative destruction process. That is, the greater market experimentation may be associated with more risk and uncertainty in the short run so that it is only after the trial and error process of the experimentation has worked its way out (through learning and selection eﬀects) that the productivity payoﬀ is realized. Thus, a business climate that encourages market experimentation might have a lower short-run contribution from entry and exit but a higher long-run contribution from entry and exit. Thus, in terms of these decompositions, the horizon over which the decomposition is measured may have a major eﬀect on the contribution of net entry in a specific country in a manner that is idiosyncratic to that country, and therefore impact any cross-country comparisons.40 In short, the gap between the productivity of entering and exiting busi39. Oviedo (2005) models the randomness of institutional enforcement as a way of capturing variations in institutional quality across countries. She shows that such randomness reduces the link between firm turnover and allocative eﬃciency. 40. Foster, Haltiwanger, and Krizan (2001) found large diﬀerences in the contribution of entry and exit between five- and ten-year horizons in the United States. Their analysis suggests that this is because entering cohorts in the United States are very heterogeneous. The selection of the least productive entrants in the first several years as well as the relatively greater increases in productivity for surviving entrants, relative to more mature incumbents over the same period, imply that the impact of net entry is much larger at a ten-year horizon than a five-year horizon. They show that this holds even taking into account the inherently higher share of activity accounted for by entering and exiting businesses over a longer horizon.

62

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

nesses is not by itself suﬃcient to gauge the contribution or eﬃciency of the creative destruction process. In addition, diﬀerent types of distortions might be acting simultaneously in a country. It might be that diﬀerent policies act to subsidize incumbents (preferential treatment for incumbents), other policies artificially increase the barriers to entry (poorly functioning financial markets and/or regulatory barriers), while other policies make exit more random for some types of businesses (e.g., poorly functioning financial markets for young and small businesses). As such, there might be too little churning on some dimensions and too much on others, and the gap between entering and exiting businesses might be too large on some margins and too small on others. With all of these caveats in mind, figure 1.6 presents the decomposition of labor productivity growth in the total business sector and figure 1.7 presents the decomposition of labor productivity for the manufacturing over the 1990s for a large sample of countries. A number of elements emerge from these decompositions: • Productivity growth is largely driven by within-firm performance. In industrial and emerging economies (outside transition), productivity within each firm accounts for the bulk of overall labor productivity growth. This is particularly the case if one focuses on the three-year horizon (not reported). Over the longer run (i.e., five-year horizon),

Fig. 1.6

Firm-level labor productivity decomposition for Total Business Sector

Notes: Chile: 1985–1999; Estonia 2000–2001; West Germany 2000–2002; Latvia 2001–2002; Portugal 1991–1994. Excluding Brazil and Venezuela. Within within-firm productivity growth. Between productivity growth due to reallocation of labor across existing firms. Entry productivity growth due to entry of new firms. Exit productivity growth due to exit of firms. Firm turnover Entry plus exit rates.

Firm-level labor productivity decomposition for manufacturing

Notes: Argentina 1995–2001; Chile 1985–1999; Colombia 1987–1998; Estonia 2000–2001; Finland 2000–2002; France 1990–1995; West Germany 2000– 2002; Korea 1988 and 1993; Latvia 2001–2002; Netherlands 1992–2001; Portugal 1991–1994; Slovenia 1997–2001; Taiwan 1986, 1991 and 1996; U.K. 2000– 2001; U.S. 1992 and 1997. Excluding Brazil and Venezuela.

Fig. 1.7

64

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

reallocation (and, in particular, the entry component) plays a stronger role in promoting productivity growth. • The impact on productivity via the reallocation of output across existing enterprises (the between eﬀect) varies significantly across countries. It is generally positive but small. This factor should be assessed together with the covariance (or cross) term, which combines changes in productivity with changes in employment shares. The covariance term is negative in most countries, including the transition economies. This implies that firms experiencing an increase in productivity were also losing market shares (i.e., their productivity growth was associated with restructuring and downsizing rather than expansion). This negative cross term, in a related way, is potentially associated with adjustment costs of labor. That is, in any given cross section there are some businesses that have recently had a productivity shock, but due to adjustment costs have not adjusted their labor inputs (at least fully). For businesses with a recent positive shock, the higher productivity will lead to a higher desired demand for labor and thus we will see such businesses increase employment, but due to diminishing returns (in the presence of any fixed factors at the micro level), a decrease in productivity. • Finally, the contribution of net entry to overall labor productivity growth is generally positive in most countries, accounting for between 20 percent and 50 percent of total productivity growth. The exit eﬀect is always positive (i.e., the least productive firms exit the market contributing to raise the productivity average of those that survive). Data for European countries show that new firms typically make a positive contribution to overall productivity growth, although the eﬀect is generally of small magnitude. By contrast, entries make a negative contribution in the United States for most industries. Interpreting these findings without more information is diﬃcult. The weak performance of entrants in the United States might reflect greater experimentation, so that for each entering cohort of entrants there is more selection and potentially more learning by doing.41 In transition economies, in all but one country (Hungary over the three-year horizon) the entry of new firms makes a positive and often strong contribution to productivity. For most countries, while the contribution of net entry is posi41. Some evidence in favor of this interpretation is provided in Haltiwanger, Jarmin, and Schank (2003); Foster, Haltiwanger, and Krizan (2001, 2002); and Bartelsman and Scarpetta (2004). The former work provides evidence of greater market experimentation in the United States relative to Germany. The latter shows that as the horizon lengthens in the United States, the contribution of net entry rises disproportionately. Moreover, Foster, Haltiwanger, and Krizan (2001, 2002) show that the increased contribution of net entry is due to both selection of the low productivity entrants and due to learning by doing to successful entrants.

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

65

tive, it is less than proportionate relative to the share of employment accounted for by firm turnover. An open question is whether the observed diﬀerences across countries are accounted for by diﬀerences in market institutions and policies or whether they reflect diﬀerent circumstances and/or problems of measurement. As discussed above, drawing such inferences from cross-country evidence is diﬃcult given that the policy environment may impact in a variety of ways and given the measurement problems. Consider, for example, the problem of measurement error in firm turnover that yields too high a measure of turnover for a country because of longitudinal linkage problems. Other things equal, spuriously high firm turnover will increase the share of activity associated with entering and exiting businesses and therefore increase the contribution of net entry to productivity growth. However, this same measurement error is likely to impact the diﬀerences in productivity between continuing, entering, and exiting businesses. If the true relationship is such that exiting businesses are less productive than continuing businesses, spurious entry and exit will tend to reduce this diﬀerence since some of the measured exiting businesses will in fact be continuing businesses. For entry, the relationship is potentially more complicated and also related to interpretation as well as diﬀerences across countries in the nature of their dynamics. For a country where entrants are immediately more productive than continuers, spurious measurement error will tend to reduce the gap and therefore decrease the contribution of net entry. For a country where entrants tend to be less productive than incumbents at entry perhaps due to market experimentation, as in the United States, spurious entry and exit will decrease the negative gap and therefore increase the contribution of net entry (since it will reduce a negative eﬀect). One set of countries where these measurement and interpretation problems appear to be interacting in interesting ways is for the transition economies (Estonia, Hungary, Latvia, Romania, and Slovenia). In these countries, there is a very high rate of firm turnover as a share of total employment and entry accounts for a large (but less than proportionate to the share of turnover) share of productivity growth. The large contribution of entry partly reflects the large rate of firm turnover, but it also reflects by construction a positive gap between entrants and incumbents productivity. In interpreting the latter finding, it is useful to put it in the context of the high pace of turnover. In general, it is diﬃcult to interpret diﬀerences across countries in the magnitude of the gap between entering and exiting businesses. For example, this gap might reflect fundamentals driving market selection with new businesses adopting the latest business practices (or in transition economies, new businesses adopting market business practices relative to incumbents), or it might reflect a very high entry barrier so that only very productive new businesses enter. However, the latter expla-

66

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

nation might suggest that firm turnover rates should be lower, which does not appear to be the case for the transition economies. Still, for these economies the contribution of net entry is far from proportionate, suggesting that there is substantial churning of businesses via entry and exit that is not productivity-enhancing. Our data also allow checking the sensitivity of the contribution of firm entry to diﬀerences in the time horizon. Table 1.8 presents the diﬀerence in the components of the decomposition as the horizon increases from three to five years for selected countries. To make the three and five year components comparable, the components have all been annualized. For the selected countries, lengthening the horizon increases the annual contribution of net entry, decreases the annual contribution of the between component, and has a mixed impact on the within component. The increase in the net entry component is largest for the transition economies, with a relatively large increase of almost three percent for Estonia. For the transition economies at least, these findings are consistent with the hypothesis that learning and selection eﬀects increase the contribution of net entry over a longer horizon. There is also an important sectoral dimension to the process of restructuring, reallocation, and creative destruction. Figure 1.8 presents the productivity decompositions for two groups of industries in manufacturing: (a) the low technology industries, and (b) the medium-high-technology industries. The large negative cross-term discussed previously (i.e., the fact that firms with strong productivity growth downsized is evident in lowtech industries, while in medium-high-tech industries this eﬀect, albeit still present, seems to be smaller). Even more interesting, the contribution of new firms to productivity growth is modest in low-tech industries, and even largely negative in a few countries, including the United States. But the entry eﬀect is strongly positive in medium-high-tech industries. This result suggests an important role for new firms in an area characterized by stronger technological changes. Given our focus on measurement issues in this chapter, these findings provide another illustration why exploiting the

Table 1.8

Time horizon differences Difference in component from 5 to 3 years

Country Argentina Chile Colombia Estonia Latvia Slovenia

Net entry

Between

Within

0.001 0.002 0.001 0.028 0.019 0.007

–0.001 –0.005 –0.005 –0.006 –0.009 –0.001

0.028 –0.007 –0.004 –0.007 0.027 0.001

Productivity decomposition by technology groups

Notes: Argentina 1995–2001; Chile 1985–1999; Colombia 1987–1998; Estonia 2000–2001; Finland 2000–2002; France 1990–1995; West Germany 2000–2002; Korea 1988 and 1993; Latvia 2001–2002; Netherlands 1992–2001; Portugal 1991–1994; Slovenia 1997–2001; Taiwan 1986, 1991, and 1996; U.K. 2000–2001; U.S. 1992 and 1997. Excluding Brazil and Venezuela.

Fig. 1.8

Low Tech Industries

(cont.)

Notes: Argentina 1995–2001; Chile 1985–1999; Colombia 1987–1998; Estonia 2000–2001; Finland 2000–2002; France 1990–1995; West Germany 2000– 2002; Korea 1988 and 1993; Latvia 2001–2002; Netherlands 1992–2001; Portugal 1991–1994; Slovenia 1997–2001; Taiwan 1986, 1991 and 1996; UK 2000– 2001; U.S. 1992 and 1997. Excluding Brazil and Venezuela.

Fig. 1.8

Medium and High Tech Industries

Measuring and Analyzing Cross-Country Differences in Firm Dynamics Table 1.9

Argentina Chile Colombia Estonia Finland France Korea, Republic Latvia Netherlands Portugal Slovenia Taiwan (China) U.K. U.S. West Germany

69

Accounting for the differences between FHK and BG decompositions Net entry difference

Exit/entry share difference

Incumbent/exit productivity difference

–0.01 –0.007 0.003 –0.001 –0.002 0.003 –0.042 0.000 0.001 –0.011 0.010 –0.014 0.005 0.002 0.000

–0.012 –0.022 0.008 –0.031 –0.013 0.034 –0.122 –0.001 0.028 –0.039 0.059 –0.077 0.148 0.012 0.001

0.098 0.432 0.627 0.28 0.251 0.107 0.495 –0.037 0.025 0.394 0.252 0.264 0.051 0.299 0.274

Notes: The reported figures are the time series averages. The first column is the product of the second and third column. However, since the reported figures are averages over time, the identity may appear not to hold (the product of the averages is not the same as the average of the product).

cross-industry variation within countries is a useful approach in crosscountry analysis. Table 1.9 presents the diﬀerence in the net entry component (annualized) for the FHK and BG methodologies. Recall that a key diﬀerence is that FHK use the initial average productivity of all plants as the benchmark from which entering and exiting plants’ productivity are compared, while BG use the exiters’ productivity. Foster, Haltiwanger, and Krizan (2001) motivate their approach as having desirable accounting properties (i.e., entering plants contribute positively to industry productivity growth over time if they are above the initial average, while exiting plants contribute positively to industry productivity growth if they are below the initial average). Baldwin and Gu (2003) motivate their approach as being more appropriate to the extent that entrants are displacing exiting plants, so the correct reference group for entrants are the exiting businesses they are displacing.42 For most countries the diﬀerence is small. It is intuitive that the eﬀects should in general be small because for both methods the net entry term depends critically on the diﬀerence between average productivity of entering and exiting businesses. In other words, both the entry and the exit term subtract oﬀ whatever base is used, so at first 42. One technical limitation of this alternative is that it implies, in turn, that the benchmark for the between component is the productivity of the exiters, which is diﬃcult to motivate.

70

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

glance, it might appear that the base is irrelevant (the base term in each component cancels out in the net). Consistent with this perspective, computing the diﬀerence between the FHK and BG net entry terms yields: (9)

FHK BG (∑itk ∑it)(Ptk PXtk) i∈X

i∈N

X where Pt–k is the average productivity of incumbents and Pt–k is the average productivity of exiting businesses in the base year. Thus, if the share of activity (in this case employment) accounted for by entering and exiting businesses is the same, then the diﬀerence is zero. As seen in section 1.5, for most countries the share of activity accounted for by entry is about the same as that for exit, typically with the latter slightly larger since exiting businesses tend to be larger than entering businesses. Thus, this diﬀerence in weights does not matter for most countries. However, for Korea—and to a lesser extent Portugal and Taiwan (China)—the share of employment accounted for by exit is substantially less than the share of employment accounted for by entry, leading to larger diﬀerences between the two decomposition methods. This diﬀerence yields an especially big eﬀect in Korea given that the gap between incumbents and exiting businesses is also large. To conclude this discussion of dynamic decompositions, it is worth highlighting the range of problems in drawing inferences from cross-country comparisons of the contribution of net entry across countries. For one, these decompositions depend critically on accurately measuring the extent of entry and exit. As we have noted, spurious entry and exit will have complex implications for the contribution of net entry with eﬀects working in potentially opposite directions. For another, horizon may play a critical role in these decompositions and such horizon diﬀerences are arguably diﬀerent across countries (and industries). The horizon problems are mitigated if very long diﬀerences are used (e.g., ten years), but this in turn poses problems of data limitations and measurement (e.g., the measurement problems may be worse over a longer horizon). We believe that these dynamic decompositions highlight some interesting patterns that appear to reflect rich actual diﬀerences in the firm dynamics.

1.6.3 The Cross-Sectional Eﬃciency of the Allocation of Activity So far, the creative destruction process has been discussed mostly from the point of view of productivity growth. This is natural in this context since the creative destruction process is inherently dynamic. However, as discussed previously at some length, measurement and interpretation problems raise questions about the comparisons of dynamic decompositions across countries. An alternative approach that is simpler and more robust is to ask the question, are resources allocated eﬃciently in a sector/ country in the cross section at a given point in time? Dynamics can also be examined here to the extent that the nature of the eﬃciency of the crosssectional allocation of businesses can vary over time.

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

71

This approach is based upon a simple cross-sectional decomposition of productivity growth developed by Olley and Pakes (1996). They note that in the cross section, the level of productivity for a sector at a point in time can be decomposed as follows: (10)

Pt (1/Nt)∑ Pit ∑ it Pit i

i

where N is the number of businesses in the sector and is the operator that represents the cross-sectional deviation of the firm-level measure from the industry simple average. The simple interpretation of this decomposition is that aggregate productivity can be decomposed into two terms involving the unweighted average of firm-level productivity, plus a cross term that reflects the cross-sectional eﬃciency of the allocation of activity. The cross term captures allocative eﬃciency since it reflects the extent to which firms with greater eﬃciency have a greater market share. This simple decomposition is very easy to implement and essentially involves just measuring the unweighted average productivity versus the weighted average productivity. Measurement problems make comparisons of the levels of either of these measures across sectors or countries very problematic, but taking the diﬀerence between these two measures reflects a form of a diﬀerence-in-diﬀerence approach. Beyond measurement advantages, this approach also has the related virtue that theoretical predictions are more straightforward as well. Distortions to market structure and institutions unambiguously imply that the diﬀerence between weighted and unweighted productivity (or equivalently the cross term) should be smaller. With these remarks in mind, figure 1.9 shows the measure of the gap between weighted and unweighted average productivity for a sample of countries. The results are obtained by applying the Olley Pakes (OP) decomposition at the industry level and then taking the weighted average across industries for the countries in the harmonized database. For virtually all countries, the gap is positive, suggesting that resources are allocated to more productive businesses. The South East Asian economies are on top, followed by the United States, while the Latin American countries (except Argentina) show higher productivity boosts through resource allocation than the EU, but lower than in Asia. The transition economies are generally weaker in terms of this measure of allocative eﬃciency. For many countries, the gap is not only positive but large. For the Asian economies and the United States, the allocative eﬃciency term accounts for about 50 percent or more of labor productivity. In the EU, the productivity boost is smaller, ranging from 15 to 38 percent. The findings in figure 1.9 are striking and suggest that this measurement approach has great potential in a cross-country context. Moreover, the allocative eﬃciency measures can be computed for diﬀerent years or for specific industries and/or other classifications of firms, suggesting that a

72

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

Fig. 1.9

The gap between weighted and unweighted labor productivity, 1990s

Note: Based on the three-year diﬀerences.

pooled country, firm-type data set of allocative eﬃciency measures would be valuable for further analysis. Note, however, that the allocative eﬃciency measures are not without problems and limitations. A key problem is that the measures by construction do not permit decomposing the contribution of entering, exiting, and continuing businesses. As such, in an analysis of the impact of institutions on reallocation and productivity dynamics, these allocative eﬃciency measures cannot be used to investigate the impact of institutions on such measures of firm dynamics and in turn, the contribution of those eﬀects on productivity. Measurement error will also cloud the interpretation of the allocative eﬃciency measures. Classical measurement error in productivity at the micro level that is uncorrelated with market share will tend to drive the allocative eﬃciency to zero. Classical measurement error in productivity that is also correlated with market share (put diﬀerently, classical measurement error in output measures at the micro level) will work in the opposite direction. 1.7 Concluding Remarks In this chapter we assess the measurement and analytic challenges for studying firm dynamics within and across countries. We use recently collected indicators of firm dynamics for a sample of more than twenty countries. Our cross-country data set has been assembled, paying great care to the harmonization of key concepts. Such harmonization is essential to conduct meaningful comparisons, but we acknowledge that our eﬀort should probably be extended, as there remain measurement problems. While simple comparisons of firm dynamics across countries remain diﬃcult to interpret, interesting inferences can be made by examining multiple indicators and by carefully considering the nature of the measurement errors. Since much of these errors are country-specific, using some form of

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

73

diﬀerence-in-diﬀerence approach that eliminates overall country-specific eﬀects helps enormously. Bearing in mind these measurement problems, there is evidence in our data of a significant heterogeneity of firms in each market and country. This heterogeneity is manifested in large disparities in firm size, firm growth, and productivity performance. More in detail, we found: • The average size of incumbent firms varies widely across sectors and countries. Diﬀerences in firm size are largely driven by within-sector diﬀerences, although in some countries sectoral specialization also plays a significant role. Smaller countries tend to have a size distribution skewed towards smaller firms, but the average size of firms does not map precisely with the overall dimension of the domestic market. An important message emerging from our analysis is that in the empirical analysis of firm dynamics, diﬀerences in the size composition across sectors and countries ought to be controlled for. • Firm churning, taken at face value, is large; gross firm turnover is in the range of 10 to 20 percent of all firms in industrial countries, and even more in transition and other emerging economies. Entering, but also exiting, firms tend to be small and thus firm flows aﬀect only about 5 to 10 percent of total employment. This may suggest that the entry of small firms is relatively easy while larger-scale entry is more diﬃcult, but survival among small firms is also more diﬃcult—many small newcomers fail before reaching the eﬃcient scale of production. Given the measurement and interpretation issues related to firm turnover data, we suggest exploring the variation in firm turnover across sectors and firms of diﬀerent sizes to shed some light on the diﬀerent nature of creative destruction. • Market selection is pretty harsh. About 20 to 40 percent of entering firms fail within the first two years of life. Confirming previous results, failure rates decline with duration; conditional on surviving the first few years, the probability of survival becomes higher. But only about 40 to 50 percent of total entering firms in a given cohort survive beyond the seventh year. • Successful entrants expand rapidly. Surviving firms are not only relatively larger but also tend to grow rapidly. The combined eﬀect of exits being concentrated among the smallest units and the growth of survivors makes the average size of a given cohort increase rapidly towards the eﬃcient scale. Measuring the post-entry performance within countries appears to be somewhat more robust than the analysis of firm dynamics, since it implies following a cohort over time within a country. • Creative destruction is important for promoting productivity growth. While the continuous process of restructuring and upgrading by incumbents is essential to boost aggregate productivity, the entry of new

74

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

firms and the exit of obsolete units also play an important role. In virtually all of our countries, the net entry process contributes positively to productivity growth. While measurement and interpretation problems associated with firm turnover cloud rank orderings across countries, within-country variations in the contribution of firm turnover to productivity growth may be an interesting avenue of research. For example, we observe a stronger contribution of net entry to productivity growth in high-technology industries compared with low-technology ones, and the diﬀerences between these two groups vary significantly across countries. This in turn may suggest a diﬀerent role of creative destruction in promoting technological adoption and experimentation. Moreover, this pattern helps highlight the usefulness of exploiting the cross-industry variation within countries and in turn comparing that cross-industry variation across countries within this context. • Allocative eﬃciency is important in productivity levels, rank ordering of countries, and in productivity growth. Allocative eﬃciency can be measured using cross-sectional data within a country or industry, or by using the covariance between market share and eﬃciency (i.e., measures of productivity). In using this measure, we find that virtually all countries exhibit positive allocative eﬃciency. Further, the rank ordering of countries on this basis appears more reasonable than other measures of the contribution of the reallocation process to growth.

References Ahn, S. 2000. Firm dynamics and productivity growth: A review of micro evidence from OECD countries. OECD Economics Department Working Paper no. 297. Paris: Organization for Economic Cooperation and Development. Audretsch, D. B. 1995. Innovation, growth and survival. International Journal of Industrial Organisation 13 (1995): 441–57. Aw, B. Y., S. Chung, and M. Roberts. 2003. Productivity, output, and failure: A comparison of Taiwanese and Korean manufacturers. Economic Journal 113 (491): F485–F510. Baily, M. N., C. Hulten, and David Campbell. 1992. Productivity dynamics in manufacturing establishments. Brookings Papers on Economic Activity: Microeconomics: 187–249. Washington, D.C.: Brookings Institution. Baldwin, J. and W. Gu. 2002. Plant turnover and productivity growth in Canadian manufacturing. OECD Science, Technology, and Industry Working Papers 2002/2. OECD Publishing. Barro, R. J., and X. Sala-i-Martin. 1995. Economic growth. New York: McGrawHill. Bartelsman, E. J. 2004. The analysis of microdata from an international perspective. OECD Statistics Directorate, STD/CSTAT, 12. Paris: Organization for Economic Cooperation and Development.

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

75

Bartelsman, E., and S. Scarpetta. 2004. Experimentation within and between firms: Any role for policy and institutions? Paper presented at the 2004 American Economic Association Meeting. 3–5 January, San Diego. Bartelsman, E. J., and M. Doms. 2000. Understanding productivity: Lessons from longitudinal microdata. Journal of Economic Literature 38 (3): 569–95. Bartelsman, E. J., J. Haltiwanger, and S. Scarpetta. 2004. Distributed analysis of firm-level data from industrial and developing countries. Mimeograph. Bastos, F., and J. Nasir. 2004. Productivity and the investment climate: What matters most? World Bank Policy Research Working Paper no. 3335. Bernard, A., and C. J. Jones. 1996a. Productivity across industries and countries: Time series theory and evidence. The Review of Economics and Statistics 78 (1): 135–46. ———. 1996b. Productivity and convergence across U.S. states and industries. Empirical Economics 21:113–35. Brown, D. J., and J. S. Earle. 2004. Economic reforms and productivity-enhancing reallocation in post-Soviet transition. Upjohn Institute Staﬀ Working Paper no. 04-98. Caves, R. E. 1998. Industrial organization and new findings on the turnover and mobility of firms. Journal of Economic Literature 36 (4): 1947–82. Davis, S. J., J. Haltiwanger, and S. Schuh. 1996. Job Creation and Destruction. Cambridge, MA: The MIT Press. Davis, S. J., and M. Henrekson. 1999. Explaining national diﬀerences in the size and industry distribution of employment. Small Business Economics 12:59–83. Dickens, W. T., and E. L. Groshen. 2003. Status of the international wage flexibility project after the authors’ conference. The Brookings Institution and the Federal Reserve Bank of New York, May. Dollar, D., M. Hallward-Driemeier, and T. Mengistae. 2003. Investment climate and firm performance in developing economies. World Bank Policy Research Working Paper no. 3323. Washington, D.C.: World Bank, Development Research Group. Dunne, T., M. Roberts, and L. Samuelson. 1989. The growth and failure of U.S. manufacturing plants. Quarterly Journal of Economics 104:671–98. ———. 1988. Patterns of firm entry and exit in U.S. manufacturing industries. RAND Journal of Economics 19 (4): 495–515. Doppelhofer, G., R. Miller, and X. Sala i Martin. 2004. Determinants of long-term growth: A bayesian averaging of classical estimates (BACE) approach. American Economic Review 94 (4): 813–35. Eslava, M., J. Haltiwanger, A. Kugler, and M. Kugler. 2004. The eﬀects of structural reforms on productivity and profitability enhancing reallocation. Journal of Development Economics 75 (2): 333–71. EUROSTAT. 1998. Enterprises in Europe, data 1994–95. Fifth Report European Commission. ———. 2004. Business demography in Europe—Results for 10 member states and Norway. Luxembourg: European Commission. Foster, L., J. Haltiwanger, and C. J. Krizan. 2001. Aggregate productivity growth: Lessons from microeconomic evidence. In New developments in productivity analysis, ed. Edward Dean, Michael Harper, and Charles Hulten pp. 303–418. Chicago: University of Chicago Press. ———. 2002. The link between aggregate and micro productivity growth: Evidence from retail trade. NBER Working Paper no. 9120. Cambridge, MA: National Bureau of Economic Research, August. Geroski, P. 1995. What do we know about entry? International Journal of Industrial Organization 13:421–40.

76

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

Griliches, Z., Regev, H. 1995. Firm productivity in Israeli industry: 1979–1988. Journal of Econometrics 65 (1): 175–203. Griﬃth, R., S. Redding, and J. Van Reenen. 2000. Mapping the two faces of R&D: productivity growth in a panel of OECD industries. Institute for Fiscal Studies Working Papers no. 2. Hallward-Driemeier, M., S. Wallsten, L. C. Xu. 2004. The investment climate and the firm: Firm-level evidence from China. World Bank Policy Research Working Paper no. 3003. Haltiwanger, J., R. Jarmin, and T. Schank. 2003. Productivity, investment in ICT and market experimentation: Micro evidence from Germany and the U.S.. Center for Economic Studies Working Paper no. 03-06. Haltiwanger, J., and H. Schweiger. 2004. Firm performance and the business climate: Where does ICA fit in? Mimeograph. Klapper, L., L. Laeven, and R. Rajan. 2006. Entry regulation as a barrier to entrepreneurship. Journal of Financial Economics 82:591–629. Levinsohn, A., and J. Petrin. 2005. Measuring industry productivity growth using plant-level data. Mimeograph. Lentz, R., and D. Mortensen. 2005. Productivity growth and worker reallocation. International Economic Review 46 (3): 731–49. Martin, R. 2005. Providing evidence based on business micro data: Methods and results. London School of Economics. Unpublished Manuscript. Micco, A., and C. Pages. 2006. The economic eﬀects of employment protection: Evidence from international industry-level data. IZA Discussion Papers no. 2433. Nicoletti, G., and S. Scarpetta. 2003. Regulation, productivity and growth: OECD evidence. Economic Policy 18 (36): 9–72. Nicoletti, G., S. Scarpetta, and O. Boylaud. 1999. Summary indicators of product market regulation with an extension to employment protection legislation. OECD Economics Department Working Paper no. 226. Paris: Organization for Economic Cooperation and Development. Olley, G. S., and A. Pakes. 1996. The dynamics of productivity in the telecommunications equipment industry. Econometrica 64 (6): 1263–97. Organization for Economic Cooperation and Development (OECD). 2004. The economic impact of ICT: Measurement, evidence, and implications. Paris: OCED. Oviedo, A. M. 2005. Doing business in developing economies: The eﬀect of regulation and institutional quality on the productivity distribution. Unpublished Manuscript. Roberts, M., and J. Tybout. 1997. Producer turnover and productivity growth in developing countries. The World Bank Research Observer 12 (1): 1–18. Scarpetta, S., ed. 2004. The sources of economic growth in OECD countries. Paris: OECD. http://ariel.sourceoecd.org/vl1234676/cl68/nw1/rpsv/cgi-bin/ fulltextew.pl?prpsv/ij/oecdthemes/99980134/v2003n1/s11/p11.idx Scarpetta, S., P. Hemmings, T. Tressel, and J. Woo. 2002. The role of policy and institutions for productivity and firm dynamics: Evidence from micro and industry data. OECD Economics Department Working Papers, no. 329. Sutton, J. 1997. Gibrat’s legacy. Journal of Economic Literature 35 (1): 40–59. World Bank. 2004. World development report: A better investment climate for everyone. Washington, D.C.: World Bank.

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

Comment

77

Timothy Dunne

Over the last twenty years, there has been a substantial growth in the empirical literature on firm dynamics. This literature has documented the tremendous churning of firms through the entry and exit process. It is now a well-established fact that industry gross entry and exit rates and the concomitant labor flows exceed net rates by a substantial amount. The impact of this churning process of firms has been examined in a number of distinct literatures. The chapter by Bartelsman, Haltiwanger, and Scarpetta is an important addition to the literature on firm dynamics and the microeconomics of productivity. First, the chapter provides a detailed comparison of the patterns of firm dynamics across a wide range of countries and focuses on the role of firm dynamics in the evolution of industry productivity across countries. Second, the chapter takes a relatively novel empirical approach to a cross-country comparison project by working with individual researchers from each country to homogenize data construction methodologies. This is important, as the measurement of firm turnover can vary markedly across countries. A main contribution of this chapter is the development of the crosscountry data set on firm dynamics. Most data on firm dynamics are generated as a by-product of a country’s administrative data collection systems or from business registers used as the basis of statistical frames in national statistics systems. The data on business dynamics are constructed by linking these cross-sectional data sources across time to create a panel structure on businesses. The definitions of what a firm is, when it is considered an entrant and an exit, how to deal with mergers and acquisitions and other such issues defining the life of the firm in the data are often determined by the administrative data collection systems (e.g., tax or unemployment insurance systems) or the nature of the data collected by the statistical agency. This creates challenges for using the systems to measure firm dynamics within an individual country but also creates challenges for comparing statistics across countries. For example, the inclusion rules for very small firms in business registers often diﬀer across countries. Since firm turnover in very small firms can be quite high, diﬀerences in inclusion rules can greatly aﬀect the firm turnover rates. One can see how such size cutoﬀ diﬀerences aﬀect firm turnover statistics by comparing the panels in figure 1.2. Alternatively, the methods that countries use to handle merger and acquisitions in various business registers can diﬀer as well. These diﬀerences in measurement aﬀect the entry and exit statistics produced and make cross-country comparisons from existing studies of firm dynamTimothy Dunne is a senior economic advisor in the Research Department at the Federal Reserve Bank of Cleveland.

78

Eric Bartelsman, John Haltiwanger, and Stefano Scarpetta

ics statistics problematic. A real contribution of this chapter is that the authors have made a serious attempt to make their cross-country data more comparable by developing a set of measurement protocols and having researchers in various countries apply these protocols to the underlying microdata. This approach is referred to in the chapter as the analysis of distributed microdata. Although diﬀerences in measurement procedures certainly remain across countries and the authors are careful to point these out, this chapter reports on the development of the most comprehensive and comparable set of cross-country industry-level statistics on firm turnover and a related set of productivity decompositions to date. Besides the basic data development contributions the chapter makes, it is also loaded with new facts about firm dynamics. In all countries, the turnover of firms (entry plus exit) greatly exceeds the net entry rates. These high turnover rates occur in large countries and small countries and in high income and moderate income countries alike. Surprisingly, a country like France—often thought to have institutions that restrict firm dynamics— has firm turnover rates similar to the United States (fig. 1.2). In fact, the United States—perceived to have low institutional barriers to the development of new firms—is usually ranked toward the middle of the distribution of countries with regard to firm turnover. Overall, industrial countries have lower firm turnover rates than less-developed countries, and manufacturing industries generally have lower turnover rates than service industries. What these striking patterns imply for thinking about the evolution of industries is that models of industry competition need to focus on equilibrium firm turnover (such as the models developed by Hopenhayn [1992] and Apslund and Nocke [2006]) and not simply on the equilibrium number of producers in a market. Firm turnover is high, and it is a persistent feature across countries and across industries. The cross-industry and cross-country turnover patterns presented in the chapter raise the question of whether the variation in country-industry turnover rates is driven primarily by industry or country eﬀects. Strong industry eﬀects suggest that industry-specific technologies are an important driver of firm turnover. Alternatively, if country eﬀects dominate, this suggests that country-specific institutional factors may play an important role. Though, to be sure, strong country eﬀects are also consistent with persistent diﬀerences in measurement procedures across countries. I analyze this issue using a simple model and the statistics presented in table 1.6 of the chapter. The model estimated is yci c i ic where yci is firm turnover in country c and industry i, c represents a set of country eﬀects, i controls for industry eﬀects, and ic is the error term. The adjusted R 2 from the model estimated with both industry and country eﬀects is .348, the adjusted R 2 with country eﬀects only is .246, and with in-

Measuring and Analyzing Cross-Country Differences in Firm Dynamics

79

dustry eﬀects only the adjusted R 2 is .079. Both sets of controls are statistically significant at conventional levels of significance. The results indicate that country eﬀects explain more of the variation in cross industry-country turnover than industry eﬀects. This finding is true if one focuses only on the industrial countries in the sample as well. This suggests the diﬀerences in turnover rates in countries are not simply driven by diﬀerences in industrial mix across countries, but that there are either systematic diﬀerences in firm turnover across countries or perhaps systematic diﬀerences in measurement. The authors are careful throughout the chapter to emphasize this latter possibility. Even with this caveat, a surprising result is the relatively low amount of the variation explained by industry controls. In a comparison of job flow data between Canada and the United States, Baldwin, Dunne, and Haltiwanger (1998) find that industry eﬀects play a dominant role in explaining cross country-industry diﬀerences in job turnover. The chapter finishes up with a set of cross-country labor productivity decompositions that show the relative importance of within-firm changes in productivity, between firm shifts in productivity and the contribution of firm turnover to overall changes in productivity. This analysis shows the novelty of the distributed microdata approach, as researchers in each country were sent computer programs to run on the microdata. The authors of the chapter only have access to a small subset of underlying microdata used in these productivity decompositions. As previous studies have found, the within-firm component dominates the between-firm component in explaining productivity growth of continuing firms in most countries. Entry and exit accounts for 20 to 50 percent of labor productivity growth across countries. Exit has the most consistent eﬀect, as the failure of low productivity firms boosts aggregate productivity in all countries. The productivity analysis illustrates the important role that firm dynamics play across a wide range of countries in the evolution of aggregate productivity growth. Overall, the chapter makes an important contribution to the empirical literature on producer dynamics. It provides many new facts and oﬀers a novel approach to analyzing cross-country data based on confidential firm and establishment-level records. References Asplund, M., and V. Nocke. 2006. Firm turnover in imperfectly competitive markets. Review of Economic Studies 73 (2): 295–327. Baldwin, J., T. Dunne, and J. Haltiwanger. 1998. A comparison of job creation and job destruction in Canada and the United States. The Review of Economics and Statistics 80 (3): 347–56. Hopenhayn, H. 1992. Entry, exit and firm dynamics in long-run equilibrium. Econometrica 60 (5): 1127–50.

II

Employment Dynamics

2 Studying the Labor Market with the Job Openings and Labor Turnover Survey R. Jason Faberman

2.1 Introduction In recent years, the Bureau of Labor Statistics (BLS) has released several new data products that describe the dynamics of the labor market. One of these is the Job Openings and Labor Turnover Survey (JOLTS). The survey is the only existing data source to measure vacancies, hires, and separations at the establishment level at a regular (monthly) frequency in the United States. The public data were released in 2002, and with its aggregate estimates, the JOLTS has already provided valuable insight on the behavior of worker recruiting and worker turnover. This chapter details the characteristics of the JOLTS data and provides some descriptive evidence at both the aggregate and establishment level. The discussion is primarily for researchers wishing to use the data in their own studies. As such, it characterizes the data scope, composition, measurement, and estimation, as well as the research potential these data have. The chapter also presents some basic evidence on the aggregate and establishment-level relations of vacancies and worker flows to state-level unemployment and other labor market conditions. The JOLTS is an evolution of earlier data series, notably the BLS Labor

R. Jason Faberman is an economist at the Federal Reserve Bank of Philadelphia. I thank Eva Nagypál, Steve Davis, John Haltiwanger, Jim Spletzer, Rick Clayton, John Wohlford, and the editors and conference participants for this volume for helpful comments on this chapter. Any remaining errors are my own. Work on this project was done while I was an employee at the Bureau of Labor Statistics. The views expressed are my own and do not necessarily reflect the positions of the U.S. Bureau of Labor Statistics, the Federal Reserve Bank of Philadelphia, or the Federal Reserve System.

83

84

R. Jason Faberman

Turnover Survey.1 The survey also builds on the research on vacancies, worker turnover, and unemployment done by Abraham (1987), Blanchard and Diamond (1989, 1990), and others, as well as theories of labor market search and matching.2 This research, and the rapidly developing research that has followed, underscores the importance of understanding labor market dynamics. As such, the BLS designed the JOLTS to capture these dynamics. The result is a high-frequency, timely survey with several major advantages over previous data. The first is its reporting of hires and separations directly by an establishment. Other data (e.g., administrative wage records, the Current Population Survey [CPS]) forced researchers to infer these flows from observed changes in a worker’s employment status. The second is its reporting of job openings or vacancies directly by an establishment. Previously, researchers had to rely on indices (such as the Help Wanted Index) for a measure of vacancies. This approach did not lend itself to studying vacancy behavior at the micro level. This was an issue because theories of labor market search often model behavior at the level of workers and firms. The final advantage is its distinction between quits and layoﬀs. The two types of separations have opposing cyclical patterns, and in general, they represent voluntary and involuntary severances, respectively. Existing research using JOLTS is currently sparse, but thanks to the ballooning of research on the theory and evidence of labor dynamics, it is expanding rapidly.3 Clark (2004) summarizes the aggregate evidence since the inception of JOLTS. Hall (2005a) and Shimer (2007a) use the JOLTS data to study whether standard theories of labor market search can match the volatility of vacancies relative to unemployment. Valetta (2005) uses the JOLTS data to study the Beveridge Curve. Besides this chapter, Davis, Faberman, and Haltiwanger (2006, 2007) and Faberman and Nagypál (2007) are the first to present analyses of the establishment-level JOLTS data. The data have also become popular with the press and various industry and policy groups. In all, the JOLTS data complement existing data and can vastly improve our understanding of the labor market. The following section defines the concepts and terminology used throughout the chapter, discusses the data sample and estimation process, and highlights the survey’s research strengths and limitations. The next section explores the relation between vacancies and unemployment at both the aggregate and establishment level. An exploration of the relations between worker flows and aggregate and local labor market conditions comes 1. The Labor Turnover Survey measured vacancies, ascensions, and separations for the manufacturing industry; the BLS discontinued the survey in 1982. See Davis and Haltiwanger (1998) and Clark and Hyson (2001) for more on this survey. 2. See, for example, Pissarides (1985) and Mortensen and Pissarides (1994). 3. Davis and Haltiwanger (1999) review the empirical work on labor dynamics, while Mortensen and Pissarides (1999) and Rogerson, Shimer, and Wright (2005) review the theoretical work on labor search.

Studying the Labor Market: the Job Openings and Labor Turnover Survey

85

next. The final section concludes and discusses potential avenues of future research. 2.2 Data and Measurement 2.2.1 Source Data The BLS uses the JOLTS data to publish monthly estimates of job openings (i.e., vacancies), hires, and separations, with separations reported as quits, layoﬀs and discharges, and other separations (e.g., retirements).4 The data start in December 2000 and are updated monthly, with the latest estimates available within two months of a month’s end. The survey covers all nonfarm establishments, the same sample frame as the Current Establishment Statistics (CES) survey. The aggregate estimates are available nationally, for four major regions and by 2-digit North American Industry Classification System (NAICS) sector.5 The BLS reports JOLTS estimates in levels and as rates. The primary unit of observation for the JOLTS survey is the establishment, which covers the operations of a firm at a single physical location. Firms can have one or more establishments. Like the CES, the JOLTS coverage of nonfarm payrolls implies that it generally excludes the selfemployed and nonprofit organizations not covered under a state unemployment insurance program. The JOLTS data are a sample of roughly 16,000 establishments surveyed each month. Establishments report their employment, hires, separations (broken out by type), and job openings for the month within the framework of the survey definitions. The survey consists of overlapping panels that are each sampled for eighteen months, and is weighted so that its employment estimates match those of the CES.6 For the analyses in this chapter, I use the JOLTS establishment data pooled over the December 2000–January 2005 period. For most aggregate statistics, I use the unrestricted sample of all observations. For the establishment-level analyses, I use a restricted sample of all observations with positive employment reported in two consecutive months. This minimizes the potential spurious eﬀects of outliers and inconsistent data reporters. The resulting sample contains 372,288 observations, which represent 92.8 percent of the pooled observations (and 92.3 percent of the pooled employment). Due to the requirement of reporting in consecutive months, the 4. The published statistics are available at http://www.bls.gov/jlt/home.htm 5. The NAICS replaces the older Standard Industrial Classification (SIC) system. The most notable change in NAICS is its classification of the service sector into several separate sectors, such as information, professional and business services, education and health, and travel and hospitality. In general, two-digit NAICS sectors correspond to major SIC industry sectors (e.g., manufacturing, services, etc.) 6. See Crankshaw and Stamas (2000) for details on the JOLTS sample weighting procedure.

86

R. Jason Faberman

restricted sample excludes the December 2000 observations.7 Results in my analyses are all sample-weighted, and where noted, also employmentweighted. Estimates are not seasonally adjusted, unless otherwise noted. 2.2.2 Concepts and Definitions The JOLTS survey form has four major data elements: employment, hires, separations, and job openings, with separations broken into three subcategories. Elements diﬀer in their timing, and their definitions are succinct in what they do (and do not) capture. These definitions are created so that BLS can optimize its measurement of changes in employment dynamics and to minimize respondent confusion in reporting. 1. Employment. Establishments report their employment for the pay period that includes the twelfth of the month. As such, it is a point-in-time measure of the employment level. An individual is counted as employed if they are on an establishment’s payroll. The reference period and definition are standard for all federal statistical establishment surveys and allows the BLS to accurately benchmark the survey to the CES. 2. Hires. Hires are new additions to the workforce of an establishment. They include new hires, rehires, seasonal and short-term hires, recalls after a layoﬀ lasting more than seven days, and transfers from other worksites. The JOLTS hires are a flow measure that is meant to capture all occurrences between the first and last day of the month. 3. Separations. Separations are subtractions from the workforce of an establishment. These removals include quits, layoﬀs lasting more than seven days, firings and other discharges, terminations of short-term and seasonal workers, retirements, and transfers to other worksites. The JOLTS separations are also a flow measure meant to capture all occurrences between the first and last day of the month. 4. Quits. Quits are the subset of separations initiated by an employee. 5. Layoﬀs and discharges. Layoﬀs and discharges are the subset of separations initiated by the employer that include all layoﬀs lasting more than seven days, firings and other discharges, and terminations of short-term and seasonal workers. 6. Other separations. Other separations include retirements, transfers, and all other separations not covered by the previous two categories. 7. Job openings (or vacancies). These are all unfilled, posted positions available at an establishment on the last day of the month. The vacancy must be for a specific position where work can start within thirty days, and an active recruiting process must be underway for the position. Vacancies are a point-in-time estimate, and its definition has two notable measurement implications. First, JOLTS does not capture vacancies for hires that 7. Even with the noted restrictions, the aggregate estimates from the unrestricted and restricted samples match each other very closely.

Studying the Labor Market: the Job Openings and Labor Turnover Survey

87

start more than a month after their posting. Second, JOLTS does not capture vacancies that are both posted and filled within the month. Note that the unemployment measure from the Current Population Survey (CPS), which is also a point-in-time measure, has a similar feature, since it must deal with individuals who both enter and leave unemployment between survey periods.8 Hires and separations are expressed as rates by dividing each by employment. The vacancy rate is slightly diﬀerent. It uses the sum of vacancies and employment in its denominator, making this rate a fraction of filled and unfilled jobs. This is analogous to the unemployment rate, which uses the labor force as its denominator (i.e., it is a fraction of employed and unemployed labor). Given the definitions of employment and worker flows, an individual who stops receiving a paycheck may not count as part of employment, but also may not count as a separation. Examples of this occurrence include teachers, temporary help workers retained but not assigned to a particular job (i.e., on call), and layoﬀs of less than seven days.9 2.2.3 Some Notes on Research with the JOLTS Data The published JOLTS data have already provided interesting evidence about the labor market, yet the survey remains relatively new and continues to evolve. The passage of time will lengthen the time series, making the survey even more useful in understanding the cyclical behavior of worker flows and vacancies. Researchers should be aware that the JOLTS sample is only representative nationwide, by major industry, and by region. With a sample size of 16,000 establishments, exploiting the data at finer industrial or geographic detail will likely face issues of precision and selection. The multiple reference periods for employment, worker flows, and vacancies can complicate some research studies (Davis, Faberman, and Haltiwanger [2007], however, have one method to deal with the timing issue). The survey does not have data on wages or other establishment characteristics, though the possibility exists for linking JOLTS data to other microdata sources, like the Quarterly Census of Employment and Wages, to obtain this information. A significant issue for JOLTS is the accurate measurement of hires and separations. Nagypál discusses some of these issues later in this volume, while Wohlford et al. (2003) and Faberman (2005) have BLS research 8. “Active recruiting” in the JOLTS is a very broad definition that includes networking and word-of-mouth recruiting. The time aggregation issue (i.e., the posting and filling of vacancies within the month) may have notable macroeconomic implications, as Shimer (2007) argues is the case with unemployment. Davis, Faberman, and Haltiwanger (2007) study the eﬀects of time aggregation on the JOLTS vacancy measure. 9. In light of this issue, the JOLTS has separate surveys for education and temporary help establishments.

88

R. Jason Faberman

Fig. 2.1

Measurement issues with labor turnover and employment

aimed at understanding and improving measurement. An important finding from this work is that the measurement of hires and separations is not as simple as theory would dictate. As noted earlier, the relations between hires, separations, and the level of employment are complicated by the fact that employed workers can exist empirically in one of two states: employed and working, or employed but not working (where working is defined as on the payroll). Other complications also exist—for instance, hires may occur months prior to the start of work.10 These nuances make hires and separations more diﬃcult to measure than a point-in-time count of employees on payroll. Figure 2.1 illustrates the possible transitions a worker can undertake (and the relative diﬃculty of measuring each) based on internal analyses by BLS program staﬀ. As one might expect, the easiest flows to measure are those where an employed and working individual either is hired or separates. Flows that deal with employed individuals not currently on payroll are where measurement diﬃculties arise, with the greatest diﬃculties occurring when an individual separates from a job match during a period of nonwork. Wohlford et al. (2003) find that separations are disproportionately harder to measure, creating an asymmetry between the measurement issues of hires and separations. Faberman (2005) further finds that contracting establishments are less likely than other establishments to respond to the survey. This asymmetry in turn results in a disparity between the CES employment trend and the cumulative diﬀerence between JOLTS hires and separations in the aggregate data. 10. The JOLTS defines a hire when the work is actually started, and asks respondents to not to count a hire until that time.

Studying the Labor Market: the Job Openings and Labor Turnover Survey

89

The BLS has taken steps (such as the creation of separate survey forms for schools and temporary help firms) to improve worker flow measurement. The BLS also continues research on JOLTS data measurement, which is obviously important for improving data quality, but can also prove useful in understanding how the employment behavior of establishments translates into the measured statistics. 2.3 Vacancies and the Beveridge Curve 2.3.1 Aggregate Relations The publicly available JOLTS estimates present a wealth of new evidence for the aggregate labor market. While the time series is relatively short, it spans a recession and slow labor market recovery, allowing researchers a glimpse of the cyclical behavior of vacancies and labor turnover. The National Bureau of Economic Research (NBER) states that a recession begins in March 2001 and ends in November of 2001, though losses in payroll employment (based on CES estimates) continue through August 2003. Figure 2.2 illustrates the aggregate behavior of vacancies and unemployment between December 2000 and January 2005. The unemployment rate estimates come from the CPS. Throughout the period, the two move in opposite directions, and the patterns are consistent with the behavior of employment growth during this period. In 2001, unemployment rises while

Fig. 2.2

Vacancy and unemployment rates, December 2000–January 2005

Source: Vacancies are from public JOLTS nonfarm estimates and unemployment is from the CPS. Both are seasonally adjusted.

90

R. Jason Faberman

vacancies fall. Unemployment rates hover around 6 percent and vacancy rates remain near 2 percent for most of 2002 and 2003. Beginning in mid2003, the unemployment rate begins to fall while the vacancy rate starts to rise; these patterns continue into the beginning of 2005. An important relation in the theory of labor search and matching is the Beveridge Curve, which predicts that the cyclical movements of vacancies and unemployment should have an inverse relation. Figure 2.3 plots the aggregate Beveridge Curve, with the JOLTS vacancy rate on the vertical axis and the CPS unemployment rate on the horizontal axis. The solid line represents the quadratic trend of the monthly vacancy-unemployment relation over the sample period. The dotted line charts the path of the vacancyunemployment relation. The labor market begins the period relatively tight, with a ratio of vacancies to unemployment of 0.85. Vacancies then fall as unemployment rises, leading to a movement downward along the trend line. This pattern continues until mid-2003, when the unemployment rate peaks and the vacancy rate reaches a trough. At this point, the ratio of vacancies to unemployment is at a low of 0.38. The relation then loops around and moves back up along the trend line, with labor market tightness increasing as a result. Given the economic downturn and recovery during this period, the evidence is consistent with the theoretical predictions of the Beveridge Curve. One can also use the aggregate JOLTS estimates to evaluate the magni-

Fig. 2.3 Vacancy vs. unemployment rates (Beveridge Curve), December 2000–January 2005 Source: Vacancies are from public JOLTS nonfarm estimates and unemployment is from the CPS. Both are seasonally adjusted. Notes: The dotted line represents the time-series path of the unemployment-vacancies relation, while the solid line represents the quadratic trend of the relation.

Studying the Labor Market: the Job Openings and Labor Turnover Survey

91

tudes, volatility, and comovement of worker flows and vacancies. Table 2.1 presents the aggregate means, standard deviations, and correlations (contemporaneous and dynamic) of vacancies, hires, and separations with relevant labor market variables (i.e., employment growth and unemployment). The vacancy rate averages 2.4 percent. It is the most volatile and persistent of the JOLTS statistics. It is strongly negatively correlated with unemployment, strongly positively correlated with hires, and to a lesser extent, positively correlated with employment growth. The dynamic correlations of vacancies to unemployment remain persistently high for both lagging and leading values, with the contemporaneous correlation being the strongest. The dynamic correlations of vacancies to net growth are significant and positive for lagging values of net growth, but insignificant, and in some cases Table 2.1

Vacancy and labor turnover aggregate summary statistics

Mean (Standard Deviation) Correlation with Unemployment (Ut) Net growth (Nt) Vacancies (Vt) Hires (Ht) Autocorrelations AR(1) AR(2) AR(3) Dynamic correlations with unemployment Ut–3 Ut–2 Ut–1 Ut Ut+1 Ut+2 Ut+3 Dynamic correlations with net growth Nt–3 Nt–2 Nt–1 Nt Nt+1 Nt+2 Nt+3

Vacancies (Vt)

Hires (Ht)

Separations (St)

Quits (Qt)

Layoffs (Lt)

0.024 (0.003)

0.033 (0.002)

0.032 (0.002)

0.018 (0.002)

0.014 (0.001)

–0.97** 0.22 1.00

–0.78** 0.54** 0.82** 1.00

–0.77** –0.29** 0.73** 0.68**

–0.93** 0.06 0.92** 0.83**

0.05 –0.75** –0.12 –0.13

0.97** 0.94** 0.90**

0.77** 0.68** 0.63**

0.78** 0.79** 0.64**

0.93** 0.91** 0.84**

0.37** 0.37** 0.00

–0.86** –0.91** –0.95** –0.97** –0.96** –0.95** –0.93**

–0.60** –0.67** –0.73** –0.78** –0.84** –0.85** –0.89**

–0.88** –0.85** –0.84** –0.77** –0.74** –0.69** –0.60**

–0.88** –0.91** –0.92** –0.93** –0.92** –0.90** –0.84**

–0.50** –0.31** –0.21** 0.05 0.09 0.16 0.24

0.43** 0.37** 0.36** 0.22 0.05 –0.14 –0.28

0.45** 0.38** 0.37** 0.54** 0.14 0.04 –0.09

0.13 –0.04 0.02 –0.29** –0.25 –0.42** –0.41**

0.31** 0.21 0.21 0.06 –0.09 –0.29** –0.39**

–0.25 –0.49** –0.37** –0.75* –0.38** –0.38** –0.18

Source: Author’s calculations based on public JOLTS and CPS aggregate data (seasonally adjusted). Notes: Net growth rates are the difference between the hires and separations rates. Statistics are based on data from December 2000 through January 2005. Asterisks (**) denote significance at the 5 percent level.

92

R. Jason Faberman

negative, for leading values of net growth, implying that growth is a good predictor of vacancies, but vacancies are not a good predictor of growth. Table 2.2 lists the summary statistics for vacancies, hires, and separations by industry and region. Vacancy rates vary considerably by industry, though industries with high worker turnover are not necessarily the industries with the highest vacancy rates. Instead, vacancy rates tend to be highest in industries with considerable expansions during the sample period, such as professional and business services, and education and health services. Education and health services have the highest vacancy rate despite also having some of the lowest turnover rates.11 Manufacturing, which underwent a large employment decline over this period, has one of the lowest vacancy rates (along with construction and resources). To a lesser extent, vacancies vary by region. In general, the South and West, which have relatively high employment growth, have higher rates of vacancies. 2.3.2 Vacancy Postings and the Local Labor Market Since the JOLTS data are collected at the establishment level, they are especially powerful for a micro-level study. Most theories of labor market search model the relation of vacancies to unemployment as the outcome of firm-level decisions of whether to post vacancies in response to current labor market conditions. Theory dictates that, controlling for outside factors, the negative aggregate relation of unemployment to vacancies should also hold at the micro level. To test this, I estimate the relation of establishment vacancy rates to local (i.e., state) unemployment rates.12 I start with the basic statistical properties of establishment-level vacancies, particularly since empirical evidence on them is sparse. Table 2.3 lists these properties for the pooled estimates of vacancy rates for establishment i in state j at month t (Vijt). The table lists separate vacancy rate statistics for all observations and for the subsample of observations with at least one vacancy reported. Statistics are employment-weighted. Only 12 percent of establishment-month observations have a vacancy posted at the end of the month, though these represent 53 percent of employment. This statistic is somewhat misleading, however, since at the monthly frequency many establishments have no net change in employment (79 percent) or hires (81 percent), and likely do not need a vacancy posting. Nevertheless, conditional on changing employment levels, only 34 percent of establishmentmonth observations (representing 67 percent of employment) have a 11. Davis, Faberman, and Haltiwanger (2007) note that the JOLTS vacancy rates tend to be higher in industries with more formal hiring practices. 12. Note that there is a timing diﬀerence in the reporting of vacancies and unemployment for a given month. Reported vacancies are those posted at the end of the month, while the unemployed are those who actively looked for work in the four weeks prior to the week of the 19th. This is true for both national and state-level unemployment. Thus, the vacancy rates used in this study will lead unemployment rates by about two weeks.

Table 2.2

Vacancy and labor turnover summary statistics by industry and region

Major industry Resources Construction Manufacturing Transportation and utilities Retail trade Information Financial activities Professional and business services Education and health Leisure and hospitality Other services Government Region Northeast Midwest South West Across-industry correlations with Net growth (Nj) Vacancies (Vj) Hires (Hj)

Vacancies (Vj)

Hires (Hj)

Separations (Sj)

Quits (Qj)

Layoffs (Lj)

Quit share (Qj /Sj)

0.011 (0.003) 0.014 (0.004) 0.014 (0.003) 0.016 (0.003) 0.019 (0.004) 0.020 (0.005) 0.021 (0.002) 0.029 (0.005) 0.033 (0.005) 0.028 (0.006) 0.019 (0.004) 0.018 (0.003)

0.031 (0.008) 0.054 (0.013) 0.022 (0.004) 0.025 (0.005) 0.044 (0.009) 0.021 (0.004) 0.022 (0.004) 0.043 (0.006) 0.027 (0.005) 0.063 (0.013) 0.032 (0.007) 0.015 (0.005)

0.031 (0.006) 0.055 (0.007) 0.027 (0.004) 0.026 (0.003) 0.043 (0.007) 0.023 (0.005) 0.023 (0.004) 0.039 (0.007) 0.023 (0.004) 0.059 (0.011) 0.032 (0.009) 0.012 (0.004)

0.013 (0.004) 0.020 (0.004) 0.012 (0.002) 0.013 (0.002) 0.027 (0.005) 0.013 (0.003) 0.013 (0.003) 0.020 (0.004) 0.015 (0.003) 0.039 (0.008) 0.019 (0.004) 0.006 (0.002)

0.013 (0.006) 0.033 (0.008) 0.012 (0.003) 0.011 (0.003) 0.013 (0.005) 0.008 (0.003) 0.007 (0.002) 0.016 (0.004) 0.007 (0.002) 0.018 (0.005) 0.011 (0.006) 0.004 (0.002)

0.421

0.021 (0.003) 0.020 (0.003) 0.023 (0.003) 0.022 (0.004)

0.029 (0.006) 0.032 (0.006) 0.035 (0.005) 0.033 (0.005)

0.028 (0.005) 0.031 (0.005) 0.034 (0.004) 0.033 (0.004)

0.014 (0.003) 0.017 (0.004) 0.020 (0.003) 0.018 (0.003)

0.012 (0.003) 0.012 (0.002) 0.012 (0.002) 0.013 (0.002)

0.498

0.74** 1.00

0.23 0.33 1.00

0.05 0.21 0.98**

0.21 0.38 0.94**

–0.20 –0.07 0.80**

0.47 0.66** 0.32

0.370 0.445 0.500 0.626 0.577 0.589 0.512 0.638 0.661 0.593 0.488

0.549 0.585 0.545

Source: Author’s tabulations from JOLTS data. Notes: Net growth rates are the difference between the hires and separations rates. Means are reported, with standard deviations in parentheses. Statistics are based on data from December 2000 through January 2005. **Significant at the 5 percent level.

94

R. Jason Faberman

Table 2.3

Local unemployment and establishment vacancy summary statistics

Mean Standard deviation Median 10th, 90th percentiles Number of observations Share of employment [Estabs.] with Vijt > 0 Share of empl. [estabs.] with Vijt > 0 | Net ≠ 0 Percent of variation explained by Month effects State effects Establishment effects

All establishments

Establishments with positive vacancies only

0.021 0.039 0.003 0.000, 0.063 372,288 0.533 [0.122] 0.674 [0.336]

0.040 0.046 0.026 0.005, 0.089 175,981 n.a.

0.5 0.7 40.7

n.a.

0.8 0.6 66.0

Source: Author’s tabulations from pooled JOLTS microdata. Notes: Estimates are based on data from December 2000 through January 2005. Estimates (except the share of establishments with positive vacancies) are weighted by employment. n.a. = not applicable.

vacancy posted at the end of the month. The vacancy rate for these observations is nearly double the rate for all observations. When looking at these statistics, remember that the JOLTS vacancy definition does not capture long-term vacancy postings or vacancies that are posted and filled within the month. Nevertheless, the statistics may reflect the fact that establishments use less formal hiring practices than vacancies with some frequency, or that some establishments may have relatively short vacancy durations. Davis, Faberman, and Haltiwanger (2007) explore these conjectures. Table 2.3 also shows that state and month diﬀerences account for less than 1 percent of the establishment-level vacancy variation. Establishment eﬀects account for 41 percent of the variation of all vacancies and 66 percent of the variation conditional on an establishment’s posting of at least one vacancy. This suggests that much of the micro-level variation stems from diﬀerent vacancy-posting behaviors among establishments rather than varying behaviors within local labor markets, or during certain points in the business cycle. To explore the relation between establishment vacancy postings and state unemployment, I regress establishment vacancy rates on state unemployment rates. The unemployment rates come from the BLS Local Area Unemployment Statistics (LAUS) data, which use the CPS and other data sources to produce its estimates. In terms of magnitudes, unemployment rates for many states are similar to the national rate, though the average rates for several states are several percentage points higher or lower than the national rate. The cyclical volatility of unemployment for some states

Studying the Labor Market: the Job Openings and Labor Turnover Survey

Fig. 2.4

95

Establishment vacancies and their relation to the local unemployment

Source: Author’s estimation of establishment vacancy rates on a fourth-order polynomial of the state unemployment rate using JOLTS establishment microdata and LAUS unemployment estimates. State and establishment fixed eﬀects are used where noted. See text for details.

also tends to be higher than the volatility at the national level. To allow for a nonlinear relation, I use a fourth-order polynomial of unemployment. Nonparametric analyses of the data (not reported here) suggest that a polynomial of this order fits the data well. I weight the regressions by employment and run separate regressions that include state and establishment fixed eﬀects.13 The predicted relations of vacancies to unemployment from these regressions are in figure 2.4. There are separate predicted trends for the unconditional relation, the relation with state eﬀects removed, and the relation with establishment eﬀects removed. As theory predicts, vacancy postings are inversely related to the local unemployment rate. The polynomial coeﬃcients for each regression are all jointly significant at the 5 percent level. The relation is steeper once I control for state or establishment eﬀects. This is likely due to the large variation in trend unemployment rates across states, suggesting that not controlling for this trend variation understates the responsiveness of vacancies to unemployment. It also suggests that the covariation of vacancies and unemployment occurs more from time variation within states than from level diﬀerences across states. Controlling for establishment rather than state eﬀects, however, makes little diﬀerence for the results. This suggests that much of the between13. Note that state fixed eﬀects are a subset of establishment fixed eﬀects, in the sense that establishments cannot change their location in the data.

96

R. Jason Faberman

establishment variation in the relation is between states, and not necessarily between establishments within states. Overall, the results suggest that a Beveridge Curve relation in fact exists at both the establishment and aggregate levels. 2.4 Worker Flows and the Labor Market 2.4.1 Aggregate Evidence I now focus on the JOLTS worker flow estimates. Figure 2.5 plots the time series of aggregate hires and separations rates over the sample period. Their patterns reflect the downturn and recovery during this time. Hires decline during the recession and remain low through mid-2003. The hiring rate then begins a gradual, steady increase though the start of 2005. Separations are high throughout most of 2001. They then decrease in early 2002, and reach a low in mid-2003. Separations then increase gradually through the end of the sample period, even though net growth is strong during this time; evidence not reported here shows that movements in the quits rate drive this increase. In figure 2.6, I plot quarterly worker flow rates calculated from the JOLTS against the gross job losses estimates from the Business Employment Dynamics (BED) program.14 Hires and gross job gains move together for the most part, though hiring has a more pronounced decline during the 2001 recession and a more pronounced rise during 2004. Gross job losses, relative to separations, show a considerably larger rise during the 2001 recession and a decline thereafter, whereas separations begin to rise again starting in mid-2003. The diﬀerence between the two series at the end of the period can be attributed to the increase in the quits rate during this time. As with vacancies, the aggregate estimates of worker flows are summarized in tables 2.1 and 2.2. Table 2.1 shows that over this period the hires rate averages 3.3 percent, while the separations rate averages 3.2 percent. More than half (54 percent) of separations, on average, are quits. Hires and separations are both negatively correlated with unemployment—the latter correlation comes primarily from a negative correlation of quits with unemployment. Layoﬀs are uncorrelated with unemployment, but strongly negatively correlated with employment growth, leading to a negative correlation between growth and total separations. Hires are positively correlated with growth, but quits are essentially uncorrelated with growth. Hires, quits, and vacancies are all strongly positively correlated with each other. Hires and quits exhibit considerable persistence, while layoﬀs exhibit little to no persistence. The latter is consistent with the notion that 14. For more on the BED, see Spletzer et al. (2004), as well as chapter 4 by Clayton and Spletzer in this volume.

Fig. 2.5

Hires and separations rates, December 2000–January 2005

Source: Public JOLTS nonfarm estimates, seasonally adjusted.

Fig. 2.6

Quarterly worker flow and job flow rates, JOLTS and BED data

Source: Quarterly worker flows are from the published JOLTS estimates and quarterly job flows are from the published BED statistics. All estimates are seasonally adjusted.

98

R. Jason Faberman

layoﬀs tend to be episodic events rather than persistent, dynamic processes. The dynamic correlations suggest that hires are a leading factor for lower future unemployment. The contemporaneous correlation between quits and unemployment is stronger than either the lagging or leading dynamic correlations. The same can be said of the contemporaneous correlation between layoﬀs and employment growth and their dynamic correlations. Because of the short sample period, one should interpret the time-series correlations with caution. Nevertheless, the patterns illustrated (particularly by quits and layoﬀs) shed some light on the cyclical behavior of worker flows. Hires and quits are clearly procyclical, though the latter are more related to unemployment than job growth. Layoﬀs, on the other hand, are countercyclical, but only with respect to job growth—they have little relation to the stock of unemployment. This evidence has implications for the recent debate on whether recessions are primarily periods of high job loss or reduced hiring. Hall (2005b) and Shimer (2007b) argue that the job-finding rate, and not necessarily the separations rate, drives cyclical movements in unemployment. The correlations in table 2.1 support that claim, but only to the extent that movements in the quits rate drives the relationship between separations and unemployment. This suggests that separations and the job-finding rate are not mutually exclusive, and that the relative importance of separations versus the job-finding rate may depend critically on the cyclical behavior of employer-to-employer transitions (described by Fallick and Fleishmann 2004; Nagypál 2005) since quits tend to dominate these flows. Table 2.2 illustrates that worker flow patterns vary widely by industry and, to a lesser extent, by region. The industry evidence is consistent with the findings of Anderson and Meyer (1994) and Burgess, Lane, and Stevens (2000). Turnover is highest in seasonal industries, such as construction, leisure, and hospitality, and low in other industries, such as manufacturing and government. Turnover is also slightly higher in the South and West than in the Northeast and Midwest. Industries and regions also vary widely in the share of their separations accounted for by quits. The majority of separations tend to be layoﬀs in goods-producing industries (resources, construction, manufacturing) and quits in other industries, such as services and retail trade. A large fraction of separations in the Northeast and Midwest, where shares of goods-producing industries are relatively high, are layoﬀs. The across-industry correlations suggest that both vacancies and growth are positively related to the share of separations made up by quits. Intuitively, expanding industries should have less layoﬀs, all else equal. The correlations also illustrate that high-turnover industries tend to have high rates of hires, quits, and layoﬀs.

Studying the Labor Market: the Job Openings and Labor Turnover Survey

99

2.4.2 Worker Flows and Establishment Growth Hires, quits, and layoﬀs are the result of continuous, dynamic interactions between workers and firms. In any period, a worker with a better job oﬀer may choose to quit a successful, expanding firm at the same time a declining firm looks to hire new employees as it restructures its workforce. Anecdotal evidence of such occurrences is quite common. Yet, even with aggregate data on labor turnover, it is diﬃcult to know what role, if any, such interactions play in the cyclical behavior of hires and separations. Another advantage of the JOLTS microdata is its ability to illustrate the relationship between establishment-level employment behavior and the aggregate behavior of worker flows. When, how, and to what extent establishments create or destroy jobs has been a topic of research for nearly two decades (e.g., Dunne, Roberts, and Samuelson 1989a, 1989b; Davis and Haltiwanger 1990, 1992). Evidence from this research shows that large fractions of establishments simultaneously create and destroy jobs each period. There is little evidence, though, on the relation between these establishment-level decisions and patterns of worker turnover. To explore this relation, I split the JOLTS microdata into three groups: establishments with expanding employment (i.e., more hires than separations); establishments with contracting employment (i.e., more separations than hires); and establishments with constant employment (i.e., either oﬀsetting hires and separations or no turnover at all). I then calculate the aggregate monthly labor turnover estimates for each group, using factors calculated from the public JOLTS estimates to seasonally adjust the data. Figures 2.7 and 2.8 show the patterns of hires and separations, respectively, by type of employment change. The figures show analogous pictures. Expanding establishments have high hires rates, while contracting establishments have high separations rates. These rates are also considerably more volatile than the other labor turnover series, with standard deviations that are between 1.5 and 3.6 times greater than those for the other groups. Establishments with no employment change have the lowest hires and separations rates. Their rates are also the least volatile. This evidence suggests that the relation of establishment-level hires and separations to net growth is nonlinear and nonmonotonic—contracting establishments have more hires and expanding establishments have more separations than establishments with no employment change. Finally, even though figure 2.5 shows a long, persistent drop in hiring during the downturn and a mild pickup in separations during the recession, the series depicted in figures 2.7 and 2.8 show little to no cyclical variation—the only exception is a moderate movement of the separations rate among contracting establishments during the 2001 recession and during the 2003–2004 recovery period. How can the evidence in the two figures be reconciled? As Davis, Faberman, and

Fig. 2.7

Hiring rates by type of establishment-level employment change

Source: Author’s tabulations of JOLTS microdata. Estimates are seasonally adjusted using factors from the aggregate public estimates.

Fig. 2.8

Separation rates by type of establishment-level employment change

Source: Author’s tabulations of JOLTS microdata. Estimates are seasonally adjusted using factors from the aggregate public estimates.

Studying the Labor Market: the Job Openings and Labor Turnover Survey

Fig. 2.9

101

Quit rates by type of establishment-level employment change

Source: Author’s tabulations of JOLTS microdata. Estimates are seasonally adjusted using factors from the aggregate public estimates.

Haltiwanger (2006) illustrate, cyclical shifts in the distribution of establishment growth account for the diﬀerences between the figures.15 Figures 2.9 and 2.10 show two notable caveats for quits and layoﬀs. In figure 2.9, the quits rate mimics the procyclical behavior of its aggregate estimates among contracting establishments and, to a lesser extent, among expanding establishments. In figure 2.10, layoﬀs among contracting establishments exhibit a mild spike in late 2001, but are otherwise acyclical. Table 2.4 summarizes worker flow rates for diﬀerent intervals of the growth rate distribution. Quit rates exceed layoﬀ rates for all but the largest contractions, but remain relatively high for all contracting establishments. Only job losses at establishments with large contractions are dominated by layoﬀs. Finally, there is an asymmetry between the tails of the growth rate distribution: separations at expanding establishments are considerably higher than hires at rapidly contracting establishments. This may suggest that a shakeout process within the hiring patterns of expanding establishments exists, but further research is warranted. 2.4.3 Worker Flow Relations to the Local Labor Market Understanding how worker flows relate to local labor market conditions can be an important aspect of understanding their aggregate movements. 15. Davis, Faberman, and Haltiwanger (2006) also note that the patterns illustrated are robust to size, industry, and establishment controls.

102

R. Jason Faberman

Fig. 2.10

Layoﬀ rates by type of establishment-level employment change

Source: Author’s tabulations of JOLTS microdata. Estimates are seasonally adjusted using factors from the aggregate public estimates.

Table 2.4 Net growth interval (Nijt) (–2, –0.3) [–0.3, –0.1) [–0.1, 0) 0 (0, 0.1) [0.1, 0.3) [0.3, 2)

Labor turnover rates by establishment growth rate interval Hiring rate (Hijt)

Separations rate (Sijt)

Quits rate (Qijt)

Layoffs rate (Lijt)

0.018 0.028 0.017 0.011 0.042 0.199 0.541

0.554 0.191 0.039 0.011 0.019 0.037 0.034

0.132 0.089 0.023 0.008 0.013 0.024 0.020

0.393 0.088 0.013 0.003 0.005 0.017 0.013

Source: Author’s tabulations from pooled JOLTS microdata. Notes: Estimates are based on data from December 2000 through January 2005. Estimates are weighted by employment.

One basic yet important question the JOLTS microdata can address is how do local worker flow rates relate to the local unemployment rate? Table 2.5 reports the basic relations of pooled establishment-month observations of hires (Hijt), quits (Qijt), and layoﬀs and discharges (Lijt) to state-level labor market statistics. These statistics include the state unemployment rate, its change from the previous month (Ujt), and the state employment growth rate (Njt) (obtained from the CES). The reported correlations appear very weak, yet nearly all are significant at the 5 percent

Studying the Labor Market: the Job Openings and Labor Turnover Survey

Table 2.5

103

Establishment labor turnover variation and local labor market conditions

Pooled correlation with Net growth rate (Njt) Unemployment (Ujt) Unemployment change (ΔUjt) Percent of variation explained by Establishment effects State × month effects

Hiring rate (Hijt)

Quits rate (Qijt)

Layoffs rate (Lijt )

0.026** –0.025** –0.012**

0.008** –0.036** –0.010**

–0.009** 0.001 0.009**

28.5 1.9

27.9 2.2

21.0 1.1

Source: Author’s tabulations from pooled JOLTS microdata (worker flows), supplemented by LAUS state data (unemployment), and CES state data (net growth). Notes: Estimates are based on data from December 2000 through January 2005. All estimates are weighted by employment. The variations explained are from the regression of each worker flow estimate on either 14,573 establishment effects or 1,887 state × month effects. **Significant at the 5 percent level.

level. This is a consequence of using pooled establishment observations, which tend to have large idiosyncratic components to their variation regardless of the variable examined. Therefore, the most relevant characteristics of these correlations are their sign and their magnitudes relative to each other.16 Establishment fixed eﬀects only explain between 21 and 29 percent of the variations of these flows; state-month eﬀects explain 1 to 2 percent. The evidence suggests a procyclical pattern for establishment hires and quits and a countercyclical pattern for layoﬀs—higher growth, lower unemployment, and decreases in unemployment at the state level are related to more hires and more quits. Layoﬀs are negatively related to growth and positively related to increases in unemployment, but consistent with the national evidence, they are essentially uncorrelated with the unemployment rate. I also estimate the establishment-level relations of hires, quits, and layoﬀs to the change in the state unemployment rate. I focus on the change rather than the level because it is more comparable to a flow measure.17 In the previous section, vacancies were a stock measure, so the level of unemployment was the appropriate metric. I regress each establishment-month observation on a fourth-order polynomial of Ujt, weighting the regressions by employment separately for each of the three labor turnover rates.18 16. Ideally, I would calculate state-level worker flow estimates and use them to estimate the reported correlations. Unfortunately, the JOLTS sample size and weighting structure do not allow for reliable estimates below the detail of its four geographic regions. 17. Note that the change in unemployment is the net eﬀect of the flows into unemployment and flows out of unemployment. 18. As with the regressions of section 2.4, the fourth-order polynomial results are consistent with similar nonparametric fits of the data.

104

R. Jason Faberman

Fig. 2.11 Establishment hirings and their relation to changes in local unemployment Source: Author’s estimation of establishment vacancy rates on a fourth-order polynomial of the state unemployment rate using JOLTS establishment microdata and LAUS unemployment estimates. State and establishment fixed eﬀects are used where noted. See text for details.

As before, I perform separate regressions for the unconditional relation, the relation with state eﬀects removed, and the relation with establishment eﬀects removed. Figures 2.11, 2.12, and 2.13 plot the results for hires, quits, and layoﬀs, respectively. Figure 2.11 shows that establishments hire less when the local unemployment rate is rising. The relation is nonlinear, with hires changing the most during large decreases in unemployment. Figure 2.12 shows that quits also decrease as unemployment rises. This relationship is also nonlinear, with quits changing the most during large increases in unemployment. In Figure 2.13, layoﬀs increase with increases in local unemployment. The relationship is close to linear. This establishment-level evidence parallels the patterns in the aggregate evidence. Controlling for state or establishment fixed eﬀects does not alter these results. 2.5 Conclusions and Further Research Potential The JOLTS data provide a wealth of labor market information at both the aggregate and establishment level. The data are the most comprehensive data source for vacancies in the United States, and have the timeliest, most frequent, and most direct measures of worker turnover. While its time series is still relatively short, the JOLTS already presents rich new evidence

Fig. 2.12

Establishment quits and their relation to changes in local unemployment

Source: Author’s estimation of establishment vacancy rates on a fourth-order polynomial of the state unemployment rate using JOLTS establishment microdata and LAUS unemployment estimates. State and establishment fixed eﬀects are used where noted. See text for details.

Fig. 2.13

Establishment layoﬀs and their relation to changes in local unemployment

Source: Author’s estimation of establishment vacancy rates on a fourth-order polynomial of the state unemployment rate using JOLTS establishment microdata and LAUS unemployment estimates. State and establishment fixed eﬀects are used where noted. See text for details.

106

R. Jason Faberman

on the time-series and cross-sectional patterns of these statistics. Vacancies, hires, and quits all exhibit persistent, procyclical behavior between 2001 and 2005, while layoﬀs exhibit an episodic, countercyclical pattern. Vacancies also exhibit a cyclical relation to unemployment consistent with the Beveridge Curve. The micro-level estimates provide several new insights into the behavior of vacancies and worker flows. Establishment-level vacancy postings are negatively related to local unemployment rates, suggesting that the Beveridge Curve relation holds even at the micro level. This result holds even though many establishments (even the ones who change their employment) often do not post vacancies. Expanding establishments have high hiring rates while contracting establishments have high separation rates. Establishments with no change exhibit a steady pattern of turnover, but have the lowest worker flow rates. The evidence suggests nonlinear, nonmonotonic relations of hires and separations to establishment growth. Finally, the evidence suggests that hires are strongly related to changes in local unemployment rates, falling nonlinearly with increases in unemployment. Quits also fall with increases in the local unemployment rate, while layoﬀs rise with these increases. These findings barely scratch the surface of what the JOLTS data can say about the labor market. I highlight three areas where the aggregate estimates and microdata can aid labor market research. The first is how firms use vacancies to attract workers. Earlier works, such as Abraham (1987) and Blanchard and Diamond (1989, 1990), study vacancies and their relation to unemployment using estimates from the Help Wanted Index. The JOLTS vacancy data has a major advantage over this index (and others like it) in that it is reported directly by establishments. This provides a representative, tangible measure of job openings and allows micro-level studies of vacancy posting behavior similar to previous work by Holzer (1994) and current work by Davis, Faberman, and Haltiwanger (2007) and Faberman and Nagypál (2007). Evidence in this chapter already suggests that the micro patterns of firms who post vacancies may diﬀer from existing theories of their behavior. The second area of potential research deals with separations and job loss. The JOLTS data can provide a better understanding of separations since it diﬀerentiates between quits and layoﬀs. This is important for macroeconomic analyses of employment adjustment, since quits are procyclical, while layoﬀs are countercyclical. The distinction between quits and layoﬀs and its importance for labor market movements is highlighted by the models of Akerlof, Rose, and Yellen (1988) and McLaughlin (1991), and the importance of this distinction is evident in the recent debate on whether recessions are hiring-driven, as argued by Hall (2005b) and Shimer (2007), or job-loss driven, which was the conventional wisdom. A final area of potential research deals with worker turnover more broadly. The aggregate national, regional, and industry estimates already

Studying the Labor Market: the Job Openings and Labor Turnover Survey

107

present many new findings. Future work with these and the micro-level estimates can build on the earlier work of Anderson and Meyer (1994), Burgess, Lane, and Stevens (2000), and others. The existence of vacancy, employment, and worker flow data reported by each establishment allows a micro-level study of their joint behavior that was previously impossible, but is essential for evaluating theories of labor market search and the matching of workers to firms. Overall, the JOLTS data provide many opportunities to increase our understanding of labor market dynamics.

References Abraham, K., and M. Wachter. 1987. Help wanted advertising, job vacancies, and unemployment. Brookings Papers on Economic Activity, Issue no. 1:207–43. Akerlof, G. A., A. K. Rose, and J. L. Yellen. 1988. Job switching and job satisfaction in the U.S. labor market. Brookings Papers on Economic Activity, Issue no. 2:495–594. Anderson, P., and B. R. Meyer. 1994. The extent and consequences of job turnover. Brookings Papers on Economic Activity, Microeconomics: 177–249. Blanchard, O. J., and P. Diamond. 1989. The Beveridge curve. Brookings Papers on Economic Activity, Issue no. 2:1–60. ———. 1990. The cyclical behavior of the gross flows of U.S. workers. Brookings Papers on Economic Activity, Issue no. 2:85–143. Burgess, S., J. I. Lane, and D. Stevens, 2000. Job flows, worker flows, and churning. Journal of Labor Economics 18 (3): 473–502. Clark, K. A. 2004. The job openings and labor turnover survey: What initial data show. Monthly Labor Review 127 (11): 14–23. Clark, K. A., and R. Hyson. 2001. New tools for labor market analysis: JOLTS. Monthly Labor Review 124 (12): 32–37. Crankshaw, M., and G. Stamas. 2000. Sample design in the job openings and labor turnover survey. 2000 Proceedings of the Annual Statistical Association. CDROM. Alexandria, VA: American Statistical Association. Davis, S. J., R. J. Faberman, and J. C. Haltiwanger. 2006. The flow approach to labor markets: New evidence and micro-macro links. Journal of Economic Perspectives 20 (3): 3–24. ———. 2007. The Establishment-level behavior of vacancies and hiring. Working Paper. Davis, S. J., and J. C. Haltiwanger. 1990. Gross job creation and destruction: Microeconomic evidence and macroeconomic implications. NBER Macroeconomics Annual 5:123–68. ———. 1992. Gross job creation, gross job destruction and employment reallocation. Quarterly Journal of Economics 107 (3): 819–63. ———. 1998. Measuring gross worker and job flows. In Labor Statistics Measurement Issues, ed. J. Haltiwanger, M. E. Manser, and R. Topel, 79–119. Chicago: The University of Chicago Press. ———. 1999. Gross job flows. In Handbook of labor economics, volume 3, ed. Orley Ashenfelter and David Card, 2711–2805. Amsterdam: Elsevier. Dunne, T., M. J. Roberts, and L. Samuelson. 1989a. Plant turnover and gross em-

108

R. Jason Faberman

ployment flows in the U.S. manufacturing sector. Journal of Labor Economics 7 (1): 48–71. ———. 1989b. The growth and failure of U.S. manufacturing plants. Quarterly Journal of Economics 104 (4): 671–98. Faberman, R. J. 2005. Analyzing the JOLTS hires and separations data. 2005 Proceedings of the Annual Statistical Association. CD-ROM. Alexandria, VA: American Statistical Association. Faberman, R. J., and É. Nagypál. 2007. The eﬀect of quits on worker recruitment: Theory and evidence. Working Paper. Fallick, B., and C. A. Fleischman. 2004. Employer-to-employer flows in the U.S. labor market: The complete picture of gross worker flows. Board of Governors of the Federal Reserve System (U.S.), Finance and Economics Discussion Series, paper no. 2004-34. Hall, R. E. 2005a. Employment fluctuations with equilibrium wage stickiness. American Economics Review 95 (1): 50–65. ———. 2005b. Job loss, job finding, and unemployment in the U.S. economy over the past fifty years. In 2005 NBER Macroeconomics Annual, ed. Mark Gertler and Kenneth Rogoﬀ, 101–37. Cambridge, MA: National Bureau of Economic Research. Holzer, H. J. 1994. Job vacancy rates in the firm: An empirical analysis. Economica 61 (1): 17–36. McLaughlin, K. J. 1991. A theory of quits and layoﬀs with eﬃcient turnover. Journal of Political Economy 99 (1): 1–29. Mortensen, D. T., and C. A. Pissarides. 1994. Job creation and job destruction and the theory of unemployment. Review of Economic Studies 61 (3): 397–415. ———. 1999. New developments in models of search in the labor market. In Handbook of Labor Economics, Volume 3, ed. Orley Ashenfelter and David Card, 2567–628. Amsterdam: Elsevier. Nagypál, E. 2005. On the extent of job-to-job transitions. Northwestern University, Working Paper. Pissarides, C. 1985. Short-run equilibrium dynamics of unemployment, vacancies, and real wages. American Economic Review 75 (4): 676–90. Rogerson, R., R. Shimer, and R. Wright. 2005. Search-theoretic models of the labor market: A survey. Journal of Economic Literature 43 (4): 959–88. Shimer, R. 2007a. Reassessing the ins and outs of unemployment. NBER Working Paper no. 13421. Cambridge, MA: National Bureau of Economic Research, September. Shimer, R. 2007b. The cyclical behavior of equilibrium unemployment and vacancies. American Economic Review 95 (1): 25–49. Spletzer, J. R., R. J. Faberman, A. Sadeghi, D. M. Talan, and R. L. Clayton. 2004. Business employment dynamics: New data on gross job gains and losses. Monthly Labor Review 127 (4): 29–42. Valetta, R. 2005. Why has the U.S. Beveridge Curve shifted back? New evidence using regional data. Federal Reserve Bank of San Francisco Working Paper no. 2005-25. Wohlford, J., M. A. Phillips, R. Clayton, and G. Werking. 2003. Reconciling labor turnover and employment statistics. 2003 Proceedings of the Annual Statistical Association. CD-ROM. Alexandria, VA: American Statistical Association.

3 What Can We Learn About Firm Recruitment from the Job Openings and Labor Turnover Survey? Éva Nagypál

The Job Openings and Labor Turnover Survey (JOLTS) has quickly captured the attention of macroeconomists studying labor markets after the survey’s launch in December 2000. The enthusiasm of macro-labor economists about JOLTS is easy to understand: job openings (more commonly referred to as vacancies) play a crucial role in equilibrium models of unemployment that have been developed in the 1980s and 1990s. These models (following the pioneering work of Diamond [1981, 1982a, 1982b], Mortensen [1982a, 1982b], and Pissarides [1984, 1985]) have proved to be very fruitful in analyzing a wide range of aggregate labor-market issues: the existence of unemployment as an equilibrium phenomenon, the ongoing high rate of worker reallocation observed in labor markets, or the eﬀect of policies that influence the operation of labor markets. Data on vacancies comparable to the series available in JOLTS had never been collected previously in the United States. Moreover, not only did JOLTS provide a much-needed superior measure of vacancies, it did so at a time when research on models emphasizing the role of vacancies has been very active. In addition, the JOLTS series had the unintended fortunate timing of beginning just as the long expansion of the 1990s was coming to an end. Capturing the state of the labor market just before the start of the 2001 recession thus allowed JOLTS to be informative about cyclical variation with a relatively short time span. In light of these facts, it is hard to overstate the enthusiasm of the macro-labor research community in response to the availability of the JOLTS. Faberman’s chapter in this volume (chapter 2) is an excellent overview of this new data source and should be on the reading list of anyone wishing to work with the JOLTS data. Éva Nagypál is an assistant professor of economics at Northwestern University.

109

110

Éva Nagypál

Despite the enthusiasm that the launch of JOLTS has created, this new data source has not yet been closely scrutinized to determine how it can be used to validate prevailing theories of recruitment. My chapter’s intent is to push the discourse on JOLTS in this direction. I start by reviewing some methodological and conceptual issues that arise when using JOLTS data. In particular, I first discuss the issue of labor turnover measurement and the problem of missing separations in the JOLTS data. I then discuss how the JOLTS definition of vacancies relates to the definition of vacancies used in theoretical models, and highlight how possible discrepancies between the definitions need to be taken into account when doing empirical work using JOLTS data. In the second part of the chapter, I use the publicly available JOLTS data to study empirically the widely used theoretical construct of the matching function. This allows me to demonstrate one of the many ways that the JOLTS data can serve to test existing theories of labor-market dynamics and provide new evidence to inform the development of these models in new directions. Throughout, the concepts and definitions that I use are equivalent to those used in Faberman’s chapter, though I limit myself to using the publicly available aggregate and industry data. 3.1 Consistency of JOLTS Turnover Data A distinct advantage of JOLTS is that it directly measures gross worker flows from the employer perspective (i.e., hires and separations) as opposed to simply measuring net employment change at establishments. Thus, the JOLTS gives a richer picture than available from other data sources about the margins that firms use to adjust their level of employment. There is, of course, a tight relationship between hires, separations, and net employment change at the level of an establishment, since, by definition, ejt ejt1 ejt hjt sjt where ejt is the level of employment at establishment j at the beginning of period t, and hjt and sjt are the number of hires and separations at establishment j during period t. Summing over all establishments in some set J (for example, the set of all nonfarm establishments, or the set of establishments in a particular industry) gives two alternative measures of employment growth over period t: (1)

e1Jt ∑ejt = ∑ejt1 ∑ejt eJt1 eJt j∈J

(2)

j∈J

j∈J

e ∑hjt ∑sjt hJt sJt 2 Jt

j∈J

j∈J

where eJt is the level of employment across all establishments in J at the beginning of period t, and hJt and sJt are the total number of hires and sepa-

Firm Recruitment and the Job Openings and Labor Turnover Survey

111

rations at all establishments in J during period t. Equation (1) gives a way to measure aggregate employment growth using employment data, the best measure of which, for the same universe of establishments as the one covered by JOLTS, is given by the Current Employment Survey (CES). This measure of employment growth can then be compared with the aggregate employment growth calculated using equation (2) based on labor turnover data in JOLTS, giving a way to assess the consistency of the JOLTS turnover data. To the extent that the JOLTS and the CES cover the same universe of establishments and the JOLTS weighting scheme is explicitly adjusted to match the CES level of employment, the correspondence between the two measures of employment growth should be very close. Beyond sampling error, there is only one reason that the correspondence between the two measures of employment growth cannot be expected to hold month by month—the diﬀerence in reference periods. The JOLTS turnover data refer to the period between the first day of the month and the last day of the month, while employment in the CES measures employment during the pay period that includes the twelfth of the month. Calculating employment growth over horizons longer than a month, however, should diminish both the eﬀect of any sampling error and the eﬀect of the diﬀerence in the reference period. Figure 3.1 plots aggregate employment growth from December 2000 onwards calculated from the CES data and from the JOLTS data using equations (1) and (2). According to the CES data, total employment declined by 59,000 workers in the United States between December 2000 and December 2004, which is in line with the poor employment performance of the U.S. economy during and following the 2001 recession. At the same time, according to the JOLTS data, the number of employed grew by 4.64 million during the same period, representing over 3.5 percent of total employment. This is a large discrepancy. To the extent that (a) the CES is a much larger survey that is designed explicitly to determine the level of employment in the United States and (b) the stock of employment is easier to measure than the flow into and out of employment, one can attribute all the discrepancy between the two measures of employment growth to measurement problems in the JOLTS turnover data. This discrepancy has been identified earlier by Wohlford et al. (2003). In fact, as a result of internal studies by BLS staﬀ that uncovered the same discrepancy, there have been some changes in 2002 in the way the JOLTS data were collected, with the survey instrument redesigned for schools and temporary help agencies. These changes have reduced the size of the above discrepancy, but have not eliminated it. To show this, figure 3.2 plots aggregate employment growth for four year-long periods based on the CES and the JOLTS data. The overstatement of employment growth by JOLTS was the largest early in the survey, between December 2000 and December 2001 (2.29 million), but it

Fig. 3.1 Aggregate employment growth in the JOLTS and in the CES data since December 2000

Fig. 3.2 Aggregate employment growth in the JOLTS and in the CES data since the beginning of the year for each year between 2001 and 2004

Firm Recruitment and the Job Openings and Labor Turnover Survey

113

remained positive in all subsequent years; it was 0.57 million between December 2001 and December 2002, 1.00 million between December 2002 and December 2003, and 0.84 million between December 2003 and December 2004. Moreover, the aggregate annual employment growth discrepancy of 0.7 percent for 2003–2004 masks substantial industry variation in annual em2 1 ployment growth discrepancy (measured as 1/2 ΣDec2004 tJan2003(eit – eit) for industry i), which is plotted on the vertical axis of figure 3.3. As can be seen for 2003–2004, the annual overstatement of employment growth by JOLTS varies from a high of 2.58 percent in the Federal Government to a low of – 3.13 percent in construction. This large industry variation implies that the mismeasurement of labor turnover in the JOLTS is a larger problem than seems at first from the aggregate data. There is reason to believe that the discrepancy in the JOLTS arises in large part due to the mismeasurement of the separation rate. To show this, I calculated for each two-digit North American Industry Classification System (NAICS) industry the average JOLTS separation rate for the period January 2003–December 2004 and the average separation rate from the Current Population Survey (CPS) for the same period.1 On average, the separation rate calculated from the CPS is 1.9 times as large as the separation rate calculated from JOLTS. This is due both to the understatement of separations in JOLTS and to the overstatement of separations in the CPS due to the well-known classification problem (Nagypál 2006). There is large cross-industry variation in the ratio of the JOLTS to the CPS separation rate, however, ranging from the JOLTS separation rate being a third of the CPS separation rate in education to two-thirds in mining. Moreover, it is exactly the industries that have a very low measured JOLTS separation rate compared to the CPS separation rate that have the largest overstatement of their employment growth in the JOLTS hires and separation data. This can be seen from figure 3.3, where I plot the average annual employment growth discrepancy for the period January 2003–December 2004 between the JOLTS and the CES against the ratio of the JOLTS separation rate to the CPS separation rate for each industry. This evidence is suggestive that the understatement of the separation rate is a key reason that the JOLTS data overstate employment growth in the U.S. economy. Further examination of the JOLTS employment growth discrepancy across industries also reveals that a relevant characteristic of industries that is correlated with the size of this discrepancy is the average level of 1. The CPS started using the NAICS industry classification of the JOLTS after January 2003. The separation rate in the CPS can be derived by matching the Basic Monthly Survey across two consecutive months and calculating the ratio of the sum of employer-to-employer and employment-to-nonemployment transitions between the two months to the number of employed workers during the first month.

114

Éva Nagypál

Fig. 3.3 Annual employment growth discrepancy between the JOLTS and the CES plotted against the JOLTS separation rate as a fraction of CPS separation rate by industry

education in the industry. Figure 3.4 plots the average years of education in each two-digit NAICS industry, calculated using CPS data from 2003– 2004 against the average annual employment growth discrepancy for the period January 2003 to December 2004 between the JOLTS and the CES. Clearly, this figure implies that the overstatement of employment growth is a larger problem for more educated workers, a pattern that is worthy of further investigation and that could inform future revisions of JOLTS data collection. To assess the impact of the employment growth discrepancy between the JOLTS and the CES on the measurement of labor turnover, I use a simple procedure to adjust hires and separations for this discrepancy by industry according to h˜ it hit max (0, e1it e2it) ˜sit sit max (0, e2it e1it) where hit (sit) and h˜it (s˜it) is the measured and adjusted number of hires (separations) in industry i in month t, respectively. To do this adjustment, I estimate the employment growth for month t for industry i in the CES by extrapolating the employment numbers for the pay period containing the twelfth of the month. To the extent that this adjustment merely requires that employment growth numbers match up industry-by-industry at the two-digit level as opposed to establishment-by-establishment, this procedure underadjusts the hires and separations numbers, thus giving a lower

Firm Recruitment and the Job Openings and Labor Turnover Survey

115

Fig. 3.4 Annual employment growth discrepancy between the JOLTS and the CES plotted against the average level of education by industry

bound on the true hiring and separation rate.2 This procedure results in an adjusted aggregate hiring rate of 3.62 percent as opposed to the measured hiring rate of 3.31 percent, and in an adjusted aggregate separation rate of 3.62 percent as opposed to the measured separation rate of 3.23 percent, a significant change. 3.2 What do the JOLTS Job Openings Measure? Beyond giving a more detailed view of labor turnover, a distinct advantage of JOLTS is that it provides information on the number of job openings for a representative sample of U.S. establishments, thereby giving a much more direct measure of vacancy creation in the U.S. economy than was previously available (primarily through the use of the Help Wanted Advertising Index). Of course, to develop a measure of job openings, the BLS had to construct an appropriate empirical definition. Faberman reviews this definition in chapter 2. Here, I would like to discuss the impact of two choices in the construction of this empirical definition: the choice to measure the stock of vacancies at a point in time as opposed to their flow during a period, and the choice to include only vacancies for positions that can start within thirty days. 2. At the same time, given that the employment growth number is estimated using extrapolation and could contain errors, this procedure could possibly overadjust the hires and separations numbers.

116

Éva Nagypál

To focus the discussion, consider the following simple continuous-time model of vacancy creation, where time is measured in months. Assume that a firm wishes to hire someone to start working at some known future date ts. Due to search frictions in the relevant labor market, appropriate candidates are not always immediately available for hire; rather, they arrive to the firm at random times if the firm has a vacancy open. In particular, assume that if the firm has a vacancy open, suitable candidates arrive at Poisson rate , which (approximately) means that during a short period of length , the probability that a suitable candidate shows up is . Assume that hiring a candidate at time te ts has a cost of ce (ts – te) to the firm. Such a cost could arise due to having to incur some expenses to keep the candidate available between the time he or she is oﬀered the position at time te and the time he or she starts working at time ts. Assume that hiring a candidate at time td ts has a cost of cd (td – ts) to the firm. Such a cost could arise due to forgone profits from starting the position late. Finally, assume that the firm chooses the time to open a vacancy to minimize the expected cost of hiring too early or too late compared to time ts. Under these assumptions, one can show3 that the firm will optimally open the vacancy at time tv ts – l, where l, the lead time to open a vacancy, is given by log (1 rd) l(, rd)

where rd cd /ce is the relative cost of delay. This simple model has the intuitive implication that the harder it is to find a suitable candidate (i.e., the lower is ) and the higher is the cost of delay relative to the cost of early hiring, the earlier will the firm decide to open a vacancy relative to the time of the intended start of the job. To see why this simple model is useful to think about the measurement of job openings in the JOLTS, assume that at each point in time firms wish to hire a fixed measure of workers. Then one can calculate the probability that a vacancy that is open at some point during the month [to – 1, to ] is observed at time to (without any restrictions on when the position starts) to be 1 Pu () . 1 The solid line in figure 3.5 plots this probability of observing a vacancy as a function of . The interpretation of this probability is simple: jobs with a higher arrival rate of suitable applicants have a vacancy open for a shorter period of time, hence these vacancies have a lower probability of being observed given a fixed frequency of observation. This is a well-known issue in duration models—whenever duration events are sampled using stock sampling (as in the JOLTS), events of short duration are less likely to be 3. Derivations of all the results shown are available upon request.

Firm Recruitment and the Job Openings and Labor Turnover Survey

117

Fig. 3.5 Probability of observing a vacancy as a function of the arrival rate of candidates and of the relative cost of delay in the simple model of vacancy creation Note: Pu is the unrestricted probability while Pr is the probability restricted to include only vacancies where the position is available within one month.

sampled. One can use statistical methods developed in duration analysis to address this issue (see Lancaster 1990) and reconstruct the flow of vacancies from the stock data. The dependence of the probability of observation on the rate of arrival of suitable applicants has at least two important implications. First, diﬀerent probability of observing vacancies due to diﬀerent arrival rate of suitable applicants could be one explanation for why the vacancy-to-hires ratio varies substantially across industries, from a ratio of 0.30 in construction to a ratio of 1.48 in health. It is possible that the number of new vacancies opened per new hire is the same in these industries, and what is diﬀerent is how long the average vacancy in the industry is open due to the relative ease with which a vacancy in construction can be filled and the relative diﬃculty with which a vacancy in health can be filled. This interpretation of the data is supported by the strong positive correlation between the vacancy-to-hires ratio and the average education of workers across industries, shown in figure 3.6. Second, to the extent that there is systematic variation in over the business cycle, with recessions being times when openings are easier to fill and hence is higher, the probability of observ-

118

Éva Nagypál

ing a vacancy is procyclical in the above simple model, implying that the cyclical variation in the stock of vacancies overstates the variation in the flow of vacancies. The above simple model also helps us understand a second potential problem with the JOLTS measurement of vacancies. Recall that the JOLTS definition of a vacancy requires that work could start within one month of the day of measurement. This means that vacancies that are opened with long lead times (either because is low or because the relative cost of delay in hiring is high) are not counted in the JOLTS definition of job openings. In particular, one can show that the probability that a vacancy that is open at some point during the month [to – 1, to] is observed at time to given that only vacancies where the position is available within a month (i.e., where ts to 1) are counted is

1 e

if log (1 rd ) 1 1 rd Pr (, rd) 1

if log (1 rd ). 1 The two dashed lines in figure 3.5 plot this probability of observation as a function of for two diﬀerent values of the relative cost of delay, a low value of rd 1 and a higher value of rd 3. Under the JOLTS definition, this simple model implies that jobs with a higher relative cost of delay and where suitable workers arrive less frequently are less likely to be observed and counted compared to the case where all vacancies are counted irrespective of the time the position is available. The reason for this is simple: for jobs with a higher relative cost of delay and where suitable workers arrive less frequently, it is optimal to open vacancies with a long lead time and, as a consequence, often workers are hired for these jobs long before they start working for the employer. Vacancies for such jobs (for example, those for academics) are systematically under measured using the JOLTS definition. This under measurement could go some way toward explaining why the education industry in figure 3.6 lies much below the regression line. The statistical tools to address this measurement issue are less readily available than the tools to use in case of stock sampling. In my opinion, the best way to address this measurement issue would be to acquire additional data on vacancies where work is expected to start further into the future than one month. 3.3 Using JOLTS to Study the Matching Function The JOLTS data on vacancies allows for the empirical examination of the theoretical construct of a matching function using more direct measure of vacancies and new hires than previously available. In equilibrium models

Firm Recruitment and the Job Openings and Labor Turnover Survey

119

Fig. 3.6 Vacancy-to-hires ratio plotted against the average level of education by industry

of unemployment, the matching function is a theoretical construct that is used to describe how workers and firms meet in a frictional labor market. In particular, it posits that the flow of new matches between workers and firms is a function of the number of workers looking for employment and the number of vacancies that are opened by firms. Assuming that only unemployed workers look for employment (a commonly maintained assumption), the matching function posits that mt m(vt1, ut1) where mt is the number of new matches created during period t, and vt–1 and ut–1 are the number of vacancies and unemployed workers looking to form employment relationships at the end of period t – 1. The number of new matches created can be measured using the hires data in JOLTS (i.e., mt ht), so the JOLTS data provides two of the three data series necessary to estimate an aggregate matching function. Assuming a log-linear functional form for the matching function and an additive error term gives the empirical specification (3)

log ht c v log vt1 u log ut1 εt.

In terms of empirical implementation, there are several issues that need to be addressed. First, should one use seasonally adjusted or unadjusted data? Second, should the matching function be estimated using aggregate or industry-level data? Both of these questions turn out to be empirically relevant.

120

Éva Nagypál

Table 3.1

Matching function estimation results using seasonally adjusted and unadjusted data

Dependent variable log vt–1 log ut–1 Seasonally adjusted Month dummies Number of observations R2

log ht

log ht

0.668 (0.180) 0.378 (0.198) Yes No

0.531 (0.158) 0.185 (0.177) No Yes

47 0.579

47 0.958

To show this, I first use the seasonally adjusted JOLTS data and the seasonally adjusted number of unemployed from the CPS and estimate equation (3) by ordinary least squares (OLS). Table 3.1 reports estimation results for this empirical specification using data from December 2000 to November 2004. Under this specification, the hypothesis that the matching function has constant returns to scale cannot be rejected, and the elasticity of the matching function with respect to vacancies is estimated to be 0.67. This estimate of the elasticity is substantially larger than the matching function elasticity of 0.3 to 0.5 derived by Petrongolo and Pissarides (2001), though given the small sample size, the standard errors on the estimates are rather large. The second column of table 3.1 reports estimation results when seasonally unadjusted data are used and cm is allowed to vary with the month m. Now, the coeﬃcients both on the vacancy rate and the unemployment rate are lower, though the hypothesis that the matching function has constant returns to scale still cannot be rejected given the small sample size. Even with the small sample size, however, one can reject the hypothesis that the scale parameter cm is the same for all months m even at the 99 percent level of confidence. Figure 3.7 plots the estimate of e , which can be thought of as an estimate of matching eﬃciency, for each month m. There is a clear seasonal pattern in this matching eﬃciency, with the summer months representing a time when the same number of inputs into the matching function produce a significantly higher number of new hires. This strong seasonal pattern can also be clearly seen in the raw data for hires and vacancies plotted in figure 3.8, which shows that the number of hires is much more volatile over the year than the number of vacancies. There are two ways to interpret these findings. First, it is possible that there is seasonal variation in the process of matching. This could be due to the nature of employment relationships created over the seasonal cycle, with more temporary jobs filled by young workers being created over the summer, for example. Second, it is possible that there is consistent seasonal mismeasurement of vacm

Firm Recruitment and the Job Openings and Labor Turnover Survey

Fig. 3.7

121

Estimated matching eﬃciency for diﬀerent months of the year

Fig. 3.8 Aggregate vacancies and hires between December 2000 and December 2004 (not seasonally adjusted)

cancies over the seasonal cycle, due to respondents’ interpretation of job openings referring to openings for permanent employment relationships. Next, I estimate industry matching functions using the empirical specification (4)

log hit i m vi log vit1 ui log uit1 εit

122

Éva Nagypál

where i are industry and m are month scale parameters4. I measure uit–1 by the number of unemployed workers whose last employment was in industry i.5 Even with the limited amount of data available, estimating this specification allows one to decisively reject the hypothesis that the elasticity of the matching function is the same across industries (i.e., v1 v2 . . . v18 and u1 u2 . . . u18), and the hypothesis that the matching eﬃciency is the same across industries (i.e., 1 2 . . . 18), thereby rejecting the hypothesis that the matching function is stable across industries.6 Again, there are two ways to interpret these findings. First, it is possible that there is variation in the process of matching across industries due to the diﬀerent characteristics of jobs and workers in these industries. Second, it is possible that the measurement issues that I discussed above systematically aﬀect the measurement of vacancies and hires across industries. In any event, the lack of similarity in the matching function across industries raises the question whether there exists an aggregate matching function at all, as assumed in theoretical studies. 3.4 Conclusion The Job Openings and Labor Turnover Survey (JOLTS) contains important new information that is useful to test existing theories of vacancy creation and to provide new insights into the process of matching in the labor market. In this volume, chapter 2 by Faberman is an excellent introduction to the data available in the JOLTS for anyone wishing to do research using these data. In this chapter, I have focused on several measurement issues that researchers using the JOLTS will have to confront and suggested ways that one might use the JOLTS data to further our understanding of labor market dynamics.

References Diamond, P. 1981. Mobility costs, frictional unemployment, and eﬃciency. Journal of Political Economy 89 (4): 798–812. ———. 1982a. Aggregate demand management in search equilibrium. Journal of Political Economy 90 (5): 881–94.

4. Since the CPS started using the NAICS industry classification of the JOLTS only after January 2003, this equation is estimated using data from January 2003 to November 2004. Given the small number of observations, it is not possible to separately estimate the month scale parameter for each industry. 5. This implicitly abstracts from industry mobility, which is a strong assumption, but without it, it is not clear what the second input of an industry matching function should be. 6. Estimation results are available upon request.

Firm Recruitment and the Job Openings and Labor Turnover Survey

123

———. 1982b. Wage determination and eﬃciency in search equilibrium. Review of Economic Studies 49 (2): 217–27. Lancaster, T. 1990. The econometric analysis of transition data: Econometric society monographs. Cambridge: Cambridge University Press. Mortensen, D. T. 1982a. The matching process as a noncooperative bargaining game. In The Economics of Information and Uncertainty, ed. J. J. McCall, 233–54. Chicago: University of Chicago Press. ———. 1982b. Property rights and eﬃciency in mating, racing, and related games. American Economic Review 72:968–79. Nagypál, E. 2008. Worker reallocation over the business cycle: The importance of employer-to-employer transitions. Northwestern University Unpublished Manuscript. Petrongolo, B., and C. Pissarides. 2001. Looking into the black box: A survey of the matching function. Journal of Economic Literature 39:390–431. Pissarides, C. A. 1984. Search intensity, job advertising, and eﬃciency. Journal of Labor Economics 2 (1): 128–43. ———. 1985. Short-run equilibrium dynamics of unemployment, vacancies and real wages. American Economic Review 75:968–79. Wohlford, J., M. A. Phillips, R. Clayton, and G. Werking. 2003. Reconciling labor turnover and employment statistics. Proceedings of the Annual Statistical Association. CD-ROM.

4 Business Employment Dynamics Richard L. Clayton and James R. Spletzer

4.1 Introduction One of the most watched economic indicators in the United States is the monthly change in nonfarm payroll employment released by the Bureau of Labor Statistics (BLS). This statistic measures the net change in the number of jobs from one month to the next. But when we think about how employment grows or declines, we realize that some establishments have opened, some establishments have expanded, some establishments have contracted, and some establishments have closed. In this chapter, we describe the new gross job gains and gross job loss statistics from the BLS Business Employment Dynamics program. These statistics not only measure the large gross job flows that underlie the substantially smaller net employment changes, but also enhance our understanding of producer dynamics across various stages of the business cycle. The development of the BLS Business Employment Dynamics data was motivated in large part by research in the academic community. The creation of longitudinal establishment datasets at the U.S. Census Bureau during the past several decades led to influential publications by Dunne, Roberts, and Samuelson (1988, 1989a, 1989b), Davis and Haltiwanger (1990, 1992), and Davis, Haltiwanger, and Schuh (1996). From this literature, we have learned that there is a large amount of establishment-level employment volatility not evident at the aggregate level, and the gross job flow Richard L. Clayton is the Division Chief of the Division of Administrative Statistics and Labor Turnover at the Bureau of Labor Statistics. James R. Spletzer is the Senior Research Economist in the Employment Research and Program Development Staﬀ at the Bureau of Labor Statistics. We thank Ken Troske for his discussant comments at the April 2005 NBER-CRIW Producer Dynamics conference.

125

126

Richard L. Clayton and James R. Spletzer

statistics have fascinating business cycle properties. Yet despite all that we have learned about the labor market from this literature, the empirical analysis in these works was restricted to data from the manufacturing sector, and the call for more comprehensive and more timely data always resonates. The second generation of analysis using longitudinal microdata from the States’ Unemployment Insurance Systems illustrates how gross job flows in manufacturing are not representative of the entire U.S. economy (see Anderson and Meyer 1994; Foote 1998; Burgess, Lane, and Stevens 2000; and Spletzer 2000). The research resulting from the creation of these longitudinal establishment data sets has not only stimulated the review and updating of existing labor market theories, but has also stimulated the U.S. statistical agencies to develop their administrative data sets in such a way so as to produce longitudinal job flow statistics. This chapter begins with a definition of gross job gains and gross job losses, followed by a description of the source data used by the BLS to generate these statistics. Because the quality of longitudinal statistics computed from administrative cross-sectional microdata depends crucially on the longitudinal linkage algorithm, we pay particular attention in this chapter to describing our record linkage methodology. We then present highlights from the new BLS Business Employment Dynamics data series; these data show that in the first quarter of 2005, the number of gross job gains from opening and expanding establishments was 7.6 million, and the number of gross job losses from closing and contracting establishments was 7.3 million. The new BLS Business Employment Dynamics data also show that the 2001 recession was characterized by a temporary spike in gross job losses accompanied by a substantial and persistent decline in gross job gains. In this chapter we introduce a new seasonally adjusted time series of the distribution of quarterly gross job flows. This new time series is motivated by several interesting questions about gross job flows over the business cycle. For example, did the temporary spike in gross job losses during the 2001 recession occur at a few establishments with large declines, or at many establishments with small declines? And did the substantial fall in gross job gains during the 2001 recession occur at a few establishments cutting back significantly on hiring, or many establishments not adding a few new positions? Our new time series shows that the relatively few establishments with large gross job gains and large gross job losses were the drivers of the 2001 recession. 4.2 The Business Employment Dynamics Program at BLS 4.2.1 Concepts and Definitions The employment statistics that are published by the Bureau of Labor Statistics are invaluable for policymakers, researchers, and the business community. The BLS report on the monthly net change in employment

Business Employment Dynamics

127

aﬀects stock market movements and interest rate decisions considerably. Yet this single macroeconomic statistic is the net result of the millions of decisions by millions of business establishments in the U.S. economy changing their employment levels. Each decision reflects the businessspecific economic conditions that face managers every day: supply, demand, labor availability, market share goals, investments in research and development, and so on. While the aggregate net employment change statistic identifies the overall growth or decline of the labor market, it does not summarize the underlying heterogeneity of the many establishments opening and expanding, or the many establishments contracting or closing. The definitions of gross job gains and gross job losses are easily derived from the definition of net employment growth. Notationally, let Ee,t denote the employment of establishment e in quarter t. Net employment growth in quarter t is defined as the change in aggregate employment from one quarter to the next: (1)

∑

Net Employment Growth (t)

∑

Ee,t

all establishments

Ee,t1.

all establishments

Noting that establishments can be classified based upon their employment dynamics from one quarter to the next, this equation for net employment growth can be manipulated as:

∑

(2) Net Employment Growth (t)

Ee,t

all establishments

∑

Ee,t1

all establishments

∑

(Ee,t Ee,t1)

∑

(Ee,t Ee,t1)

all establishments

establishments increasing employment

∑

(Ee,t Ee,t1)

∑

(Ee,t Ee,t1)

establishments decreasing employment

establishments with no change in employment

∑

(Ee,t 0)

opening establishments

∑

(Ee,t Ee,t1)

contracting establishments

∑

∑

(Ee,t Ee,t1)

expanding establishments

(0 Ee,t1).

closing establishments

128

Richard L. Clayton and James R. Spletzer

Note that the quarterly employment change for the set of establishments that do not change their level of employment from one quarter to the next is zero, and this term drops out of the final version of equation (2). In the Business Employment Dynamics data, there are 3.2 million establishments with positive employment that do not change their employment between the fourth quarter of 2004 and the first quarter of 2005. The definitions for gross job gains and gross job losses fall immediately out of the previous equation. Gross job gains are the sum of all employment increases at opening and expanding establishments: (3)

∑

Gross Job Gains (t)

(Ee,t 0)

opening establishments

∑

(Ee,t Ee,t1).

expanding establishments

Gross job losses are the sum of all employment losses at contracting and closing establishments: (4)

Gross Job Losses (t)

∑

(Ee,t Ee,t1)

contracting establishments

∑

(0 Ee,t1).

closing establishments

An expanding establishment is defined as a continuous unit that increases its employment from a positive level in the previous quarter to a higher level in the current quarter, and a contracting establishment is a continuous unit that decreases its employment from the previous quarter to a lower positive level in the current quarter. An opening establishment is one that has positive employment in the current quarter, and either had zero employment or was not in the database the previous quarter. A closing establishment is one that had positive employment in the previous quarter, and has either zero employment or is not in the database the current quarter. Because it is not possible to define business deaths on a contemporaneous basis, the definitions of establishment openings and closings used in the BLS Business Employment Dynamics program are conceptually diﬀerent than the more familiar definitions of establishment births and deaths. In the State Unemployment Insurance (UI) systems, businesses are allowed to and often do report zero employment for several quarters after they have eﬀectively closed. This undoubtedly occurs when a business owner temporarily shuts down but anticipates starting up the business again when economic conditions improve. By reporting zero employment and wages on the quarterly contributions form, the business owner can keep their UI account active. This results in many observed business closings, but which of these closings will start up again and which will die will not be observed for several more quarters. It is important to note that gross job gains and gross job loss statistics measure the sum of establishment-level net employment changes, and do not measure the flow of workers into and out of the establishment. For ex-

Business Employment Dynamics

129

ample, if an establishment increases employment from fifty workers to sixty workers, these ten additional jobs are classified as gross job gains. This addition of ten jobs during the quarter might have occurred with the addition of ten new hires, or by the net of twenty new hires and ten separations. Counts of hires and separations are published monthly by the Job Openings and Labor Turnover Survey (JOLTS) program at the BLS. Both Clark (2004) and Faberman (chapter 2, this volume) present a thorough description of the conceptual foundations and the empirical estimates from the JOLTS program. 4.2.2 Source Data The quarterly BLS Business Employment Dynamics data series is constructed from microdata originating from the Quarterly Census of Employment and Wages (QCEW), also known as the ES-202 program. A complete description of the underlying source data and the data flows can be found in the longer conference version of this chapter (Clayton and Spletzer 2005) and in the April 2004 Monthly Labor Review (Spletzer et al. 2004); the following is a bare-bones description of the source data. All employers subject to state Unemployment Insurance (UI) laws are required to submit quarterly contribution reports detailing their monthly employment and quarterly wages to the State Employment Security Agencies. The raw UI data require substantial edit and review. In addition, the BLS directs the states to conduct two supplemental surveys that are necessary to yield accurate data at the local level. The first is the Annual Refiling Survey (ARS), where nearly two million businesses each year are contacted to obtain or update business name, addresses, industry codes, and related contact information. The second is the Multiple Worksite Report (MWR), which collects employment and wages for each establishment in multiunit firms within the state. The MWR covers about 110,000 businesses (1.4 percent of all businesses, 16 percent of all establishments, and 39 percent of employment) each quarter, allowing the accurate distribution of employment and wages to the correct county and industry. Without these two additions to the UI data, the resulting QCEW economic information would not be accurate at the industry level or at the MSA, county, or city level. In addition, state QCEW staﬀs review and reconcile complex cases including mergers and acquisitions where correctly determining and linking predecessors and successors is critical to the accuracy of the QCEW and the Business Employment Dynamics data. After the microdata are augmented and thoroughly edited by the State Labor Market Information staﬀ, the states submit these data and other business identification information to the Bureau of Labor Statistics as part of the federal-state cooperative QCEW program. The data gathered in the QCEW program are a comprehensive and accurate source of employment and wages, and provide a virtual census (98 percent) of employees on

130

Richard L. Clayton and James R. Spletzer

nonfarm payrolls. In the first quarter of 2005, the QCEW statistics show an employment level of 129.8 million, with 8.5 million establishments in the U.S. economy. The BLS publishes the Business Employment Dynamics data approximately seven-and-a-half months after the end of the quarter. 4.2.3 Longitudinal Linkages The quarterly gross job gains and gross job loss statistics created in the BLS Business Employment Dynamics program are tabulated by linking establishments across quarters; establishments are then classified as opening, expanding, contracting, closing, or not changing their employment level. The accuracy of the Business Employment Dynamics statistics depends on two primary factors: the quality of the establishment-level microdata being reported by businesses to the states, and the record linkage methodology used by the BLS to link establishments across quarters. Following establishments across time using administrative UI microdata is a complex and challenging exercise. Creating the Business Employment Dynamics data series requires a thorough understanding of how businesses operate and how they file their UI tax forms. The manner in which businesses report administrative changes and ownership changes can result in establishments changing UI identifiers even though no economic changes occurred. Failing to identify and link such noneconomic changes would result in an overstatement of establishment openings and closings, and thus an overstatement of gross job gains and gross job losses. The BLS has developed a multistep process to accurately link business establishment microdata over time. This linkage process consists of four steps: two distinct administrative matches, a probability-based weighted match, and an analyst intervention match. The linkage process is based on the unique establishment identifier maintained by the states. This identifier is composed of two pieces: the UI number and the reporting unit number. The UI number refers to the taxpaying entity within the state. The reporting unit number refers to establishments within the firm. Although the reporting unit number is not used in the administration of the UI system, it is assigned by the state using information collected from the Multiple Worksite Reports. The first step in the Business Employment Dynamics record linkage methodology is to link establishments that maintain the same establishment identifier across quarters. This step identifies almost all of the establishments linked as continuous across quarters. This is followed by a match using predecessor and successor information. Predecessors and successors refer to establishments that are continuous across quarters, yet the establishment identifier changes as a result of a change in ownership or a change in the reporting configuration of a multi-establishment company. The vast majority of predecessor and successor linkages are businesses buying another business (the assumption of liability for UI taxes must be reported to

Business Employment Dynamics

131

the state); other predecessor and successor linkages are identified by the State Labor Market Information Staﬀ. The third step in the linkage process, conducted by the BLS, is a probability-based weighted match process. This probability-based weighted match uses information such as establishment name, street address, and telephone number to link—as continuous—a closing establishment in the previous quarter with an opening establishment in the current quarter. The theoretical foundation for the BLS record linkage methodology is based on the work of Ivan P. Felligi and Alan B. Sunter, and is more fully explained in Robertson et al. (1997). The final step in the matching process is an analyst review and possible manual linkage of selected large unmatched records. Although this analyst review and manual linkage is very resource intensive, it is crucial for the quality of the detailed industry and geography statistics. The BLS has undertaken many detailed reviews and analyses of the quality of its longitudinal linkage algorithm, and continues to conduct research to explore the sources and consequences of any additional valid establishment links. Furthermore, as part of the annual cooperative agreement between BLS and the states, the BLS is now requiring that the states examine and attempt to explain any unlinked records with employment above a certain threshold; this review of opening and closing records by state analysts before it is transmitted to the BLS will certainly increase the quality of the Business Employment Dynamics data. 4.3 The Business Employment Dynamics Data The basic products from the new BLS Business Employment Dynamics program are statistics measuring quarterly gross job gains and gross job losses. The gross job gains can be decomposed into the gains from both expansions and openings, and the gross job losses can be decomposed into the losses from both contractions and closings. The Business Employment Dynamics program also publishes the establishment counts underlying the employment gains and losses. All these statistics are available from the BLS website (http://www.bls.gov/bdm) as both levels and percents, and seasonally adjusted or unadjusted. The time series of historical statistics starts in the third quarter of 1992. The following summary of the data is a shortened version of what can be found in the longer conference version of this chapter (Clayton and Spletzer 2005) and in the April 2004 Monthly Labor Review (Spletzer et al. 2004). 4.3.1 Point-in-Time Results The seasonally adjusted gross job gains and gross job loss statistics for the first quarter of 2005 are presented in table 4.1 (data for the first quarter of 2005 were the most recent available data when we submitted this article for publication in January 2006). We see that the economy gained 325,000

132 Table 4.1

Richard L. Clayton and James R. Spletzer Gross job gains and job losses, March 2005a Net Change, Employment Gross Job Gains Total Expanding Establishments Opening Establishments Gross Job Losses Total Contracting Establishments Closing Establishments

a

325

7,635 6,171 1,464 7,310 5,852 1,458

Seasonally adjusted quarterly data, in thousands.

net new jobs (seasonally adjusted) between December 2004 and March 2005. This growth in employment is the net result of two components: the gross job gains of 7.635 million jobs and the gross job losses of 7.310 million jobs. The gross job gains and gross job loss statistics are substantially larger than the net employment change. Gross job gains come from both expanding and opening establishments. In table 4.1, we see that employment in expanding establishments grew by 6.171 million jobs and employment in opening establishments grew by 1.464 million jobs. These statistics indicate that expanding establishments account for 81 percent of quarterly gross job gains, whereas opening establishments account for 19 percent of quarterly gross job gains. With regard to gross job losses, employment in contracting establishments declined by 5.852 million jobs, and closing establishments accounted for the loss of 1.458 million jobs. Contracting establishments account for 80 percent of quarterly gross job losses, whereas closing establishments account for 20 percent of quarterly gross job losses. Expanding and contracting establishments account for most jobs gained and most jobs lost when measured on a quarterly frequency. An important component of the Business Employment Dynamics data series is the establishment counts underlying the gross job gains and gross job losses. These establishment counts for the first quarter of 2005, on a seasonally adjusted basis, are reported in table 4.2. There were 1.506 million expanding establishments and 1.504 million contracting establishments during the first quarter of 2005. There were 345,000 establishments opening during the quarter, and 347,000 establishments closing during the quarter. The diﬀerence between the number of opening and closing establishments (–2,000) is the net change in the number of active establishments during the quarter. By revealing the tremendous amount of churning underlying the net growth rates, the Business Employment Dynamics data enhance the labor market statistics currently available from the Bureau of Labor Statistics.

Business Employment Dynamics Table 4.2

Number of establishments, by direction of employment change, March 2005a Net Change, Establishments Establishments Gaining Jobs Total Expanding Establishments Opening Establishments Establishments Losing Jobs Total Contracting Establishments Closing Establishments

a

133

–2

1,851 1,506 345 1,851 1,504 347

Seasonally adjusted quarterly data, in thousands.

The traditional measure of net employment change produced by the BLS indicates that employment grew by 325,000 jobs during the first quarter of 2005 (seasonally adjusted). The gross job gains and gross job loss statistics indicate that this net employment loss is the result of 6.171 million jobs added at 1.506 million expanding establishments, 1.464 million jobs added at 345,000 opening establishments, 5.852 million jobs lost at 1.504 million contracting establishments, and 1.458 million jobs lost at 347,000 closing establishments. These gross job flows that underlie the net employment growth statistic demonstrate that there are a sizable number of jobs and establishments that appear and disappear in the short time frame of three months. These statistics are calculated without additional data collection eﬀorts or additional respondent burden. 4.3.2 Time-Series Results—Business Cycle Analysis The business cycle, to a large degree, is defined by the growth of employment (or lack thereof). The new BLS Business Employment Dynamics data will enable researchers to analyze the extent to which economic recessions and expansions are characterized by changes in business expansions and openings, by changes in business contractions and closings, or by a combination of the two. The seasonally adjusted time series of quarterly net employment growth is shown in figure 4.1. The recent recession, which was dated by the National Bureau of Economic Research (NBER) as occurring between March 2001 to November 2001, is clearly evident in this chart. Prior to the recession, between the third quarter of 1992 and the fourth quarter of 2000, net employment growth had been positive every quarter, averaging 637,000 net new jobs per quarter. But during the recession, as seen in figure 4.1, net employment growth was negative for all quarters of 2001, with a low of 1.380 million net jobs lost in the third quarter of 2001. The seasonally adjusted gross job gains and gross job loss statistics are

134

Fig. 4.1

Richard L. Clayton and James R. Spletzer

Quarterly net employment growth (seasonally adjusted, in thousands)

plotted in figure 4.2. The diﬀerence between the gross job gains and the gross job losses in figure 4.2 is the familiar net employment change depicted in figure 4.1. The most recent business cycle is evident in figure 4.2. Between 1992 and 1999, both the gross job gains and the gross job loss series were climbing at relatively constant rates. The gross job gains started to decline in early 2000, and then dropped substantially in 2001. After a peak of 9.144 million gross job gains in the fourth quarter of 1999, the gross job gains fell to 7.749 million jobs in the third quarter of 2001. The gross job losses continued to increase through 2001, rising from 8.354 million gross jobs lost in the fourth quarter of 2000 to a high of 9.129 million gross jobs lost in the third quarter of 2001. Thus, the declining net employment growth during the first three quarters of 2001 can be attributed to both falling gross job gains and rising gross job losses. As the oﬃcial NBER-dated recession ended in late 2001, the gross job losses significantly declined and by early 2002 had returned to a level comparable to its prerecessionary level in early 2000. The same cannot be said for the gross job gains. Following the recession, the gross job gains statistic has remained in the range of 7.4 to 8.1 million jobs gained each quarter, which is substantially lower than its prerecessionary levels (the gross job gains in calendar year 2000 averaged 8.8 million jobs per quarter). The gross job gains started to increase in late 2003. There has been positive net employment growth since the third quarter of 2003, as this recent increase in gross job gains has been accompanied by a gross job loss series that steadily declined through 2003 and remained relatively constant through 2004.

Business Employment Dynamics

Fig. 4.2

135

Quarterly gross job gains and losses (seasonally adjusted, in thousands)

The seasonally adjusted time series of gross job gains at expanding and opening establishments—and the gross job losses at contracting and closing establishments—are presented in figure 4.3. Immediately obvious is the prior-stated observation that, for any given quarter, expanding and contracting establishments account for roughly 80 percent of gross jobs gained and gross jobs lost, respectively, when measured on a quarterly frequency. Also obvious in figure 4.3 is that the business cycle is most evident in the expansionary and contractionary establishments. The diﬀerence between the gross job gains due to expansions and the gross job losses due to contractions mirrors the overall diﬀerence between the gross job gains and the gross job losses. The diﬀerence between the gross job gains due to openings and the gross job losses due to closings does exhibit some business cycle properties, but this diﬀerence is quite small relative to the diﬀerence between expansions and contractions. 4.3.3 Additional Research Results In addition to the basic results just described, the BLS has also released several other data products from the Business Employment Dynamics program. Statistics for major industry sectors were released in May 2004, statistics by firm size class were released in December 2005 (Butani et al. [2006], discuss and empirically analyze the interesting methodological issues underlying longitudinal size class statistics), and statistics by state were released in August 2007. There have also been several recent research papers using the longitudinal establishment microdata from the Business Employment Dynamics program—Pinkston and Spletzer (2004) present

136

Fig. 4.3

Richard L. Clayton and James R. Spletzer

Quarterly gross job gains and losses (seasonally adjusted, in thousands)

annual tabulations of gross job gains and gross job losses, Knaup (2005) and Knaup and Piazza (2007) present survival statistics of business births, Sadeghi (2008) computes establishment birth and death statistics, Butani, Werking, and Kapani (2005) analyze how net employment growth diﬀers in single-establishment employers versus multi-establishment firms, Clayton and Mousa (2004) describe linking the Business Employment Dynamics data with state wage records, Hyson and Spletzer (2002) analyze the employment and wage dynamics associated with mass layoﬀs, Brown and Spletzer (2005) analyze the employment and wage dynamics of businesses involved in oﬀshoring, and Faberman (2004) creates quarterly gross job gains and gross job loss statistics for the 1990 to 1991 recession. 4.3.4 Comparison to Other Data We have been asked many times how the Business Employment Dynamics data compares to gross job flow statistics from other datasets. This is a diﬃcult question to answer precisely due to diﬀerences in time periods, diﬀerences in industry sectors, diﬀerences in reporting frequency, and differences in definitions. We are aware of two research papers that have attempted to compare the Business Employment Dynamics data to the manufacturing statistics in the heavily cited work of Davis, Haltiwanger, and Schuh (1996). Pinkston and Spletzer (2004) compute annual gross job gains and losses statistics for the manufacturing sector, and conclude that the Business Employment Dynamics statistics are broadly similar to those of Davis, Haltiwanger, and Schuh. Faberman (2004) plots the quarterly

Business Employment Dynamics

137

Business Employment Dynamics manufacturing statistics on the same chart as the 1972 to 1993 quarterly statistics from Davis, Haltiwanger, and Schuh, and concludes that the data are relatively comparable. There is also interest in how the Business Employment Dynamics data compare to the data from the Job Openings and Labor Turnover Survey (JOLTS). The JOLTS data are from a sample of approximately 16,000 U.S. business establishments collected by the BLS. The JOLTS program publishes monthly data on hires, separations (quits, layoﬀs and discharges, and other separations), and job openings. These data are meant to serve as demand-side indicators of labor shortages at the national level. Further information about the JOLTS and some research using the JOLTS can be found in Clark and Hyson (2001), Clark (2004), Faberman (chapter 2, this volume), and Nagypál (chapter 3, this volume). Several previous authors have compared the JOLTS hires and separations data to the gross job gains and gross job losses data from the Business Employment Dynamics. Davis, Faberman, and Haltiwanger (2006) characterize the relationship of hires, separations, quits, and layoﬀs to the employer-level gross job gains and gross job loss statistics. In table 4.1 of their article, Davis, Faberman, and Haltiwanger report average job and worker flow rates for the U.S. economy measured at various frequencies using the JOLTS and the Business Employment Dynamics data. Boon et al. (2008) compare the concepts and the data from the JOLTS, the Business Employment Dynamics, and the CPS gross flows. In charts 7 and 8 of their article, Boon et al. compare the time series movements of the JOLTS and the Business Employment Dynamics data. 4.4 The Distribution of Gross Job Gains and Gross Job Losses 4.4.1 Concepts and Definitions The Business Employment Dynamics data have given us several interesting facts about producer dynamics during and immediately following the 2001 recession. As seen in figure 4.2 of this chapter, the recent business cycle is characterized by a large temporary spike in gross job losses accompanied by a substantial and persistent decline in gross job gains. In this section of the chapter, we present seasonally adjusted time series of the distribution of gross job gains and gross job losses underlying the BLS Business Employment Dynamics statistics. Distribution statistics will allow us to analyze (a) whether the temporary spike in gross job losses occurred at a few establishments with large declines, or at many establishments with small declines, and (b) whether the decline in gross job gains occurred at a few establishments cutting back significantly on hiring or at many establishments not adding a few new positions. Recall from equation (2) earlier in this chapter that the net employment

138

Richard L. Clayton and James R. Spletzer

growth in any given quarter can be written as the sum of gross job gains from establishments increasing employment and the sum of gross job losses from establishments decreasing employment: (2) Net Employment Growth (t)

∑

(Ee,t Ee,t1)

establishments increasing employment

∑

(Ee,t Ee,t1).

establishments decreasing employment

This equation can be rewritten as:

(5)

Net Employment Growth (t) ∑

∑

(Ee,t Ee,t1)

x1 establishments increasing employment by x jobs

∑

∑

(Ee,t Ee,t1).

x1 establishments decreasing employment by x jobs

In equation (5) we have decomposed both gross job gains and gross job losses into an empirical distribution defined by the number of jobs gained or lost. For practical purposes, it is infeasible to calculate and report statistics for every possible level of net employment change x in equation (5). We have calculated gross job gains and gross job losses for establishments gaining or losing {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11–14, 15–19, 20–24, 25–29, 30– 39, 40–49, 50–74, 75–99, 100} jobs. However, for the graphical analysis we wish to present, 19 series is too many, and we have aggregated further. We have chosen to present statistics for the following intervals of gross job gains and gross job losses: {1–3, 4–19, 20}. In the fourth quarter of 2004, 16 percent of employment is in establishments that do not change their employment level, 33 percent of employment is in establishments that change their employment level by 1 to 3 jobs, 30 percent of employment is in establishments that change their employment level by 4 to 19 jobs, and 21 percent of employment is in establishments that change their employment level by 20 or more jobs. We have looked extensively at other possible aggregations and have determined that the main conclusions we present in this section are not sensitive to the particular aggregation we have chosen. To be precise, we have decomposed gross job gains in quarter t as: (6)

∑

establishments increasing employment

∑

(Ee,t Ee,t1)

(Ee,t Ee,t1)

establishments increasing employment by 1–3 jobs

∑

establishments increasing employment by 20+ jobs

∑

establishments increasing employment by 4–19 jobs

(Ee,t Ee,t1).

(Ee,t Ee,t1)

Business Employment Dynamics

139

Similarly, we have decomposed gross job losses in quarter t as: (7)

∑

(Ee,t Ee,t1)

establishments decreasing employment

∑

(Ee,t Ee,t1)

establishments decreasing employment by 1–3 jobs

∑

∑

(Ee,t Ee,t1)

establishments decreasing employment by 4–19 jobs

(Ee,t Ee,t1).

establishments decreasing employment by 20+ jobs

The issue of whether to present our distribution statistics in levels or in rates deserves mention. Much of the existing literature has used rates; for example, figure 2.2 of Davis, Haltiwanger, and Schuh (1996) reports the distribution of job creation rates and job destruction rates for intervals spanning 5 percentage points. We have chosen to use levels because we are concerned about the interpretation of rates for small establishments. Based upon analysis of the QCEW microdata, most establishments in the United States are small: 61 percent of establishments have less than five employees, and 88 percent of establishments have less than twenty employees. The comparable statistics for the employment distribution are as follows: 7 percent of employment is in establishments with less than five employees, and 26 percent of employment is in establishments with less than twenty employees. When calculating percentages using average employment in the denominator, as is standard, a small establishment with less than five employees that grows or declines by one job has a percentage change (in absolute value) of between 22 and 200 percent, whereas a large establishment with more than 500 employees that grows or declines by one job has a percentage change (in absolute value) of less than 0.2 percent. Because we are interested in decomposing the time series variation of net employment growth based upon the distribution of establishment-level changes, the use of levels as expressed in equation (5) strikes us as most appropriate for our first pass through the microdata. Research that calculates rates rather than levels, and that conditions on the size of the establishment to make rates comparable across establishments, is in progress. 4.4.2 Empirical Results In the top panel of figure 4.4, we present the establishment counts for establishments gaining or losing 1 to 3 jobs, 4 to 19 jobs, and 20 or more jobs. The bottom panel of figure 4.4 reports the net number of establishments gaining 1 to 3 jobs, 4 to 19 jobs, and 20 or more jobs, where the net is calculated as the number of establishments gaining minus the number of establishments losing a given amount of jobs. In the fourth quarter of 2004, there were 1.456 million establishments (seasonally adjusted) that gained 1 to 3 jobs, and 1.400 million establishments that lost 1 to 3 jobs. This indicates that 56 thousand more establishments were gaining 1 to 3 jobs than

140

Richard L. Clayton and James R. Spletzer

Fig. 4.4 Quarterly gross job gains and losses, establishment counts (seasonally adjusted)

were losing 1 to 3 jobs; this 56 thousand figure is plotted in the bottom panel of figure 4.4. There were 381 thousand establishments gaining 4 to 19 jobs, and 344 thousand establishments losing 4 to 19 jobs. There were 56 thousand establishments gaining 20 or more jobs, and 50 thousand establishments losing 20 or more jobs. The establishment counts in figure 4.4 clearly show business cycle properties. Looking at the bottom panel of figure 4.4, the net number of establishments gaining 1 to 3 jobs falls from 87 thousand in the fourth quarter of

Business Employment Dynamics

141

1999 to negative 69 thousand in the third quarter of 2001. The net number of establishments gaining 20 or more jobs also falls from 8 thousand in the fourth quarter of 1999 to negative 11 thousand in the third quarter of 2001. The statistics in figure 4.5 show the employment gains and losses associated with the establishments gaining or losing 1 to 3 jobs, 4 to 19 jobs, and 20 or more jobs. The ordering of the series in figure 4.5 is opposite than in figure 4.4. In the top panel of figure 4.5, we see that the 1.5 million estab-

Fig. 4.5

Quarterly gross job gains and losses (seasonally adjusted)

142

Richard L. Clayton and James R. Spletzer

lishments gaining 1 to 3 jobs contributed 2.2 million jobs to the gross job gains in the fourth quarter of 2004. The 381 thousand establishments growing by 4 to 19 jobs contributed 2.8 million jobs to the count of gross job gains in the fourth quarter of 2004 (the average growth of these job-gaining establishments is 7.3 jobs), and the 56 thousand establishments growing by 20 or more jobs added 3.1 million new jobs to the economy (an average growth of 56 jobs per establishment). The key graph is in the bottom panel of figure 4.5. Between the third quarter of 1992 and the fourth quarter of 1999, establishments gaining or losing 1 to 3 jobs created an average of 99 thousand net new jobs per quarter. During this same time period, establishments gaining or losing 4 to 19 jobs created an average of 228 thousand net new jobs per quarter, and establishments gaining or losing 20 or more jobs created an average of 331 thousand jobs per quarter. These three statistics sum to the average net employment growth of 657 thousand per quarter during the 1990s (the three series in the bottom panel of figure 4.5 sum to the series graphed in figure 4.1). The 2001 recession is clearly evident in both the top and bottom panels of figure 4.5. Establishments that were gaining or losing 1 to 3 jobs lost a net 110 thousand jobs during the third quarter of 2001, establishments that were gaining or losing 4 to 19 jobs lost a net of 325 thousand jobs in that quarter, and establishments that were gaining or losing 20 or more jobs lost a net of 758 thousand jobs in the third quarter of 2001. These statistics indicate that 64 percent of the net job losses in the most severe recessionary quarter are attributable to the relatively few establishments gaining or losing 20 or more jobs. To return to the motivating question, this new seasonally adjusted time series of quarterly distribution statistics illustrates where the temporary spike in gross job losses occurred in the 2001 recession. The spike in gross job losses did not occur because many establishments had small declines in employment, but rather from a relatively few number of establishments with large declines. Similarly, the substantial and persistent fall in gross job gains during and following the 2001 recession did not occur because many establishments did not add a few positions, but rather this fall can be attributed to a relatively few number of establishments cutting back significantly on their hiring. The analysis we have presented in this section is quite simple. There are many empirical extensions that could be done. As mentioned above in the discussion of levels versus rates, it would be interesting to know whether the establishments that are adding or losing twenty or more jobs are relatively small establishments with a large percentage change in employment, or whether they are large establishments with a relatively small percentage change in employment. Furthermore, the statistics we have presented are quarterly; annual distribution statistics would enable us to analyze whether the large (twenty or more) establishment-level gains or losses in a quarter are onetime changes within a year, or whether they are one incremental

Business Employment Dynamics

143

step towards even larger gains or losses within the year. We hope that our presentation and simple analysis of distribution statistics that we have provided in this section will spur on additional empirical and theoretical work about producer dynamics and the causes and consequences of employment growth over the business cycle. 4.4.3 Sectoral Detail The editors of this conference volume have asked us present some sectoral detail. The statistics in figure 4.6 show the distribution of employ-

Fig. 4.6

Quarterly gross job gains and losses, manufacturing (seasonally adjusted)

144

Fig. 4.7

Richard L. Clayton and James R. Spletzer

Quarterly gross job gains and losses, services (seasonally adjusted)

ment gains and losses for the manufacturing sector, and the statistics in figure 4.7 show the distribution of employment gains and losses for the services sector. The basic results for these two sectors mimic the analysis we presented for the national statistics. During the 1990s, establishments with large gains or losses in employment are the biggest contributors to the gross job gains and gross job losses. During the 2001 recession, the em-

Business Employment Dynamics

145

ployment losses are most evident for the establishments with the largest gains and losses. 4.5 Conclusion Our goals in this chapter were threefold: to describe the BLS Business Employment Dynamics program, to summarize the data from this program and how it has informed us about the U.S. labor market, and to present a new seasonally adjusted time series of the distribution of quarterly gross job gains and gross job losses. The first two objectives are described in the text, and are not summarized here. This chapter released for the first time a seasonally adjusted time series of the distribution of quarterly gross job gains and gross job losses for the entire U.S. economy. This new data series is motivated by the earlier work of Davis and Haltiwanger (1990, 1992), Davis, Haltiwanger, and Schuh (1996), and Spletzer (2000). We have learned from these earlier studies that gross job gains and gross job losses are concentrated at establishments with large percentage changes in employment. We mimic this finding with the Business Employment Dynamics data—in the fourth quarter of 2004, we find that 39 percent of all gross job gains are contributed by just 3 percent of establishments who gain twenty or more jobs, and 38 percent of all gross job losses are contributed by just 3 percent of establishments who lose twenty or more jobs. Our seasonally adjusted time series shows that these relatively few establishments with large gross job gains and large gross job losses are the drivers of the 2001 business cycle. The Business Employment Dynamics data is now routinely cited in the economic, statistical, and policy communities, as well as in the popular press. This high level of attention by the user community reinforces our belief that the relatively new BLS Business Employment Dynamics data is a major contributor to our understanding of producer dynamics in the U.S. economy. We do not find this surprising: the data are timely, high quality, high frequency, and historically consistent. And in conclusion, we note that the BLS was able to create the Business Employment Dynamics data with no new data collection eﬀorts and with no new additional respondent burden.

References Anderson, P. M., and B. D. Meyer. 1994. The extent and consequences of job turnover. Brookings Papers on Economic Activity, Vol. 1994:177–236. Boon, Z., C. M. Carson, R. J. Faberman, and R. E. Ilg. 2008. Studying the Labor Market with BLS Labor Dynamics Data. Monthly Labor Review 131 (2): 3–16.

146

Richard L. Clayton and James R. Spletzer

Brown, S., and J. Spletzer. “Labor Market Dynamics Associated with the Movement of Work Overseas.” Paper presented at the Organization for Economic Cooperation and Development (OECD) conference on the Globalisation of Production. November 2005. Burgess, S., J. Lane, and D. Stevens. 2000. Job flows, worker flows, and churning. Journal of Labor Economics 18 (3): 473–502. Butani, S. J., R. L. Clayton, V. Kapani, J. R. Spletzer, D. M. Talan, and G. S. Werking Jr. 2006. Business employment dynamics: Tabulations by employer size. Monthly Labor Review 129 (2): 3–22. Butani, S., G. Werking, and V. Kapani. 2005. Employment dynamics of individual companies versus multicorporations. Monthly Labor Review 128 (12): 3–15. Clark, K. A. 2004. The job openings and labor turnover survey: What initial data show. Monthly Labor Review 127 (11): 14–23. Clark, K. A., and R. Hyson. 2001. New tools for labor market analysis: JOLTS. Monthly Labor Review 124 (12): 32–37. Clayton, R. L., and J. A. Mousa. 2004. Measuring labor dynamics: The next generation in labor market information. Monthly Labor Review 127 (5): 3–8. Clayton, R. L., and J. R. Spletzer. Business employment dynamics. Paper presented at the April 2005 NBER-CRIW conference on Producer Dynamics. Online at http://www.nber.org/confer/2005/CRIWs05/clayton.pdf Davis, S. J., R. J. Faberman, and J. Haltiwanger. 2006. The flow approach to labor markets: New data sources and micro-macro links. Journal of Economic Perspectives 20 (3): 3–26. Davis, S. J., and J. C. Haltiwanger. 1990. Gross job creation and destruction: Microeconomic evidence and macroeconomic implications. NBER Macroeconomics Annual 5:123–68. ———. 1992. Gross job creation, gross job destruction, and employment reallocation. Quarterly Journal of Economics 57 (3): 819–63. Davis, S. J., J. C. Haltiwanger, and S. Schuh. 1996. Job creation and destruction. Cambridge, MA: The MIT Press. Dunne, T., M. J. Roberts, and L. Samuelson. 1988. Patterns of firm entry and exit in U.S. manufacturing industries. RAND Journal of Economics 19 (4): 495–515. ———. 1989a. Plant turnover and gross employment flows in the U.S. manufacturing sector. Journal of Labor Economics 7 (1): 48–71. ———. 1989b. The growth and failure of U.S. manufacturing plants. Quarterly Journal of Economics 54 (4): 671–98. Faberman, R. J. 2004. Gross job flows over the past two business cycles: Not all “recoveries” are created equal. Bureau of Labor Statistics Working Paper no. 372. Available at http://www.bls.gov/ore/pdf/ec040020.pdf Foote, C. L. 1998. Trend employment growth and the bunching of job creation and destruction. Quarterly Journal of Economics 63 (3): 809–34. Hyson, R. T., and J. R. Spletzer. 2002. Large-scale layoﬀs, employment dynamics, and firm survival. Paper presented at the Society of Labor Economists annual conference. May 2002. Knaup, A. E. 2005. Survival and longevity in the Business Employment Dynamics data. Monthly Labor Review 128 (5): 50–56. Knaup, A. E., and M. C. Piazza. 2007. Business employment dynamics data: Survival and longevity, II. Monthly Labor Review 130 (9): 3–10. Pinkston, J. C., and J. R. Spletzer. 2004. Annual measures of gross job gains and gross job losses. Monthly Labor Review 127 (11): 3–13. Robertson, K., L. Huﬀ, G. Mikkelson, T. Pivetz, and A. Winkler. 1999. Improvements in record linkage processes for the Bureau of Labor Statistics’ Business Establishment list. In Record Linkage Techniques—1997: Proceedings of an In-

Business Employment Dynamics

147

ternational Workshop and Exposition, 212–221. Washington, D.C.: National Academy Press. Sadeghi, A. 2008. Measuring births and deaths in business employment dynamics data series. Monthly Labor Review, forthcoming. Spletzer, J. R. 2000. The contribution of establishment births and deaths to employment growth. Journal of Business and Economic Statistics 18 (1): 113–26. Spletzer, J. R., R. J. Faberman, A. Sadeghi, D. M. Talan, and R. L. Clayton. 2004. Business employment dynamics: New data on gross job gains and losses. Monthly Labor Review 127 (4): 29–42.

5 The LEHD Infrastructure Files and the Creation of the Quarterly Workforce Indicators John M. Abowd, Bryce E. Stephens, Lars Vilhuber, Fredrik Andersson, Kevin L. McKinney, Marc Roemer, and Simon Woodcock

5.1 Introduction Since 2003, the U.S. Census Bureau has published a new and innovative statistical series: the Quarterly Workforce Indicators (QWI). Compiled from administrative records data collected by a large number of states for both jobs and firms, and enhanced with information integrated from other John M. Abowd is the Edmund Ezra Day Professor of Industrial and Labor Relations at Cornell University, and a research associate of the National Bureau of Economic Research. Bryce E. Stephens is a senior consultant with the economics consulting firm Bates White. Lars Vilhuber is a senior research associate at the Cornell Institute for Social and Economic Research, and a senior research associate in the Longitudinal Employer-Household Dynamics program at the U.S. Census Bureau. Fredrik Andersson is a senior research associate of the Cornell Institute for Social and Economic Research (CISER), and a research fellow of the Longitudinal Employer-Households Dynamics Program (LEHD) of the U.S. Bureau of the Census. Kevin L. McKinney is an economist in the Longitudinal Employer-Household Dynamics program at the U.S. Census Bureau, and an administrator of the California Census Research Data Center. Marc Roemer is a mathematical statistician at the U.S. Census Bureau and independent researcher. Simon Woodcock is an assistant professor of economics at Simon Fraser University, and a consultant for the Cornell Institute for Social and Economic Research (CISER). The authors acknowledge the substantial contributions of the staﬀ and senior research fellows of the U.S. Census Bureau’s Longitudinal Employer-Household Dynamics (LEHD) Program. We thank participants of the 2005 CRIW “Conference on Producer Dynamics: New Evidence from Micro Data” and an anonymous discussant for their comments, and Mark Roberts, Tim Dunne, and Brad Jensen for their valuable input during the editorial process. This document is based in part on a presentation first given at the NBER Summer Institute Conference on Personnel Economics, 2002, by John Abowd, Paul Lengermann, and Lars Vilhuber. It replaces LEHD Technical Paper TP-2002-05-rev1 (Longitudinal EmployerHousehold Dynamics Program 2002). This research is a part of the U.S. Census Bureau’s Longitudinal Employer-Household Dynamics Program (LEHD), which is partially supported by the National Science Foundation Grant SES-9978093 to Cornell University (Cornell Institute for Social and Economic Research), the National Institute on Aging Grant R01 AG018854, and the Alfred P. Sloan Foundation. This research is also partially supported by

149

150

J. Abowd et al.

data sources at the Census Bureau, these statistics oﬀer unprecedented detail on the local dynamics of labor markets. Despite the fine geographic and industry detail, the confidentiality of the underlying micro-data is maintained by the application of new, state-of-the-art protection methods. The underlying data infrastructure was designed by the Longitudinal Employer-Household Dynamics (LEHD) Program at the Census Bureau (Abowd, Haltiwanger, and Lane 2004). The Census Bureau collaborates with its state partners, the suppliers of critical administrative records from the state unemployment insurance programs, through the Local Employment Dynamics (LED) cooperative federal-state program. Although the QWI are the flagship statistical product published from the LEHD Infrastructure Files, the latter have found a much more widespread application. The infrastructure constitutes an encompassing and almost universal data source for individuals and firms of all forty-six currently participating states.1 When complete, the LEHD Infrastructure Files will be the first nationally comprehensive statistical product developed from a universe that covers jobs—a statutory employment relation between an individual and employer—as distinct from ones that cover households (e.g., the Decennial Census of Population and Housing) or establishments (e.g., the Economic Censuses or the Quarterly Census of Employment and Wages [QCEW]). In this chapter, we describe the primary input data underlying the LEHD Infrastructure Files, the methods by which the Infrastructure Files are compiled, and how these files are integrated to create the Quarterly Workforce Indicators. We also provide details about the statistical models used to improve the basic administrative data, and describe enhancements and limitations imposed by both data and legal constraints. Many of the infrastructure and derivative micro-data files are now available within the Research Data Centers of the U.S. Census Bureau, and we indicate these files during the discussion. the National Science Foundation Information Technologies Research Grant SES-0427889, which provides financial resources to the Census Research Data Centers. This document reports the results of research and analysis undertaken by U.S. Census Bureau staﬀ. It has undergone a Census Bureau review more limited in scope than that given to oﬃcial Census Bureau publications. This document is released to inform interested parties of ongoing research and to encourage discussion of work in progress. The views expressed herein are attributable only to the authors and do not represent the views of the U.S. Census Bureau, its program sponsors, Cornell University, or data providers. Some or all of the data used in this paper are confidential data from the LEHD Program. The U.S. Census Bureau supports external researchers’ use of these data through the Research Data Centers (see www.ces.census.gov). For other questions regarding the data, please contact Jeremy S. Wu, Program Manager, U.S. Census Bureau, LEHD Program, Data Integration Division, Room 6H136C, 4600 Silver Hill Rd., Suitland, MD 20233, USA. http://lehd.did.census.gov 1. The number of participating states still increases regularly as new Memoranda of Understanding are signed and new states begin shipping data. As of January 15, 2008, there are 46 states with signed MOUs, and 43 states with public use data available at http:// lehd.did.census.gov/

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

151

The QWI use a bewildering array of data sources—administrative records, demographic surveys and censuses, and economic surveys and censuses. The Census Bureau receives Unemployment Insurance (UI) wage records and ES-202 (QCEW) establishment records from each state participating in the LED federal/state partnership. The Census Bureau then uses these products to integrate information about the individuals (place of residence, sex, birth date, place of birth, race, education) with information about the employer (place of work, industry, employment, sales). Not all of the integration methods are exact one-to-one matches based on stable identifiers. In some cases, statistical matching techniques are used, and in other cases critical linking values are imputed. Throughout the process, critical imputations are done multiple times, improving the accuracy of the final estimates and permitting an assessment of the additional variability due to the imputations. Data integration is a two-way street. Not only do the Census Bureau’s surveys and censuses improve the detail on the administrative files, allowing the creation of new statistical products without any increase in respondent burden, but also as a part of its Title 13 mission, the Census Bureau uses the integrated files to improve its other demographic and economic products. The demographic products that have been improved include the Current Population Survey, the Survey of Income and Program Participation, and the American Community Survey. In addition, the LEHD Infrastructure Files are used for research to improve the Census Bureau’s Business Register, which is the sampling frame for all its economic data and the initial contact frame for the Economic Censuses. We give an overview of the diﬀerent raw data inputs and how they are treated and adjusted in section 5.2. In a system that focuses on the dynamics at the individual, establishment, and firm level, proper identification of the entities is important, and we briefly highlight the steps undertaken to edit and verify the identifiers. A more detailed analysis of the longitudinal editing of individual record identifiers using probabilistic record linking has been published elsewhere (Abowd and Vilhuber 2005). The raw data are then aggregated and standardized into a series of component files, which we call the “Infrastructure Files,” as described in section 5.3. Finally, sections 5.4 and 5.5 illustrate how these Infrastructure Files are brought together to create the QWI. It will soon become clear to the reader that the level of detail potentially available with these statistics requires special attention to the confidentiality of the micro data supplied by the underlying entities. How their identities and data are protected is described in section 5.6. Many of the files described in this chapter are accessible in either a public-use or restricted-access version. A brief description of these files with pointers to more detailed documentation is provided in section 5.8. Section 5.9 concludes and provides a glimpse at the ongoing research into improving the infrastructure files.

152

J. Abowd et al.

We should note that this chapter has far too few authors. Over the years, many individuals have contributed to the eﬀort documented in this chapter. As far as we are aware, in addition to the authors of this chapter, the following individuals, who are or were part of the LEHD Program and other parts of the Census Bureau, contributed to the design, implementation, and dissemination of the Infrastructure Files and the Quarterly Workforce Indicators. We thank Romain Aeberhardt, Charlotte Andersson, Matt Armstrong, B.K. Atrostic, Sasan Bakhtiari, Nancy Bates, Gary Benedetto, Melissa Bjelland, Lisa Blumerman, Holly Brown, Bahattin Buyuksahin, Barry Bye, John Carpenter, Nick Carroll, Pinky Chandra, Hyowook Chiang, Karen Conneely, Rob Creecy, Anja Decressin, Pat Doyle, Lisa Dragoset, Robert Dy, Colleen Flannery, Matt Freedman, Kaj Gittings, Cheryl Grim, Matt Graham, Matt Harlin, Sam Hawala, Sam Highsmith, Tomeka Hill, Rich Kihlthau, Charlene Leggieri, Paul Lengermann, Cyr Linonis, Cindy Ma, Jennifer Marks, Kristin McCue, Erika McEntarfer, John Messier, Harry Meyers, Jeronimo Mulato, Dawn Nelson, Nicole Nestoriak, Sally Obenski, Robert Pedace, Barry Plotch, Ron Prevost, George Putnam, Bryan Ricchetti, Kristin Sandusky, Lou Schwarz, David Stevens, Martha Stinson, Cynthia Taeuber, Jan Tin, Dennis Vaughn, Pete Welbrock, Greg Weyland, Karen Wheeless, Bill Winkler, and Laura Zayatz. In addition, continuing guidance was provided by Census Bureau executive staﬀ and Senior Research Fellows, including Chet Bowie, Cynthia Clark, Gerald Gates, Nancy Gordon, John Haltiwanger, Hermann Habermann, Ron Jarmin, Brad Jensen, Frederick Knickerbocker, Julia Lane, Tom Mesenbourg, Paula Schneider, Rick Swartz, John Thompson, Dan Weinberg, and Jeremy Wu. 5.2 Input Files The LEHD Infrastructure File system is, fundamentally, a job-based frame designed to represent the universe of individual-employer pairs covered by state unemployment insurance system reporting requirements.2 Thus, the underlying data are wage records extracted from Unemployment Insurance (UI) administrative files from each LED partner state. In addition to the UI wage records, LED partner states also deliver an extract of the file reported to the Bureau of Labor Statistic’s Quarterly Census of Employment and Wages (QCEW, formerly known as ES-202). These data are received by LEHD on a quarterly basis, with historical time series extending back to the early 1990s for many states. 2. The frame is intended to be comprehensive for legal employment relations and selfemployment. Current development eﬀorts include the addition of federal employment via records provided by the Oﬃce of Personnel Management and the addition of self-employment via records constructed from the Employer and Nonemployer Business Registers.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

153

5.2.1 Wage Records: UI Wage records correspond to the report of an individual’s UI-covered earnings by an employing entity, identified by a state UI account number (recoded to the State Employer Identification Number [SEIN] in the LEHD system). An individual’s UI wage record is retained in the processing if at least one employer reports earnings of at least one dollar for that individual during the quarter. Thus, an in-scope job must produce at least one dollar of UI-covered earnings during a given quarter in the LEHD universe. Maximum earnings reported are defined in a specific state’s unemployment insurance system, and observed top-coding varies across states and over time. A record is completed with information on the individual’s Social Security Number (later replaced with the Protected Identification Key [PIK] within the LEHD system), first name, last name, and middle initial. A few states include additional information: the firm’s reporting unit or establishment (recoded to SEINUNIT in the LEHD system), available for Minnesota, and a crucial component to the Unit-to-Worker imputation described later; weeks worked, available for some years in Florida; hours worked, available for Washington state. Current UI wage records are reported for the quarter that ended approximately six months prior to the reporting date at Census (the first day of the calendar quarter). Wage records are also reported for the quarter that the state considers final in the sense that revisions to its administrative UI wage record database after that date are relatively rare. This quarter typically ends nine months prior to the reporting date. Historical UI wage records were assembled by the partner states from their administrative record backup systems. 5.2.2 Employer Reports: ES-202 The employer reports are based on information from each state’s Department of Employment Security. The data are collected as part of the Covered Employment and Wages (CEW) program, also known as the ES202 program, which is jointly administered by the U.S. Bureau of Labor Statistics (BLS) and the Employment Security Agencies in a federal-state partnership. This cooperative program between the states and the federal government collects employment, payroll, economic activity, and physical location information from employers covered by state unemployment insurance programs and from employers subject to the reporting requirements of the ES-202 system. The employer and workplace reports from this system are the same as the data reported to the BLS as part of the Quarterly Census of Employment and Wages (QCEW), but are referred to in the LEHD system by their old acronym, ES-202. The universe for these data is a reporting unit, which is the QCEW establishment—the place

154

J. Abowd et al.

where the employees actually perform their work. Most employers have one establishment (single-units), but most employment is with employers who have multiple establishments (multi-units). One report per establishment per quarter is filed.3 The information contained in the ES-202 reports has increased substantially over the years. Employers report wages subject to statutory payroll taxes on this form, together with some other information. Common to all years, and critical to LEHD processing, are information on the employer’s identity (the SEIN), the reporting unit’s identity (SEINUNIT), ownership information, employment on the twelfth of each month covered by the quarter, and total wages paid over the course of the quarter. Additional information pertains to industry classifications (initially Standard Industrial Classification [SIC] and later, North American Industry Classification System [NAICS]). Other information includes the federal Employer Identification Number (EIN), and geography both at an aggregated civil level (county or Metropolitan Statistical Area [MSA]) and at a detailed level (physical location street address and mailing address). A recent expansion of the standard report’s record layout has increased the informational content substantially. 5.2.3 Administrative Demographic Information: PCF and CPR The UI and ES-202 files are the core data files describing the economic activity of individuals, jobs, and employers. Although these files contain a tremendous amount of detail on the economic activity, they contain little or no demographic information on the individuals. Demographic information comes from two administrative data sources—the Person Characteristics File (PCF) and the Composite Person Record (CPR), compiled by the Planning, Research, and Evaluation Division at the Census Bureau.4 The PCF contains information on sex, date of birth, place of birth, citizenship, and race, most of which is extracted from the Social Security Administration’s Numident file—the database containing application information for Social Security Numbers (SSN) sorted in SSN order. The CPR information contains annual place of residence data compiled from the Statistical Administrative Records System (StARS). 5.2.4 Demographic Product Integration As part of the integration of individual and household demographic information, the LEHD system uses the fact that many individuals were part of respondent households in the Survey of Income and Program Partici3. These data are also used to compile the Covered Employment and Wages (CEW) and Business Employment Dynamics (BED) data at the BLS. 4. This Division has now been reconstituted as part of the Data Integration Division in the Demographic Programs Directorate at the Census Bureau.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

155

pation (SIPP) or the March Current Population Survey (CPS). Identifier information from the 1984, 1990–1993, and 1996 SIPP panels as well as from March Demographic Supplement to the CPS from 1983 forward have been integrated into the system. See the discussion of the Individual Characteristics File system in section 5.3.2. 5.2.5 Economic Censuses and Annual Surveys Integration The LEHD Infrastructure Files include a crosswalk between the SEIN/ SEINUNIT and the federal Employer Identification Number (EIN). This crosswalk can be used to integrate data from the 1987, 1992, 1997, and 2002 economic censuses, all annual surveys of manufacturing, service, trade, transportation, and communication industries and selected, approved fields from the Census Bureau’s Employer and Nonemployer Business Registers. The integration is used for research to improve the economic activity and geocoding information in both the Infrastructure Files and the Business Registers. The integration of these data is based upon exact EIN matches, supplemented with statistical matching to recover establishments. See the discussion of the Business Register Bridge in section 5.8.2. 5.2.6 Identifiers and Their Longitudinal Consistency Both the wage records and employer reports are administrative datacomprehensive, but sometimes less than perfect. Spurious changes in the entity identifiers (Social Security Number for individuals, SEIN/SEINUNIT for employers and establishments) used for longitudinal matching can have a significant impact on most economic uses of the data. This section discusses the procedures implemented in the LEHD Infrastructure Files to detect, edit, and manage these identifiers. Scope of Data and Identifiers In the LEHD system, a person is identified initially by Social Security Number, and later by the Protected Identification Key (PIK). This identifier is national in scope, and individuals can be tracked across all states and time periods. Not all individuals are in-scope at all times. To be included in the wage record database, an individual’s job must be covered by the reporting requirements of the state’s unemployment insurance system. The prime exclusions are agriculture and some parts of the public sector, particularly federal, military, and postal works. Coverage varies across states and time, although on average, 96 percent of all private-sector jobs are covered. The BLS Handbook of Methods (Bureau of Labor Statistics l997a) describes UI coverage as “broad and basically comparable from state to state,” and claims “over 96 percent of total wage and salary civilian jobs” were covered in 1994. Stevens (2007) provides a survey of coverage for a subset of the current participant states in the LEHD system. An employer is identified primarily by its state UI account number (re-

156

J. Abowd et al.

coded to SEIN). A single legal employer might have multiple SEINs, but regardless of its operations in other states a legal employer has a diﬀerent unemployment insurance account in each state in which it has statutory employees. In particular, the QWI are based exclusively on SEIN-based entities and their associated establishments. Since the SEIN is specific to a state, the QWI does not account for simultaneous activity of individuals across state lines, but within the same multi-state employer. Such activity appears as distinct jobs in the universe. Time-consistency is also not guaranteed, since the UI account number associated with an employer can also change (see later discussions). Although the QWI are based on SEIN/SEINUNIT establishments, this restriction does not apply to the Infrastructure Files themselves. Using the federal EIN, reported on the ES-202 extract and stored on the Employer Characteristics File (ECF) and the Business Register Bridge (BRB), research links to the Census Employer and Nonemployer Business Registers (BR) permit analyses that map entities from the QCEW universe to the Census establishment universe even when the employer-entity operates across state lines. (See section 5.8.2 for more information on the Business Register Bridge.) Error Correction of Person Identifiers Coding errors in the SSN can occur for a variety of reasons. A survey of fifty-three state employment security agencies in the United States over the 1996–1997 time period found that most errors are due to coding errors by employers, but that when errors were attributable to state agencies, data entry was the culprit (Bureau of Labor Statistics 1997b). The report noted that 38 percent of all records were entered by key entry, while another 11 percent were read in by optical character readers (OCRs) Optical character readers and magnetic media tend to be less prone to errors. Errors can be random digit coding errors that do not persist, typically generated when data are transferred from one format (paper) to another (digital), or they can be persistent, typically occurring when a firm’s payroll system contains an erroneous SSN. While the latter is harder to identify and to correct, the LEHD system uses statistical matching techniques, primarily probabilistic record linking, to correct for spurious and nonpersistent coding errors. The incidence of errors and the success rate of the error correction methods diﬀers widely by state. In particular, it depends critically on the quality of the available individual name information on the wage records. Abowd and Vilhuber (2005) describe and analyze the LEHD SSN editing process as it was applied to data provided by the state of California. The process verified over half a billion records for that state and is now routinely applied to all states in the LEHD Infrastructure Files. The number of records that are recoded is slightly less than 10 percent of the total num-

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

157

ber of unique individuals appearing in the original data, and only a little more than 0.5 percent of all wage records. The authors estimate that the true error rate in the data is higher, in part due to the conservative setup of the process. Over 800,000 job history interruptions in the original data are eliminated, representing 0.9 percent of all jobs, but 11 percent of all interrupted jobs. Despite the small number of records that are found to be miscoded, the impact on flow statistics can be large. Accessions in the uncorrected data are overestimated by 2 percent, and recalls are biased upwards by nearly 6 percent. Payroll for accessions and separations are biased upward by up to 7 percent. The wage record editing occurs prior to the construction of any of the Infrastructure Files for two reasons. First, the wage record edit process requires access to the original Social Security Numbers as well as to the names on the wage records, both of which, because they are covered by the Privacy Act, are replaced by the Protected Identification Key (PIK) early in the processing of wage records. The PIK is used for all individual data integration. The original SSN and the individual’s name are not part of the LEHD Infrastructure Files. Second, because the identifier changes underlying the wage record edit are deemed spurious, and because individuals have no economic reason at all to change Social Security Numbers, there is little ambiguity about the applicability of the edit. This is diﬀerent from the editing of employer identifiers, as shown in section 5.3. The Census Bureau designed the PIK as a replacement for the Privacy Act-protected SSN. The PIK itself is a random number related to the SSN solely through a one-to-one correspondence table that is stored and maintained by the Census Bureau on a computing system that is isolated from all LEHD systems and from most other systems at the Census Bureau. To avoid any commingling of SSN-laden data with PIK-laden data, which might compromise the protection aﬀorded by the PIK, the wage record editing process takes place in a secure computing area distinct from the rest of the LEHD processing. Correcting for Changes in Firm Identifiers Firms in the QCEW system are identified by a UI account number assigned by the state. As with all employer identifiers, an account number can change over time for a number of reasons, not all of which are due to economically meaningful changes. State administrative units take great care to follow the legal entities in their system, but account numbers may nevertheless change for reasons which economists may not consider legitimate economic reasons. For instance, a change in ownership of a firm without any change in economic activity may lead to a change in the account number. Often, but not always, such a change is noted in the successor/predecessor fields of the ES-202 record. Other times, without changes in ownership, employees migrate en masse from one UI account to another. In this

158

J. Abowd et al.

case, one might make a reasonable inference that there were continuous economic operations. Because changes in the employer identifiers are correlated with some elements of economic choice, albeit imperfectly, these identifiers are managed in the LEHD Infrastructure File system. Because the system is designed to operate from regular reports of the administrative record systems in the partner states, the original employer identifiers must be retained in all files in the system. The LEHD system then builds a database of entity demographics that traces the formal successor/predecessor relations among these identifiers. In addition, entity-level summary inferences about undocumented successor/predecessor relations, which are based on worker flow statistical analysis, are also stored in this entity demography database. An auxiliary file, the Successor-Predecessor File (SPF), is created from the entity demographic histories and used to selectively apply successor/predecessor edits to the input files for the QWI. Handling the entity identifiers in this manner allows the LEHD system to receive and integrate updates of input data from partner state (because these share common entity identifiers) and to purge statistical analyses of the spurious changes due to noneconomic changes in the entity demography over time. Benedetto et al. (2007) provide more detail on the development of the SPF and its validity. The SPF is described in more detail later in this chapter. 5.3 Infrastructure Files This section describes the creation of the core Infrastructure Files from the raw input files. These files form the core of the integrated system that supports the job-based statistical frame that LEHD created. Each Infrastructure File is integrated into the system with longitudinally consistent identifiers that satisfy fundamental database rules, allowing them to be used as unique record keys. Thus, the core Infrastructure File system can be used to create valid statistical views of data for jobs, individuals, employers, or establishments. The system is programmed entirely in SAS and all files are maintained in SAS format with SAS indices. The raw input files, quarterly UI wage records, and ES-202 reports are first standardized.5 The UI wage record files are edited for longitudinal identifier consistency, and the SSN is then replaced by the PIK. The ES-202 files are standardized, but no identifier or longitudinal edits are performed at this stage. Thus, the raw input files with only the edits noted here are preserved for future research. Beyond these standardizing steps, no further processing of the raw files occurs. Instead, all the editing and imputation are done in the process of building the Infrastructure Files. The LEHD system builds the Infrastructure Files from the standardized 5. The ES-202 files, in particular, have been received in a bewildering array of physical file layouts and formats, reflecting the wide diversity in computer systems installed in state agencies.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

159

input files augmented by a large number of additional Census-internal demographic and economic surveys and censuses. The Employment History File (EHF) provides a full time series of earnings at all within-state jobs for all quarters covered by the LEHD system and provided by the state.6 It also provides activity calendars at a job, SEINUNIT, and SEIN level. The Individual Characteristics File (ICF) provides time-invariant personal characteristics and some address information.7 The Employer Characteristics File (ECF) provides a complete database of firm and establishment characteristics, most of which are time-varying. The ECF includes a subset of the data available on the Geocoded Address List (GAL), which contains geocodes for the block-level Census geography and latitude/longitude coordinates for the physical location addresses from a large set of administrative and survey data, including address information in the ES-202 input files. We will describe each of these files in detail in this section. 5.3.1 Employment History File: EHF The Employment History File (EHF) is designed to store the complete in-state work history for each individual that appears in the UI wage records. The EHF for each state contains one record for each employeeemployer combination—in other words, a job—in that state in each year. Both annual and quarterly earnings variables are available in the EHF. Individuals who never have strictly positive earnings at their employing SEIN (a theoretical possibility) in a given year do not have a record in the EHF for that year. The EHF data are restructured into a file containing one observation per job (PIK-SEIN combination), with all quarterly earnings and activity information available on that record. The restructured file is called the Person History File (PHF).8 An active job within a quarter, the primary job-level economic activity measure, is defined as having strictly positive quarterly earnings for the individual-employer pair that define the job. A similar time series, based on observed activity (positive employment) in the ES-202 records, is computed at the SEINUNIT level (UNIT History File, UHF) and the SEIN level (SEIN History File, SHF). At this stage of the data processing the first major integrated quality con6. The earliest data accepted by the LEHD system are 1990, quarter 1. Most states provided data beginning some time in the early 1990s. All partner states provide data beginning in 1997, quarter 1. Current input raw data files are delivered six months after the close of the quarter. The QWI data are produced within three months of the receipt of the raw input files from the unemployment insurance system. The LEHD system maintains all of the data reported by a partner state (or nationally for the national files). The QWI system uses as much of these data as possible. 7. A longitudinal enhancement of the ICF, which updates residential address information annually and contains some data from 2000 Census of Population and Housing, is under development. 8. It should be noted that the actual file structure is at the PIK-SEIN-SEINUNIT-YEAR level for the EHF, and at the PIK-SEIN-SEINUNIT level for the PHF. Although only one state (Minnesota) has nonzero values for SEINUNIT, this allows the file structure to be homogeneous across states.

160

J. Abowd et al.

trol checks occur. The system performs a quarter-by-quarter comparison of the earnings and employment information from the UI wage records (beginning-of-quarter employment, see the appendix for the definition, and total quarterly payroll) and ES-202 records (month one employment and total quarterly payroll). Large discrepancies in any quarter are highlighted and the problematic input files are passed to an expert analyst for study. Discrepancies that have already been investigated and that will, therefore, be automatically corrected in the subsequent processing of a state’s data are allowed to pass. Other discrepancies are investigated by the analyst. The analyst’s function is to find the cause of the discrepancy and take one of three courses of action: • Arrange for corrected data from the state supplier. • Develop an edit that can be applied to correct the problem. • Flag the data as problematic so that they are not used in the QWI estimation system. The first two actions result in a continuation of the Infrastructure File processing and no change in the QWI estimation period. The third action results in continuation of the Infrastructure File processing and either the suppression of a state’s QWI data until the problem can be corrected or a shortening of the time period over which QWI data are produced for that state. Often, a state-supplied corrected data file is imported into the LEHD system. Equally often, a state-specific edit is built into the data processing. Each time the state’s data are reprocessed, this edit is invoked. Unfortunately, not all data discrepancies can be resolved. Then, the third action occurs. In particular, the state’s archival historical UI wage record and ES202 data are sometimes permanently damaged or defective. In these cases, the data have been lost or permanently corrupted. The quality control during the EHF processing identifies the state and quarter when such problems occur. In the current Infrastructure File system, such data are not used for the QWI estimation but may be used by analysts for specific research projects. In the course of such research projects, the analyst often develops a statistical method for improving the defective data. These improvements are then ported into the Infrastructure File system.9 5.3.2 Individual Characteristics File: ICF The Individual Characteristics File (ICF) for each state contains one record for every person who is ever employed in that state over the time period spanned by the state’s unemployment insurance records. 9. For example, research on wage dynamics associated with estimates of firm-level human capital use has produced a statistical missing data edit for the UI wage records that detects missing wage records and imputes them by drawing from an appropriate posterior predictive distribution. The statistical models that detect and correct this problem will be imported into a future version of the EHF Infrastructure File.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

161

The ICF is constructed in the following manner. First, the universe of individuals is defined by compiling the list of unique PIKs from the EHF. Basic demographic information from the PCF is merged using the PIK, and records without a valid match are flagged. PIK-survey identifier crosswalks link the CPS and SIPP ID variables into the ICF. Sex and age information from the CPS is used to complement and verify the PCF-provided information. Age and Sex Imputation Approximately 3 percent of the PIKs found in the UI wage records do not link to the PCF. Multiple imputation methods are used to impute date of birth and sex for these individuals. To impute sex, the probability of being male is estimated using a state-specific logit model: (1)

P(male) f(Xiss)

where Xis contains a full set of yearly log earnings and squared log earnings, and full set of employment indicators covering the time period spanned by the state’s records, for each individual i with strictly positive earnings within state s and non-missing PCF sex. The state-specific ˆs, as estimated from equation (1), is then used to predict the probability of being male for individuals with missing sex within state s, and sex is assigned as (2)

ˆ male if Xis s l

where l ~ U [0, 1] is one of l 1, . . . , 10 independent draws from the distribution. Thus, each individual with missing sex is assigned ten independent missing data implicates, all of which are used in the QWI processing.10 The imputation of date of birth is done in a similar fashion using a multinomial logit to predict the probability of being in one of eight birth date decades and then assigning a birth date within decade based on this probability and the distribution of birth dates within the decade. Again, ten implicates are imputed for birth date. If an individual is missing sex or birth date in the PCF, but not in the CPS, then the CPS values are used, not the imputed values. Before the imputation model for date of birth is implemented, basic editing of the date of birth variable eliminates obvious coding errors, such as a negative age at

10. Note that this imputation does not account for estimation error in ˆ. This was one of the first missing data imputations developed at LEHD. At the time, techniques for sampling from the posterior predictive distribution of a binary outcome where the likelihood function is based on a logistic regression were not feasible on the LEHD computer system. Since only three percent of the observations in the ICF are subject to this missing data edit, it was implemented as described in the text. A longitudinal, enhanced ICF is under development (see section 5.9). All missing data imputations in the new ICF will be performed by sampling from an appropriate posterior predictive distribution. This will properly account for estimation error.

162

J. Abowd et al.

the time when UI earnings are first reported for the individual. In those relatively rare cases where the date of birth information is deemed unrealistic, birth date is set to missing and imputed based on the model described previously. Place of Residence Imputation Place of residence information on the ICF is derived from the StARS (Statistical Administrative Records System), which for the vast majority of the individuals found in the UI wage records contains information on the place of residence down to the exact geographical coordinates. However, in less than ten percent of all cases the geography information is incomplete or missing. Since the QWI estimation relies on completed place of residence information, because this information is a critical conditioning variable in the unit-to-worker (U2W) imputation model (see section 5.4.2), all missing residential addresses are imputed. County of residence is imputed based on a categorical model of the data that is a fully saturated contingency table. Separately for each state, unique combinations of categories of sex, age, race, income, and county of work are used to form i 1, . . . , I populations. For each sample i, the probability of residing in a particular county as of 1999, ij, is estimated by the sample proportion, pij nij /ni , where j 1, . . . , J indexes all the counties in the state plus an extra category for out-of-state residents. County of residence is then imputed based on (3)

county j if Pij1 uk Pij

where Pi is the CDF corresponding to pi for the ith population and kl ~ U [0, 1] is one of k 1, . . . , 10 independent draws for the i th individual belonging to the ith population.11 In its current version, no geography below the county level is imputed and in those cases where exact geographical coordinates are incomplete the centroid of the finest geographical area is used. Thus, in cases where no geography information is available this amounts to the centroid of the imputed county. Geographical coordinates are not assigned to individuals whose county of residence has been imputed to be out-of-state. Education Imputation The imputation model for education relies on a statistical match between the Decennial Census 1990 and LEHD data. The probability of belonging to one of thirteen education categories is estimated using 1990 Decennial data conditional on characteristics that are common to both Decennial and LEHD data, using a state-specific logit model: 11. The longitudinal, enhanced ICF that is under development augments the model in the text with a Dirichlet prior distribution for the Pij. The imputations are then made by sampling from the posterior predictive distribution, which is also Dirichlet.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

(4)

163

P(educat ) f(Zis s)

where Zis contains age categories, earnings categories, and industry dummies for individuals age fourteen and older in the 1990 Census Long Form residing in the state being estimated, and who reported strictly positive wage earnings. The education category is imputed based on (5)

educat j if cpj1 l cpj

where cpj Zis s and l ~ U [0, 1] is one of l 11, . . . , 20 independent draws, and i ∈ EHF.12 5.3.3 The Geocoded Address List: GAL The Geocoded Address List (GAL) is a file system containing the unique commercial and residential addresses in a state geocoded to the Census block and latitude/longitude coordinates. The file encompasses addresses from the state ES-202 data, the Census Bureau’s Employer Business Register (BR), the Census Bureau’s Master Address File (MAF), the American Community Survey Place of Work file (ACS-POW), the American Housing Survey (AHS) and others. Addresses from these source files are processed by geocoding software (Group1’s Code1), address standardizers (Ascential/Vality), and record-matching software (Ascential/Vality) for unduplication. The remaining processing is done in SAS and the final files are in SAS format. The final output file system consists of the address list and a crosswalk for each processed file-year. The GAL contains each unique address, identified by a GAL identifier called GALID, its geocodes, a flag for each fileyear in which it appears, data quality indicators, and data processing information, including the release date of the Geographic Reference File (GRF). The GAL Crosswalk contains the ID of each input entity and the ID of its address (GALID). Geographic Codes and Their Sources A geocode on the GAL is constructed as the concatenation of FIPS (Federal Information Processing Standard) state, county and Census tract: FIPS-state (2) || FIPS-county (3) || Census-tract (6) This geocode uniquely identifies the Census tract in the United States. The tract is the lowest level of geography recommended for analysis. The Census block within the tract is also available on the GAL, but the uncertainties in block-coding make some block-level analyses unreliable. Geocoding 12. In the longitudinally enhanced ICF that is under development, this imputation is replaced by a probablistic record link to Census 2000 long form data. Approximately one person in six acquires directly reported educational attainment as of 2000. The remaining individuals get 10 multiple imputations from a Dirichlet/Multinomial posterior predictive distribution.

164

J. Abowd et al.

Table 5.1

Value

Sources of geocodes on GAL Typical percent

C

12.20

M E

81.86 0.00

W

0.03

O S I D missing

1.23 1.17 0.01 0.00 3.50

Meaning Code1, or the address matches an address for which Code1 supplied the block code The MAF—the address is a MAF address or matches a MAF address The MAF, the street address is exactly the same as a MAF address in the same tract The MAF, the street address is between 2 MAF addresses on the same block face Imputed using the distribution of commercial addresses in the tract Imputed using the distribution of residential addresses in the tract Imputed using the distribution of mixed-use addresses in the tract Imputed using the distribution of all addresses in the tract Block code is missing

100.00

to the block allows the addition of all the higher-level geocodes associated with the addresses. Latitude and longitude coordinates are also included in the file.13 Block Coding. Block coding is achieved by a combination of geocoding software (Group1’s Code1), a match to the MAF, or an imputation based on addresses within the tract. Table 5.1 describes the typical distribution of geocode sources. In all states processed to date, except California, no address required the D method. That is, almost every tract where an address lacks a block code contains commercial, residential, and mixed-use addresses. The Census Bureau splits blocks to accommodate changes in political boundaries. Most commonly, these are place boundaries (a place is a city, village, or similar municipality). The resulting block parts are identified by 2 suﬃxes, each taking a value from A to Z. The GAL assigns the block part directly from the MAF, or by using the one whose internal point is closest to the address by the straight-line distance. The GAL also provides the following components of the geocodes as separate variables, for convenience: Federal Information Processing Standards (FIPS) code (5 digits), FIPS state code (the first 2 digits of the FIPS code), FIPS county code within state (the rightmost 3 digits of the FIPS code), and Census tract code (a tract within the county, a 6-digit code). Higher-level geographic codes originate from the Block Map File (BMF). 13. An enhanced geocoding system was developed for the newer LEHD product called OnTheMap, which published to the block level. These enhancements are being integrated into an enhanced version of the GAL, which will be used for both QWI and OnTheMap.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

165

The BMF is an extract of the GRF-C (Geographic Reference File-Codes). All geocodes are character variables. Federal Information Processing Standards (FIPS) codes are unique within the United States; Census codes are not. Table 5.2 lists the available higher-level geocodes. Geographic Coordinates. The geographic coordinates of each address are available as latitude and longitude with 6 implied decimals. The coordinates are not always as accurate as 6 decimal places implies. An indicator of their quality is provided. Table 5.3 provides the typical distribution of codes, which range from 1 (highest quality) to 9 (lowest quality). Variables indicating the source of the geographic coordinates (Block internal point, geocoding software, MAF, or otherwise derived) are also available. Most coordinates are provided by either commercial geocoding software or the MAF. Finally, a set of flags also indicates, for each year and source file, whether an address appears on that file. For example, the flag variable b1997 equals 1 if the address is on the 1997 BR; otherwise it equals 0. As another example, if a state partner supplies 1991 ES-202 data with no address information, then e1991 will be 0 for all addresses. In a typical GAL year, between 3 and 6 percent of addresses are present on that year’s ES202 files, between 4 and 10 percent are present on a specific BR year file, and between 80 and 90 percent are present on the MAF. Less than one percent of addresses are found on the ACS-POW and AHS data, because these are sample surveys. Note that this distribution indicates where the GAL found a geocoded address, not the percentage of addresses that could be geocoded. Table 5.2

Higher-level geocodes on GAL

a_ fipsmcd a_mcd a_ fipspl a_ place a_msapmsa a_wib

5-digit FIPS Minor Civil Division (a division of a county) 3-digit Census Minor Civil Division (a division of a county) 5-digit FIPS Place 4-digit Census Place Metropolitan-Statistical-Area(4)—Primary-Metropolitan-Statistical-Area(4) 6-digit Workforce Investment Board area

Table 5.3

Quality of geographic coordinates

Value

Typical percent

Meaning

80.15 1.59 10.12 4.65 3.50

Rooftop or MAF (most accurate) ZIP4 or block face, block face is certain Block group is certain Tract is certain Coordinates are missing

1 2 3 4 9

100.00

166

J. Abowd et al.

Accessing the GAL: The GAL Crosswalks The GAL crosswalks allow data users to extract geographic and address information about any entity whose address went into the GAL. Each crosswalk contains the identifiers of the entity, its GALID, and sometimes flags. To attach geocodes, coordinates, or address information to an entity, users merge the GAL crosswalk to the GAL by GALID, selecting only observations existing on the required entities on the GAL crosswalk. Then they merge the resulting file to the entities of interest using the entity identifiers. An entity whose address was not processed (because it is out of state or lacks address information) will have blank GAL data. Table 5.4 lists the entity identifiers by data set or survey. 5.3.4 The Employer Characteristics File: ECF The Employer Characteristics File (ECF), which is actually a file system, consolidates most employer and establishment-level information (size, location, industry, etc.) into two files. The employer SEIN-level file contains one record for every year-quarter in which a SEIN is present in either the ES-202 or the UI wage records, with more detailed information available for the establishments of multi-unit SEINs in the SEINUNIT-level file. The SEIN file is built up from the SEINUNIT file and contains no additional information, but is an easier and more eﬃcient way to access SEINlevel summary data. A number of inputs are used to build the ECF. The primary input is the ES-202 data. Unemployment Insurance (UI) wage record summary data are used to supplement information from the ES-202; in particular, SEINlevel employment (beginning of quarter, see appendix, for definitions) and quarterly payroll measures are built from the wage records. Unemployment Insurance (UI) wage record data are also used to supplement published BLS county-level employment data, which are used to construct weights for use in the QWI processing. Geocoded address information

Table 5.4

GAL crosswalk entity identifiers

Dataset

Entity identifier variables

AHS ES-202

control and year sein, seinunit, year, and quarter

ACS-POW BR

acsfileseq, cmid, seq, and pnum. cfn, year, and singmult

MAF

mafid and year

Note

e_ flag = p for physical addresses, e_ flag = m for mailing addresses as source of address info singmult indicates whether the entity resides in the single-unit (su) or the multi-unit (mu) data set. b_ flag = P if physical address, b_ flag = M for mailing address.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

167

from the GAL file contributes latitude-longitude coordinates of most establishments, as well as updated Workforce Investment Board (WIB) area and MSA information. The state-provided extracts from the BLS Longitudinal Database (LDB) and LEHD-developed imputation mechanisms are used to backfill NAICS information for periods in which NAICS was not collected. Finally, the QWI disclosure avoidance mechanism is initiated in the ECF. We will describe basic methods for constructing the ECF in the next section. Details of the NAICS imputation algorithm are described in the section titled “NAICS Codes on the ECF”. The entire disclosure-proofing mechanism is described in section 5.6. Constructing the ECF ECF processing starts by integrating yearly summary files for each SEIN and SEINUNIT in the ES-202 data files. General and state-specific consistency checks are then performed. The county, NAICS, SIC, and federal EIN data are checked for invalid values. The industry code edit goes beyond a simple validity check. If a four-digit SIC code or NAICS industry code (six-digit) is present, but is not valid, then the industry code undergoes a conditional missing data imputation based on the first two and three (SIC) or three, four, and five (NAICS) digits.14 All other invalid or missing industry codes are subjected to the longitudinal edit and missing data imputation described in the following paragraphs. Based on the EHF, SEIN-level quarterly employment (beginning of quarter) and payroll totals are computed. Unemployment Insurance (UI) wage record data are used as an imputation source for either payroll or employment in the following situations: • If ES-202 month one employment is missing, but ES-202 payroll is reported, then UI wage record beginning-of-quarter employment is used. • If ES-202 month one employment is zero, then UI employment is not used, since this may be a correct report of zero employment for an existing SEIN. The situation may arise when bonuses or benefits were retroactively paid, even though no employees were actively employed. • If ES-202 quarterly payroll is zero and ES-202 employment is positive, then UI wage record quarterly payroll is used. • If ES-202 quarterly payroll and employment are both zero or both missing, then UI wage record quarterly payroll and beginning-ofquarter employment are used. The ES-202 data contain a master record for multi-unit SEINs, which is removed after preserving information not available in the establishment records. Various inconsistencies in the record structure are also handled at 14. The NAICS 1997 are updated to NAICS 2002. Then, NAICS 2002 are used for the imputation. The same procedure is later used for LDB data.

168

J. Abowd et al.

this stage of the processing. For a single-unit SEIN, which has two records (master and establishment), information from the master records is used to impute missing data items directly for the establishment record. For a multi-unit SEIN, a flat prior is used in the allocation process; missing establishment data are imputed, assuming that each establishment has an equal share of unallocated employment and payroll. A subsequent longitudinal edit reexamines this allocation and improves it if there is historical information that is better than the equal-size assumption. The allocation process implemented above (master to establishments) does not incorporate any information on the structure of the SEIN. To improve on this, SEINs that are missing establishment structure for some periods—but reported a valid multi-unit structure in other periods—are inspected. The absence of information on establishment structure typically occurs when a SEIN record is missing due to a data processing error. A SEIN with a valid multi-unit structure in a previous period is a candidate for structure imputation. The employer’s establishment structure is then imputed using the last available record with a multi-unit structure. Payroll and employment are allocated appropriately. From this point on, the employer’s establishment structure (number of establishments per SEIN) is defined for all periods. Geocoded data from the GAL are incorporated to obtain geographic information on all establishments. Once the multi-unit structure has been edited and the geocoding data have been integrated, the ECF records undergo a longitudinal edit. Geographic data, industry codes (SIC and NAICS), and EIN data from quarters with valid data are used to fill missing data in other quarters for the same establishment (SEINUNIT). If at least one industry variable among the several sources (SIC, NAICS1997, NAICS2002, NAICS 2007, LDB) has valid data, it is used to impute missing values in other fields. Geography, if still missing, is imputed conditional on industry, if available. Counties with larger employment in a SEINUNIT’s industry have a higher probability of being selected. All missing data imputations are single draws from posterior predictive distributions that are multinomial based on an improper uniform Dirichlet prior. The imputation probabilities are the ratio of employment in each possible value to total employment in the support of the distribution.15 For SEINs, the (employment and establishment-weighted) modal values of county, industry codes, ownership codes, and EIN are calculated for 15. The posterior predictive distribution is multinomial because the employment proportions are derived from the population of employing establishments in the quarter, which is assumed to be nonrandom. Only a single imputation is performed because the unit-to-worker missing data model imputes 10 establishments to each job in a multi-unit SEIN. Multiple imputation of the missing data in those establishments would have meant that 100 implicates would have to be processed for each multi-unit job. This processing requirement was deemed impractical for the current QWI system.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

169

each SEIN and year-quarter. The SEIN-level records with missing data are filled in with data from the closest time period with valid data. At this point, if a SEIN mode variable has a missing value, then no information was ever available for that SEIN. Additional attention is devoted to industry codes, which are critical for QWI processing. Missing SIC and NAICS are randomly imputed with probability proportional to the statewide share of employment in four-digit SIC code or five-digit NAICS code. The SIC and NAICS codes with a larger share of employment have a higher probability of selection. If an industry code is imputed, it is done so once for each SEIN and remains constant across time. These industry codes are then propagated to all SEINUNITs as well. With most data items complete, provisional weights are calculated. These weights are discussed in the section on QWI processing (section 5.5). The disclosure avoidance noise-infusion factors are also prepared at the SEIN and SEINUNIT level and added to the ECF at this point. Disclosure avoidance methods are discussed in detail in section 5.6. Imputations in the ECF All employer or establishment data items used in the QWI processing, when missing, are imputed. These items include employer-establishment structure, employment, payroll, geography, industry, ownership, and EIN. This subsection describes these imputations, which are of two types: longitudinal edits—data from another period closest in time to the period with missing data are copied into the missing data items, and probabilistic imputation—missing data are imputed by sampling from a posterior predictive multinomial distribution based on a uniform Dirichlet prior, conditional on as much sample information as possible. The analyst is responsible for developing the likelihood component of the posterior predictive distribution. The employer-establishment structure refers to the structure of establishments within single-unit and multi-unit employers. In the ECF, the SEIN master record summarizes the information from all establishments. This record is either based on the comparable record in the raw ES-202 data (input directly or aggregated from the establishment records), or imputed by calculating summary information on beginning-of-quarter employment and total quarterly payroll directly from all UI wage records in a given quarter that come from the indicated SEIN (in the case where the SEIN does not have a record in the raw ES-202 data for that quarter). In either case, a SEIN master record is always available for every SEIN that exists in a given quarter in either the ES-202 or UI wage record data for that quarter. However, the establishment structure of this SEIN may be missing in a given quarter; that is, the SEINUNITs associated with the SEIN for this quarter are not input directly from the ES-202 data. In this case, the establishment structure is imputed by a longitudinal edit that looks for the

170

J. Abowd et al.

nearest quarter in which the establishment structure is not missing and copies this structure to the quarter with the missing structure. Then, the missing SEINUNIT employment and payroll are imputed from the SEIN master record by proportionally allocating the current quarter SEIN-level values to the SEINUNITs based on the proportions of the same variables in the donor quarter’s establishments. Only longitudinal edits are used in this process. If no donor quarter can be found, then the establishment structure is assumed to be single unit and a single SEINUNIT record is built from the SEIN master record. At this point, the employer-establishment structure is available for all SEINs, and all missing employment and payroll data have been imputed for every SEIN and SEINUNIT that exists in a state’s complete ECF. The ECF records are then geocoded from the GAL, as described in section 5.3.3. Hence, the missing geocode items are completed before the remainder of the missing data in the ECF are imputed. The geography subprocess of the ECF combines information about the entity history with the geocoding information from the GAL. Geocodes in the GAL are determined exclusively by contemporaneous address information, but contain information on the quality of the geocode information—whether a geocode reflects a rooftop geocode, a block, a block group, a tract, or only a county. The ECF geography subprocess takes this information, and applies a longitudinal edit, conditional on the SEINUNIT not changing locations. The inference of a geographical move for a SEINUNIT occurs whenever the geocode delivered by the GAL is diﬀerent for two diﬀerent time periods in a way that is not due to variations in the quality of geography coding. For example, a rooftop and a block group geocode will always necessarily have diﬀerent geocodes. However, if the block groups corresponding to each entity diﬀer, then the system assumes that the entity has physically moved. If the two SEINUNITs have been geocoded to the same block group, the diﬀerence in geocodes is considered a change in geography quality, not a move. Finally, the GALID associated with the best quality geography is copied to all quarters within the nonmove time period for that SEINUNIT. SEINUNITs with missing geography are excluded from the longitudinal edit. These units are assigned geography by a probabilistic imputation based on employment shares across counties given SIC (if the industry for the SEINUNIT is available), or by unconditional employment shares across counties (if it is not available). Each SEINUNIT with missing geography is assigned a pseudo-GALID reflecting the imputed county’s centroid. Additional geographic information (MSA or Core Based Statistical Area [CBSA], and WIB area) is attached to the ECF based on the GALID or pseudo-GALID. At this point all SEINUNIT-level records have completed geocoding. When the records are returned from geocoding, missing industry codes,

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

171

ownership, and EIN are imputed by longitudinal edit, if possible. The values of SIC, NAICS, ownership, and EIN are copied from the nearest nonmissing quarter. No further editing occurs for ownership or EIN. Industry codes that are still missing after the longitudinal edit are imputed with probabilistic methods based on the empirical distribution conditional on the same unit’s observed other industry data items. For instance, if SIC is missing, but NAICS1997 is available, the relative observed distribution of SIC-NAICS1997 pairs is used to impute the missing data item. If all previous imputation mechanisms fail, SIC is imputed unconditionally based on the observed distribution of within-state employment across SIC industries. Once SIC is assigned, the previous conditional imputation mechanisms are again used to impute other industry data items. Geocoding and industry coding are supplied for the SEIN-level record based on the following edit. The unweighted and employment-weighted modal values across SEINUNITs from the same SEIN are computed for WIB, MSA/CBSA, state, county, best sub-county geography, ownership, SIC, NAICS, and EIN. All SEIN-level records get assigned the both modal values (weighted and unweighted) and a researcher or analyst may choose the appropriate value.16 NAICS Codes on the ECF Enhanced NAICS variables on the ECF can be diﬀerentiated by the sources and coding systems used in their creation. There are two sources of data—the ES-202 and the BLS-created LDB—and three coding systems for NAICS—NAICS1997, NAICS2002, and NAICS2007. Every NAICS variable uses at least one source and one coding system. The ESO (ES-202-only) and FNL (final) variables are of primary importance to the user community. The ESO variables use information from the ES-202 exclusively and ignore any information that may be available on the LDB. In section 5.7.2 we provide an analysis on why this may be preferred. The FNL variables incorporate information from both the ES-202 and the LDB, with the LDB being the primary source. The QWI uses FNL variables for its NAICS statistics. Neither ESO nor FNL variables contain missing values. NAICS algorithm precedence ordering. Four basic sources of industry information are available on the ECF: NAICS and NAICS_AUX as well as SIC from ES-202 records, and the LDB-sourced NAICS_LDB codes. The NAICS, NAICS_AUX, and NAICS_LDB, when missing (no valid 6-digit industry code), are imputed based on the following algorithm. The SIC is 16. The employment weighted modal values that are on the SEIN-level record are only used in the QWI processing when the unit-to-work imputation described in the next section fails to impute a SEINUNIT to a job history.

172

J. Abowd et al.

filled similarly. Depending on the imputation used, a miss variable is defined, which is used in building the ESO and FNL variables. 1. Valid 6-digit industry code (miss 0). 2. Imputed code based on first 3, 4, or 5 digits when no valid 6-digit code is available in another period (miss 0). 3. Imputed code based on contemporaneous SIC if SIC changed prior to 2000 (miss 1.5). 4. Valid 6-digit code from another period (miss 2). 5. Valid code from another source (for example, if NAICS1997 is missing, NAICS2002 or SIC may be available) (miss 3). 6. Use employment-weighted SEIN modal value (miss 5 if contemporaneous modal value, miss 7 if the modal value stems from another time period). 7. Unconditional impute (miss 6 if only the SEIN-level modal value is imputed unconditionally, miss 11 if the SEIN-level value was unconditionally imputed and propagated to all SEINUNITs). ESO and FNL Variables. The ESO and FNL variables are made up of combinations of the various sources of industry information. The ESO variable uses the NAICS and NAICS_AUX variables as input. Information from the variable with the lowest miss value is preferred, although in case of a tie the NAICS_AUX value is used. The FNL variable uses the ESO and LDB variables. Information from the variable with the lowest miss value is preferred, although in case of a tie the NAICS_LDB value is used. 5.4 Completing the Missing Job-Level Data The Infrastructure Files contain most of the information necessary to compute the QWI. However, there are two important sources of missing job-level data that must be addressed before those estimates can be formed using substate levels of geography and detailed levels of industry: spurious employer-level identifier changes and missing establishment-level geography and economic activity data. We discuss the edits and imputations associated with these problems in this section. Fundamentally, the QWI are based on the job-level employment histories. Dynamic inconsistencies in these histories that are caused by individual identifier breaks are handled by the wage record edit described previously. Dynamic inconsistencies in these histories that are caused by employer or establishment identifier breaks that are not due to real economic activity are handled by creating the Successor-Predecessor File from the entity demographics, then extracting information from this file to suppress spurious employment and job flows. We describe this process in section 5.4.1.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

173

Job histories that are part of a multi-unit SEIN do not contain the establishment (SEINUNIT) associated with the job, except for the state of Minnesota. This means that establishment-level characteristics—geography and industry, in particular—are missing data for job histories that relate to multi-units. The missing imputation associated with this problem is discussed in section 5.4.2. 5.4.1 Connecting Firms Intertemporally: The Successor-Predecessor File (SPF) The firm identifier used in all of LEHD’s files is a state-specific account number from that state’s unemployment insurance accounting system, used, in particular, to administer the tax and benefits of the UI system. These account numbers, recoded and augmented by a state identifier, become the entity identifier called the SEIN in the Infrastructure File system. The SEINs can, and do, change for a number of reasons, including a change in legal form or a merger. Typically, the separation of a worker from an employer is identified by a change in the SEIN on that worker’s UI wage records. If an employer changes SEINs, but makes no other changes, the worker would appear to have left the original firm even though his or her employment status remains unchanged from an economic viewpoint. These spurious apparent employer changes are known to induce biases in both employment and job flow statistics. For example, a simple change in account numbers would lead to the observation of a firm closing even though all workers remain employed. To identify such events, the Successor Predecessor File (SPF) tracks large worker movements between SEINs. Benedetto et al. (2007) used the SPF for an early analysis in one particular state of the impact of such an exercise. The SPF provides a variety of link characteristics, based on the number of workers leaving a SEIN, in both absolute and relative terms, and the number of workers entering a SEIN, again in absolute and relative terms. For the QWI, only the strongest links are used to filter out spurious employer identifier changes. If 80 percent of a SEIN’s workers (the predecessor) are observed to move to a single successor, and that successor absorbs 80 percent of its employees from a single predecessor, then all flows between those two account numbers are filtered out and treated as if they had never existed. This is accomplished by coding in the QWI processing, not by changing any of the information in the infrastructure files.17 Of importance to the unit-to-worker imputation (described in section 5.4.2) is a similar measure, computed within a SEIN. For most states, and employers within states, the breakout of units into SEINUNITs is at the discretion of the employer, and the employer may decide to change such a 17. A more extensive evaluation of the impact of the SPF on the aggregate QWI statistics is currently under way.

174

J. Abowd et al.

breakout. The SPF, by following groups of workers as they move between SEINUNITs, identifies spurious intra-SEIN flows, which are then ignored when doing the unit-to-worker imputation for multi-unit job histories. 5.4.2 Allocating Workers to Workplaces: Unit-to-Worker Imputation (U2W) Early versions of the QWI (then called the Employment Dynamics Estimates [EDE]), were computed only at the SEIN level, with employment allocated to a single location per SEIN. This approach was driven by the absence of workplace information on almost all state-provided wage records. Only the state of Minnesota requires the identification of a worker’s workplace (SEINUNIT) on its UI wage records. A primary objective of the QWI is to provide employment, job and worker flows, and wage measures at a very detailed level of geography (place-of-work) and industry. The structure of the administrative data received by LEHD from state partners, however, poses a challenge to achieving this goal. The QWI measures are primarily based on the processing of UI wage records that report, with the exception of Minnesota, only the legal employer (SEIN) of the workers. The ES-202 micro-data, however, are comprised of establishment-level records which provide the geographic and industry detail needed to produce the QWI. For employers operating only one establishment within a state, the assignment of establishmentlevel characteristics to UI wage records is straightforward because there is no distinction between the employer and the establishment. However, approximately 30 to 40 percent of state-level employment is concentrated in employers that operate more than one establishment in that state. For these multi-unit employers, the SEIN on workers’ wage records identifies the legal employer in the ES-202 data, but not the employing establishment (place-of-work). Thus, establishment level characteristics—geography and industry, in particular—are missing data for these multi-unit job histories. In order to impute establishment-level characteristics to job histories of multi-unit employers, a nonignorable missing data model with multiple imputation was developed. The model imputes establishment-of-employment using two key characteristics available in the LEHD Infrastructure Files: (a) distance between place-of-work and place-of-residence and (b) the distribution of employment across establishments of multi-unit employers. The distance to work model is estimated using data from Minnesota, where both the SEIN and SEINUNIT identifiers appear on a UI wage record. Then, the posterior distribution of the parameters from this estimation, combined with the actual SEIN and SEINUNIT employment histories from the ES-202 data, are used for multiple imputation of the SEINUNIT associated with workers in a given SEIN in the data from states other than

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

175

Minnesota.18 Emerging from this process is an output file, called the Unitto-Worker (U2W) file, containing ten imputed establishments for each worker of a multi-unit employer. These implicates are then used in the downstream processing of the QWI. The U2W process relies on information from each of the four Infrastructure Files—ECF, GAL, EHF, and ICF—as well as the auxiliary SPF file. Within the ECF, the universe of multi-unit employers is identified. For these employers, the ECF also provides establishment-level employment, date-of-birth, and geocodes (which are acquired from the GAL). The SPF contains information on predecessor relationships, which may lead to the revision of date-of-birth implied by the ECF. Finally, job histories in the EHF in conjunction with place-of-residence information stored in the ICF provide the necessary worker information needed to estimate and apply the imputation model. A Probability Model for Employment Location Definitions. Let i 1, . . . , I index workers, j 1, . . . , J index employers (SEINs), and t 1, . . . , T index time (quarters). Let Rjt denote the number of active establishments at employer j in quarter t, let maxj,t Rjt, and r 1, . . . , index establishments. Note that the index r is nested within j. Let Njrt denote the quarter t employment of establishment r in employer j. Finally, if worker i was employed at employer j in t, denote by yijt the establishment at which the worker was employed. Let t denote the set of employers active in quarter t, let jt denote the set of individuals employed at employer j in quarter t, let jt denote the set of active (Njrt 0) establishments at employer j in t, and let ijt ⊂ jt denote the set of active establishments that are feasible for worker i. Feasibility is defined as follows: an establishment r ∈ ijt if Njrs 0 for every quarter s that i was employed at j. The probability model. Let pijrt Pr( yijt r). At the core of the model is the probability statement: e x (6) pijrt i e x jrt

∑

ijrt

jst

ijst

s∈jt

where jrt is a establishment- and quarter-specific eﬀect, xijrt is a time-varying vector of characteristics of the worker and establishment, and measures the eﬀect of characteristics on the probability of being employed at a partic18. The actual SEINUNIT coded on the UI wage records is used for Minnesota, and would be used for any other state that provided such data. Note that there are occasional, and rare, discrepancies between the unit structure on the Minnesota wage records and the unit structure on the Minnesota ES-202 data for the same quarter. These discrepancies are resolved during the initial processing of the Minnesota data in its state-specific read-in procedures.

176

J. Abowd et al.

ular establishment. In the current implementation, xijrt is a linear spline in the (great-circle) distance between worker i’s residence and the physical location of establishment r. The spline has knots at 25, 50, and 100 miles. Using equation (6), the following likelihood is defined T

(7)

p( y| , , x) ∏ ∏ ∏ ∏ ( pijrt)d

ijrt

t1 j∈t i∈jt r∈ijt

where dijrt

(8)

1 if yijt r 0 0 otherwise

and where y is the appropriately-dimensioned vector of the outcome variables yijt, is the appropriately dimensioned vector of the jrt, and x is the appropriately-dimensioned matrix of characteristics xijrt. For jrt, a hierarchical Bayesian model based on employment counts Njrt is specified. The object of interest is the joint posterior distribution of and . A uniform prior on , p () ∝ 1 is assumed. The characterization of p( , |x, y, N) is based on the factorization (9)

p( , |x, y, N) p( |N)p(| , x, y) ∝ p( |N)p()p(y| , , x) ∝ p( |N)p(y| , , x).

Thus, the joint posterior (9) is completely characterized by the posterior of and the likelihood of y in (7). Note (7) and (9) assume that the employment counts N aﬀect employment location y only through the parameters . Estimation. The joint posterior p( , |x, y, N) is approximated at the posterior mode. In particular, we estimate the posterior mode of p(| , x, y) evaluated at the posterior mode of . From these we compute the posterior modal values of the jrt, then, maximize the log posterior density T

(10) log p(| , x, y) ∝ ∑ ∑ ∑ ∑ dijrt jrt xijrt log t1 j∈t i∈jt r∈jti

∑e

jst xijst

s∈ijt

which is evaluated at the posterior modal values of the jrt, using a modified Newton-Raphson method. The mode-finding exercise is based on the gradient and Hessian of (10). In practice, (10) is estimated for three employer-employment size classes: 1 to 100 employees, 101 to 500 employees, and greater than 500 employees, using data for Minnesota. Imputing Place of Work After estimating the probability model using Minnesota data, the posterior distribution of the estimated parameters is combined with the entityspecific posterior distribution of the parameters in the imputation pro-

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

177

cess for other states. A brief outline of the imputation method, as it relates to the probability model previously discussed, is provided in this section. Emphasis is placed on not only the imputation process itself, but also the preparation of input data. Sketch of the imputation method. Ignoring temporal considerations, 10 implicates are generated as follows. First, using the posterior mean and variance of estimated from the Minnesota data, we take 10 draws of from the normal approximation (at the mode) to p(| , x, y). Next, using ES-202 employment counts for the establishments, we compute 10 values of jt based on the hierarchical model for these parameters. Note that these are draws from the exact posterior distribution of the jrt. The drawn values of and are used to draw 10 imputed values of place of work from the asymptotic approximation to the posterior predictive distribution (11)

p( y˜ |x, y) ∫ ∫ p( y˜ | , , x, y)p( |N)p(| , x, y) d d.

Implementation Establishment data. Using state-level micro-data, the set of employers (SEINs) that ever operate more that one establishment in a given quarter is identified; these SEINs represent the set of ever-multi-unit employers defined above as the set t. For each of these employers, its establishmentlevel records are identified. For each establishment, latitude and longitude coordinates, parent employer (SEIN) employment, and ES-202 monthone employment19 for the entire history of the establishment are retained. Those establishments with positive month-one employment in a given quarter characterize jt, the set of all active establishments. An establishment birth date is identified and, in most cases, is the first quarter in the ES-202 time series in which the establishment has positive month-one employment. For some employers, predecessor relationships are identified in the SPF; in those instances, the establishment date-of-birth is adjusted to coincide with that of the predecessor’s. Worker data. The EHF provides the earnings histories for employees of the ever-multi-unit employers. For each in-scope job (a worker-employer pair), one observation is generated for the end of each job spell, where a job spell is defined as a continuum of quarters of positive earnings for a worker at a particular employer during which there are no more than three consecutive periods of nonpositive earnings.20 The start date of the job history 19. In rare instances where no ES-202 employment is available, an alternative employment measure based on UI wage record counts may be used. 20. A new hire is defined in the QWI as a worker who accedes to a firm in the current period but was not employed by the same firm in any of the 4 previous periods. A new job spell is created if, for example, a worker leaves a firm for more than 4 quarters and is subsequently reemployed by the same firm.

178

J. Abowd et al.

is identified as the first quarter of positive earnings; the end date is the last date of positive earnings.21 These job spells characterize the set jt. Candidates. Once the universe of establishments and workers is identified, data are combined and a priori restrictions and feasibility assumptions are imposed. For each quarter of the date series, the history of every job spell that ends in that quarter is compared to the history of every active (in terms of ES-202 first month employment) establishment of the employing employer (SEIN). The start date of the job spell is compared to the birth date of each establishment. Establishments that were born after the start of a job spell are immediately discarded from the set of candidate establishments. The remaining establishments constitute the set ijt ⊂ jt for a job spell (worker) at a given employer.22 Given the structure of the pairing of job spells with candidate establishments, it is clear that within job spell changes of establishment are ruled out. An establishment is imputed once for each job spell,23 thereby creating no spurious labor market transitions. Imputation and output data. Once the input data are organized, a set of 10 imputed establishment identifiers are generated for each job spell ending in every quarter for which both ES-202 and UI wage records exist. For each quarter, implicate, and size class, s 1, 2, 3, the parameters on the linear spline in distance between place-of-work and place-of-residence ˆ s are sampled from the normal approximation of the posterior predictive distribution of s conditional on Minnesota (MN) (12)

p(s| MN, xMN, yMN).

The draws from this distribution vary across implicates, but not across time, employers, and individuals. Next, for each employer j at time t, a set of ˆ jrt are drawn from (13)

p( ST |NST)

which are based on the ES-202 month-one employment totals (Njrt) for all candidate establishments rjt ⊂ Rjt at employer j within the state (ST ) being processed. The initial draws of ˆ jrt from this distribution vary across time and employers but not across job spells. Combining (12) and (13) yields 21. By definition, an end-date for a job spell is not assigned in cases where a quarter of positive earnings at a firm is succeeded by 4 or fewer quarters of nonemployment and subsequent reemployment by the same firm. 22. The sample of UI wage and QCEW data chosen for processing of the QWI is such that the start and end dates are the same. Birth and death dates of establishments are, more precisely, the dates associated with the beginning and ending of employment activity observed in the data. The same is true for the dates assigned to the job spells. 23. More specifically, an establishment is imputed to a job spell only once within each implicate.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

(14)

179

p( ST|NST)p(s| MN, xMN, yMN) ≈ p( ST|NST)p(s| ST, xST, yST) p( ST, ST|xST, yST, NST)

an approximation of the joint posterior distribution of and s (9) conditional on data from the state being processed. The draws ˆ s and ˆ jrt in conjunction with the establishment, employer, and job spell data are used to construct the pijrt in (1) for all candidate establishments r ∈ ijt. For each job spell and candidate establishment combination, the ˆ s are applied to the calculated distance between place-ofresidence (of the worker holding the job spell) and the location of the establishment, where the choice of ˆ s depends on the size class of the establishment’s parent employer. For each combination an ˆ jrt is drawn, which is based primarily on the size (in terms of employment) of the establishment relative to other active establishments at the parent employer. In conjunction, these determine the conditional probability pijrt of a candidate establishment’s assignment to a given job spell. Finally, from this distribution of probabilities is drawn an establishment of employment. The imputation process yields a data file containing a set of 10 imputed establishment identifiers for each job spell. In a very small set of cases, the model fails to impute an establishment to a job spell. This is often due to unanticipated idiosyncrasies in the underlying administrative data. Furthermore, across states, the proportion of these failures relative to successful imputation is well under 0.5 percent. For these job spells, a dummy establishment identifier is assigned and in downstream processing, the employment-weighted modal employer-level characteristics are used. 5.5 Forming Aggregated Estimates: QWI 5.5.1 What are the QWI? The Quarterly Workforce Indicators (QWI) provide detailed local estimates of a variety of employment and earnings indicators. Employment, earnings, gross job creation and destruction, and worker turnover are available at diﬀerent levels of geography, including the county, Workforce Investment Area, and Core Based Statistical Area.24 At each level of geography, the QWI are available by detailed industry (SIC and NAICS), sex, and age of workers. As of January 2008, QWI for forty-three states had been published, three additional states were in prerelease analysis, and a total of forty-six states, including the District of Columbia, had signed 24. The original QWI release used Metropolitan Statistical Areas. The older MSA definitions were replaced with CBSA definitions in 2005.

180

J. Abowd et al.

Memorandums of Understanding (MOUs). The program was still expanding with the goal of national coverage. 5.5.2 Computing the Estimates The establishment of the LEHD Infrastructure Files was driven in large part, although not exclusively, by the needs of the QWI computations. Completed and representative job-level data, with worker and workplace characteristics, are the primary input for the QWI. The ICF (section 5.3.2) and the ECF (section 5.3.4) draw on a large number of data sources, and use a set of editing and imputation procedures described previously, to provide a detailed picture of each economic actor. The ECF also provides the input data for the weighting, which is explained in more detail in section 5.5.3. The wage record edit (section 5.2.6) and the SPF (section 5.4.1) apply longitudinal edits and probabilistic matching rules to the improve the longitudinal linking of entities. The U2W (section 5.4.2) completes the picture, by multiply imputing an employing establishment to each job reported by the multi-unit employers. Figure 5.1 provides a graphical overview of how these data sources are used in QWI processing. These data are then combined and aggregated to compute the QWI statistics. The aggregation is a four-step process: 1. A job—a unique PIK-SEIN-SEINUNIT combination—is identified, and the job’s complete activity history (when the worker had positive earnings at the SEIN-SEINUNIT, and when the worker did not have positive earnings) was recorded. Note that for job history associated with multi-unit SEINs, there are 10 implicate SEINUNITs (possibly nonunique) for each job, and these implicates each get a weight of 0.1 in the downstream processing.25 2. Job-level variables are computed as a set of indicators. The computation of each of these variables is described in detail in section 2.2 of the appendix. 3. Job-level variables are aggregated to the establishment level (SEINUNIT), using appropriate implicate weights. The aggregation is done using formulae described in section 2.3 of the appendix. For many variables, aggregation to the establishment-level is achieved by summing the job-level variables (beginning-of-period employment, end-of-period employment, accessions, new hires, recalls, separations, full-quarter employment, full-quarter accessions, full-quarter new hires, total earnings of full-quarter employees, total earnings of full-quarter accessions, and total earnings of full-quarter new hires). Some aggregate flow variables are computed using the beginning- and end-of-quarter employment estimates for 25. In the underlying frame, a job is a PIK-SEIN pair. For single-unit employers, this is equivalent to a PIK-SEIN-SEINUNIT triple. For multi-unit employers within a single state, the original pair is completed to a triple by the unit-to-worker multiple imputation.

Fig. 5.1

Overview of LEHD data flow

182

J. Abowd et al.

that workplace. Examples are net job flows (see equation (A43) in appendix section 2), average employment (A44), job creations (A46) and job destructions (A48). The file created in this step, internally known as the Unit Flow File (UFF_B), is also available in the RDC system (see section 5.8.2 for details). 4. The variables necessary for applying the QWI disclosure avoidance algorithm—SEINUNIT-specific noise infusion called “fuzz factors”—are attached, and the establishment-level file is summed to the desired level of geographic and demographic detail, using the noise-infused values. Some flow variables are computed directly from other aggregated variables (see appendix section 2.5). An undistorted version of all aggregates is also created. All aggregations use weights (see section 5.5.3). 5. The tables created in the previous step are processed by the disclosure avoidance procedure (see section 5.6), using a comparison with the undistorted version of each indicator and appropriate cell counts. If necessary, items in some cells are suppressed, and noisy estimates are flagged as such. 5.5.3 Weighting in the QWI The QWI are estimates formed from weighted sums where the weights have been controlled to state-level QCEW statistics for all private employers as published by the BLS. The control is approximate, however, because the weights are calculated from the unfuzzed beginning-of-quarter employment data whereas the publication estimates are based on the weighted sums of the noise-infused data. When building the ECF, weights are computed such that the measured beginning-of-quarter UI employment of in-scope units, when properly weighted, is equal to the published QCEW statewide employment in the first month of the quarter for all private employers. A preliminary weight is computed as part of the ECF processing. An adjustment factor that accounts for system-wide missing data imputation and other edits, is computed in the downstream (UFF_B) processing. This adjustment factor is computed for all private establishments. The final weight is computed in the UFF_B processing to control the product of the initial weight and the adjustment factor to the state total for all private employment in that quarter’s QCEW data. The same overall adjustment factor that was calculated for all private establishments is used to produce the final weights for all the establishments QWI estimates. Selection, editing, longitudinal linking, and disclosure avoidance procedures in the micro data used to build the QWI all change the in-scope units’ data somewhat, causing the preliminary and final weights to disagree. When the final weight is used for all published QWI statistics, the diﬀerence between the published QCEW statistic and the appropriate statistic in the QWI system is less than 0.5 percent.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

183

5.6 Disclosure Avoidance Procedures for the QWI The disclosure avoidance procedures for the QWI consist of a set of methods used to protect the confidentiality of the identity and attributes of the individuals and businesses that form the underlying data in the system. In the QWI system, disclosure avoidance is required to protect the information about individuals and businesses that contribute to the UI wage records, the ES-202 quarterly reports, and the Census Bureau demographic data that have been integrated with these sources. The QWI disclosure avoidance mechanism is described and analyzed in more detail in Abowd et al. (2006); we present an overview here. 5.6.1 Three Layers of Confidentiality Protection There are three layers of confidentiality protection and disclosure avoidance protections in the QWI system. The first layer occurs when job-level estimates (computed from the EHF) are aggregated to the establishment level. The QWI system infuses specially constructed noise into the estimates of all of the workplace-level measures. We will describe the noiseinfusion process in more detail in section 5.6.2. After this noise infusion, the distorted micro data item is used as the source for all published QWIs. A second layer of confidentiality protection occurs when the workplacelevel measures are aggregated to higher levels (e.g., substate geography and industry detail). The data from many individuals and establishments are combined into a (relatively) few estimates using a dynamic weight that controls the state-level beginning of quarter employment for all private employers to match the first month in quarter employment as tabulated from the QCEW. The weighting procedure introduces an additional diﬀerence between the confidential data item and the released data item, and in combination with the noise infusion, the published data are moved away from the value contained in the underlying micro data, contributing to the protection of the confidentiality of the micro data. Third, some of the aggregate estimates turn out to be based on fewer than three persons or establishments. These estimates are suppressed and a flag set to indicate suppression. Suppression is only used when the combination of noise infusion and weighting may not distort the publication data with a high enough probability to meet the criteria laid out above. Estimates such as employment are subject to suppression. Continuous dollar measures like payroll are not. All published estimates are influenced by the noise that was infused in the first layer of the protection system. When the distortion exceeds certain limits, the estimates are still published, but flagged as substantially distorted. Each observation on any one of the published QWI tables thus has an associated flag that describes its disclosure status. Table 5.5 lists all possible flags in the published QWI tables.

184

J. Abowd et al.

Table 5.5

Disclosure flags in the QWI

Flag –2 –1 0 1 5 9

Explanation No data available in this category for this quarter Data not available to compute this estimate Zero employment estimated or zero estimated denominator in a ratio, zero released OK, distorted value released Value suppressed because it does not meet U.S. Census Bureau publication standards Data significantly distorted, distorted value released

5.6.2 Details of the QWI Noise Infusion Process The noise infused into the QWI data is designed to have three very important properties. First, every data item is distorted by some minimum amount. Second, for a given workplace, the data are always distorted in the same direction (increased or decreased) by the same percentage amount in every period, and in every revision of the QWI series. Third, the statistical properties of this distortion are such that when the estimates are aggregated, the eﬀects of the distortion cancel out for the vast majority of the estimates, preserving both cross-sectional and time series analytical validity. We describe below the algorithms by which the above goals are achieved. A statistical analysis providing evidence of the third goal is provided in section 5.7.2. Disclosure Avoidance Using Noise Infusion Factors To implement the multiplicative noise model in section 5.6, a random fuzz factor j is drawn for each establishment j according to the following process: (b )/(b a)2, ∈ [a, b]

p(j) (b 2)/(b a)2, ∈ [2 b, 2 a]

0, otherwise

0, 2 b

[( b 2)2]/[2(b a)2], ∈ [2 b, 2 a]

F(j)

0.5, ∈ (2 a, a) 0.5 [(b a)2 (b )2]/[2(b a)2], ∈ [a, b]

1, b

where a 1 c/100 and b 1 d/100 are constants chosen such that the true value is distorted by a minimum of c percent and a maximum of d percent (the exact numbers are confidential). Note that 1 a b 2. This produces a random noise factor centered around 1 with distortion of at

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

Fig. 5.2

185

Distribution of fuzz factors

least c and at most d percent. Figure 5.2 depicts such a distribution. A fuzz factor is drawn for each employer and for each of the establishments associated with that employer. Although fuzz factors vary across establishments of the same employer, the fuzz factors attached to all establishments of the same employer are drawn from the same (upper or lower) tail of the fuzz factor distribution. Thus, if the fuzz factor associated with a particular employer (SEIN) is less than unity, then all that employer’s establishments (SEINUNITs) will also have fuzz factors less than unity. It is also important to point out that a fuzz factor is attached to each SEIN and SEINUNIT only once and retained for all time periods after the initial assignment. Applying the Fuzz Factors to Estimates Although all estimates are distorted based on the multiplicative noise model, the exact implementation depends on the type of estimate that is computed. For completeness we show all the relevant formulas here, referring the reader to Abowd, Stephens, and Vilhuber (2006) for details. In all cases, the micro data noise infusion occurs at the level of an establishment estimate. However, for QWI involving ratios and changes, the basic fuzzed and unfuzzed values are combined at the publication level of aggregation to produce the released estimates. In what follows, distorted values are distinguished from their undistorted counterparts by an asterisk, that is, the

186

J. Abowd et al.

true (unfuzzed) value of beginning-of-quarter employment is B, its noiseinfused (fuzzed) counterpart is B∗. Fuzzing of estimates of employment. The fuzz factor j is used to fuzz all estimates of employment totals by scaling of the true establishment level statistic according to the formula: (15)

X∗jt j Xjt

where Xjt is an establishment level employment estimate: B, E, M, F, A, S, H, R, FA, FS, and FH. All variable definitions are provided in section 2 of the appendix. Fuzzing of averages of magnitude estimates where the denominator is an employment estimate. Ratios of magnitude estimates to employment estimates are protected by using fuzzed numerators and unfuzzed denominators according the formula: Y∗jt Yjt ZY∗jt j B(Y)jt B(Y)jt where ZYjt is a ratio of a magnitude estimate, Yjt, (dollars or quarters) and B(Yjt) is an estimate of employment. The ratio has the interpretation of an average in most cases. The variables protected according to this method are: ZW2, ZW3, ZWFH, ZWA, ZWS, ZNA, ZNH, ZNR, and ZNS. The relevant values of Yjt and B(Yjt) are shown in the establishment level statistics in the previous equation. In the actual QWI processing, the numerator and denominator of these confidentiality-protected ratios are tabulated separately for each publication category (ownership state substategeography industry age group sex). Then, the publication ratio is computed when the public-use release files are created. Fuzzing of diﬀerences of counts and magnitudes. Fuzzed net job flow (JF ) is computed at the aggregate level for k (ownership state substategeography industry age group sex) cell as the product of the aggregated, unfuzzed rate of growth of net jobs and the aggregated fuzzed employment: E ∗kt J F ∗kt Gkt E ∗kt J Fkt . E kt This method of fuzzing net job flow will consistently estimate net job flow because it takes the product of two consistent estimators. The formulas for fuzzing gross job creation (JC ) and job destruction (JD) are similar: ∗kt E JC∗kt JCRkt E ∗kt JCkt E kt

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

187

and E ∗kt JD∗kt JDRkt E ∗kt JDkt . E kt The same method was used to protect estimates of wage changes for diﬀerent employment estimates. The unfuzzed estimated total changes were divided by the unfuzzed denominators then multiplied by the ratio of the fuzzed denominator to the unfuzzed denominator, as in the formula: Y∗kt WYkt ZWY ∗kt Ykt Ykt where, again, Y denotes a particular employment, WY denotes the estimated change in wages for that employment estimate, and ZWY∗ is the confidentiality-protected estimate of the ratio. This method is used for ZWA, ZWS, ZWFA, and ZWFS. The ratio FT involves three QWI that are also in the release file. In order to protect the ratio of the fuzzed to unfuzzed estimate of full-quarter employment, the release value of FT is protected by the formula: (FA∗kt FS∗kt) /2 F ∗kt FT ∗kt Fkt Fkt In the actual QWI processing the numerator and denominator of these confidentiality-protected changes and ratios are tabulated separately for each publication category (ownership state substate-geography industry age group sex). Then, the publication change or ratio is computed when the public-use release files are created. 5.7 Analysis of the QWI Files In this section, we will provide some basic analysis highlighting the usefulness of the QWI as time series data on local labor market conditions and measuring the impact of the various corrections that are applied to the series. 5.7.1 Basic Trends of Some Variables The QWI are uniquely positioned to provide a picture of a dynamic workforce at a highly disaggregated level with both demographic and economic detail. In this section, we consider three variables, and provide examples of analyses that can be easily produced with the QWI. We consider employment (more precisely, begininning-of-quarter employment), job creation, and recalls. We have picked the states of Illinois and Montana to illustrate the analyses. Figures 5.3 and 5.4 show the basic data trends for the three variables, stated in thousands of workers, for both sexes combined and separately. In general, all three time series show considerable seasonality, but job creations

Fig. 5.3

Basic data trends, Illinois

Fig. 5.4

Basic data trends, Montana

190

J. Abowd et al.

and recalls are considerably more variable. However, when looking at the time series by sex, there appears to be less volatility in job creations and recalls for women than for men. Figures 5.5 and 5.6 restate these series as the percentage of women in the total for each variable. In Illinois, the percentage of jobs created that are filled by recalls is significantly lower for women (46.2 percent) than it is for men (53 percent), and persistently so over time (fig. 5.3), although there is strong seasonality in this pattern as well. A similar pattern, although not quite as stable, emerges in Montana (fig. 5.3). Thus, it would seem that although women participate as much as men in job creation, they are more likely to have found a new job than to have been recalled to an old job. Of course, this is a very simple analysis. A further breakdown by industry (also feasible using the public-use QWI) might reveal that the lower recall rate of women is a phenomenon specific to certain industries that employ a higher fraction of women for other reasons. However, it is an example that serves to highlight the utility of the demographic, geographic, and industry breakdown that is possible with the QWI. Disaggregating statistics by geography is one of the more common strategies for policy analysts, and several data sources are available to perform such an analysis. With the QWI, geographic analysis can be extended to distinct demographic groups. In figure 5.7, the geographic distribution of job creation is plotted for young workers nineteen to twenty-one years of age, by counties in Illinois. A policy analyst could perform such an analysis for eight age groups and both sexes, using the complete QWI. Note that net job creation is computed with both the numerator and the denominator computed for workers aged nineteen to twenty-one, so it is not simply

Fig. 5.5

Proportion of women, select variables, Illinois

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

Fig. 5.6

191

Proportion of women, select variables, Montana

a decomposition of an aggregate job creation statistics; it is computed from the ground up using data only on workers between nineteen and twentyone years of age. 5.7.2 Importance of LEHD Adjustments to Raw Data There are numerous data edits, corrections, and imputations performed in the processing of the QWI data series. We summarize the eﬀect of some of these adjustments here. Choosing Between LDB and LEHD Coding of NAICS Variables As noted in section 5.3.4, the ECF provides enhanced NAICS variables that expand on the information available on the ES-202 files. Information is imputed based on all available industry information, and backcoded to time periods that precede the introduction and widespread implementation of NAICS coding on ES-202 data. The creation of the enhanced NAICS variables was described in section 5.3.4. In this section, we present a summary of research done on a comparison of the ESO (ES-202 only) and FNL (final) NAICS codes on the Illinois ECF. The imputation algorithm used by the BLS to create the LDB stably backfills NAICS codes once it has imputed a code for a later year; that is, once an establishment has received a backcoded NAICS, that code is used for all prior years of data for the establishment. The LEHD algorithm allows the backcoded NAICS to change if the contemporaneously coded SIC changes. Thus, we expect the two backcodes to have diﬀerent statistical properties for historical NAICS-based QWI. Although some of the SIC changes over time may be spurious, a SEINUNIT’s SIC code history

Fig. 5.7

Job creation for young workers, by county, Illinois

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

193

contains valuable information that we have attempted to preserve in the LEHD imputation algorithm. Overall, the eﬀect of the diﬀerent approaches is relatively small, since very few SEINUNITs change industry, in particular relative to the proportion of SEINUNITs that change geography. The LDB-sourced NAICS variable is used for about 85 percent of the records for Illinois; the rest are filled with information from the ES-202. It is unclear why only 85 percent of ES-202 records are in the LDB. The results weighted by employment are about the same, suggesting that activity was not a criterion for being included on the LDB. First and not surprisingly, in later years and quarters (1999 ) when NAICS is actively coded by the states, the ESO and FNL codes look almost identical when available. Second, there is little variation in the LDB NAICS codes over time compared with SIC. Among all of the active SEINSEINUNITs over the period covered by the Illinois data, only slightly more than 8 percent experience at least one SIC change, compared with about 1.5 percent on the LDB. Almost all NAICS code changes occur after 1999. While this is not entirely unexpected, it is something to keep in mind when comparing NAICS FNL versus SIC or NAICS ESO employment totals. Many of these changes in industry appear to be real and are not captured on the LDB. As we go back in time, a larger portion of employment can be found in NAICS FNL codes that are diﬀerent from what one would expect given the SIC code on the ECF. For example, in 1990 about 13 percent of employment is in a NAICS FNL code that is diﬀerent from what we would expect based on the SIC. By 2001, the proportion of employment that is in a NAICS code outside of the set of possible values predicted by the SICNAICS crosswalk falls to 3 percent. The ES-202 based NAICS variable does a better job tracking SIC, since more SIC information is used in putting it together. The main source of the discrepancy is due to entities that experience a change in their SIC code prior to 2000. The LDB appears to ignore this change, while the ES-202 based NAICS variable uses an SIC-based imputation for these SEINUNITs. The result is a series that exhibits similar patterns of change over time as SIC, while still preserving the value added in the NAICS codes for entities that did not experience a change. Users should keep in mind that for early years (before 1997) some of the NAICS industries have yet to come into existence. The prevalence of this problem has not yet been investigated. Correcting for Coding Errors in Personal Identifiers Abowd and Vilhuber (2005) describe and analyze the method used at LEHD to identify coding errors in the person identifier (Social Security Number [SSN]), and provide an analysis of the impact that correcting for

194

J. Abowd et al.

such errors has on statistics generated from the corrected and the uncorrected data for one state (California). A simplified version of the same analysis is used as a quality assurance method during the wage record edit, and the results are similar for other states, but vary with length of available data and with prior state processing of name and SSN fields. For California, the process verified over half a billion records. Slightly less than 10 percent of the total number of unique individuals appear in the original data, and only a little more than 0.5 percent of all wage records require some corrective measures, which is considered conservative relative to other analyses done (see Abowd and Vilhuber [2005] for further references). Table 5.6 presents patterns of job histories for uncorrected and corrected data. The unit of observation is a worker-employer match (a job), potentially interrupted. For each such observation, the longest interruption is tabulated if there is one. If no interruption was observed during the worker’s tenure with the employer, then the type of continuous job spell is tabulated. By definition, the absence of a hole implies continuous tenure, but that spell may have been ongoing in the first (left-truncated) or last (right-truncated) quarter of the data, or in both (entire period). If the spell was continuous, with both the beginning and the end of the job spell observed within the data, then the default code of C is assigned.

Table 5.6

Wage record edit: Comparing job histories before and after editing process Original data

Pattern in job history

1 quarter 2 quarters 3 quarters 4 quarters 5 quarters 6 quarters 7 quarters 8 quarters 9 or more quarters C Continuous F Entire period L Left-truncated R Right-truncated

Frequency

Edited data

Percent (%)

Frequency

Percent (%)

Noncontinuous, length of longest interruption 5,315,869 5.50 4,710,673 4.87 2,357,942 2.44 2,359,374 2.44 1,764,701 1.83 1,755,814 1.82 750,910 0.78 747,707 0.77 532,174 0.55 529,777 0.55 466,301 0.48 463,878 0.48 430,549 0.45 429,179 0.44 241,573 0.25 240,214 0.25 1,172,039 1.21 1,163,420 1.20

Change Frequency

Percent (%)

–605,196 1,432 –8,887 –3,203 –2,397 –2,423 –1,370 –1,359 –8,619

–11.38 0.06 –0.50 –0.42 –0.45 –0.51 –0.31 –0.56 –0.73

59,990,419 1,735,340 9,871,084 12,001,245

62.08 1.80 10.22 12.42

Continuous 60,311,626 1,807,775 10,032,149 12,144,959

62.37 1.87 10.37 12.56

321,207 72,435 161,065 143,714

0.53 4.17 1.63 1.19

96,630,146

100.00

96,696,545

100.00

66,399

0.06

Notes: From table 6, Abowd and Vilhuber (2005). For definitions of job history patterns, see text.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

195

Over 800,000 job history interruptions in the original data are eliminated by the corrections, representing 0.9 percent of all jobs, but 11 percent of all interrupted jobs (Table 5.6). Despite the small number of records that are found to be miscoded, the impact on flow statistics can be large. Accessions in the uncorrected data are overestimated by 2 percent, and recalls are biased upwards by nearly 6 percent. On the other hand, and as expected, overall payroll W1 are not biased, but payroll for accessions (WA) and separations (WS) are biased upward by up to 7 percent (Table 5.7). Identification of Successor-Predecessor Links of Firms and Establishments As noted in section 5.4.1, care is taken when tracking firms and establishments over time by tracking worker movements between firms. These corrections should have little or no impact on the time series of pure stock

Table 5.7

Distribution of percentage bias in aggregate QWI statistics All age groups, both sexes, SEIN-level micro data

Variable (bias) A

B

F

R

S

W1

WA

WS

Unit

Mean (%)

Std (%)

N

Firm County Industry Firm County Industry Firm County Industry Firm County Industry Firm County Industry Firm County Industry Firm County Industry Firm County Industry

2.17 1.56 1.97 –0.74 –0.46 –0.31 –1.23 –0.78 –0.53 4.71 5.26 5.95 2.31 1.66 2.01 –0.01 –0.01 0.04 15.57 4.92 3.95 18.77 4.87 3.64

13.98 1.01 2.29 6.14 0.31 0.31 8.05 0.36 0.31 26.86 3.61 3.49 14.29 1.11 2.08 4.96 0.15 0.35 1111.78 3.34 4.94 1094.50 3.17 4.48

11,755,355 2,006 374 20,717,508 1,947 363 18,454,708 1,888 352 3,242,186 1,888 352 11,161,916 1,947 363 23,229,843 2,006 374 11,755,355 2,006 374 11,161,916 1,947 363

P10 (%)

P50 (%)

P90 (%)

0.62 0.51

1.42 1.47

2.64 3.40

–0.75 –0.59

–0.45 –0.34

–0.25 –0.14

–1.21 –0.90

–0.74 –0.53

–0.43 –0.24

1.70 1.93

4.59 5.46

9.18 10.29

0.67 0.63

1.46 1.53

2.72 3.41

–0.05 –0.04

–0.02 –0.02

0.00 0.08

1.89 0.77

4.38 3.35

8.44 6.79

2.02 1.00

4.31 3.18

8.06 5.71

Note: From table 9, Abowd and Vilhuber (2005). There are 23,232,068 firm-quarter cells, 2006 county-quarter cells, and 374 industry-quarter cells. Percentiles for firm-quarter cells are all zero and not reported for simplification.

196

J. Abowd et al.

measures (total wage bill W1), but should influence a number of flow measures. In particular, separations (S) and accessions (A) will be reduced when between-firm (successor-predecessor) links are identified. A small experiment was run using the standard processing stream for the QWI for a single state. Transitions associated with observed successorpredecessor flows as identified by the SPF, which are normally suppressed, were left intact. In other words, the SPF was removed from the processing stream. Comparing the resultant (unreleased) QWI with published QWI from the same time period provides an estimate of the bias due to firm links that unadjusted QWI would otherwise have. The suppression of flows due to successor-predecessor links also aﬀects B, beginning-of-quarter employment, which in turn is used to weight the QWI (section 5.5.3). Thus, all statistics will be aﬀected, either directly through the statistic itself, or indirectly through a change in the weights. Analysis performed on Montana reveals that earnings and separations are 4 percent lower if successor-predecessor transitions are filtered out. Beginning-of-period employment estimates are 0.4 percent lower. For more results, consult Benedetto et al. (2007), who have used the successor-predecessor flows in the analysis of the firm. Analytical Validity of the Unit-to-Worker Imputation This subsection presents some results of the assessment of the analytical validity of the unit-to-worker imputation process (section 5.4.2). For five QWI measures—beginning-of-quarter employment (B), full-quarter employment (F ), accessions (A), separations (S), and total payroll (W1)—percentiles of the distribution of the bias induced by the imputation process for two levels of industry aggregation are presented. A complete evaluation of the validity of the unit-to-worker imputation process is provided in Stephens (2006, chapter on Imputation of Place-of-Work in the Quarterly Workforce Indicators). To assess the analytic validity of the imputation process, two sets of QWI measures for the 1994:1 to 2003:4 time period were generated using the Minnesota data. The first set, True, is produced using the establishment of work reported on the Minnesota UI wage records. The second set, Imputed, is generated treating the establishment of work as unknown; thus, Imputed is generated using the same imputation process that is applied to other states in the QWI system. Measures for both sets were tabulated using data for all establishments in Minnesota and produced for two levels of industry aggregation—SIC Division and two-digit SIC—as well as by county, sex, and age. For each measure, the discrepancies between values X, prior to the application of multiplicative noise factors, contained in Imputed and True for each interior quarter industry county age sex cell are calculated as:

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

197

XImputed XTrue bias . XTrue Table 5.8 presents percentiles of both the weighted and unweighted distribution of the bias statistic for five QWI measures across interior data cells for SIC Division and two-digit SIC industry aggregations. For all distributions, the median discrepancy never exceeds 0.005 in absolute value, suggesting that on average the bias induced by the imputation of place of work is relatively insignificant. For all bias statistics, the unweighted distribution is tighter than the weighted distribution, illustrating that the bias is less severe in data cells with higher levels of employment. This is expected, as fewer establishments and workers contribute data to cells with relatively low levels of employment, and the lower are the number of establishments and workers contributing data, the more detectable are outliers that emerge from the imputation process. Also expected is the relative tightness of the distributions of the bias when comparing across levels of industry aggregation. The SIC Division level distributions of the bias are tighter than the two-digit SIC distributions, as more establishments and workers contribute more data to each SIC Division level cell. The tightening distributions are clear when examining the 90-10 diﬀerential. For B at the SIC Division level, for example, the spread between the 90th and 10th percentile falls by 0.19 when the distribution of the bias is weighted. It is also clear that the spread between the 90th and 10th percentiles is smaller for the SIC Division level as compared to the two-digit SIC level of aggregation. Time-Series Properties of Disclosure Avoidance System The disclosure avoidance algorithm described in section 5.6 has the dual goals of preserving confidentiality and maintaining a high level of analytical validity of the public-use data. This section draws on Abowd, Stephens, and Vilhuber (2006), who provide an in-depth analysis of the extent of disclosure protection and the degree to which analytical validity is maintained. The analysis presented in this subsection focuses on the time series properties of the published QWI, after noise-infusion and suppressions. Abowd, Stephens, and Vilhuber (2006) also show the cross-sectional unbiasedness of the published data. In each case, data from two states (Illinois and Maryland) were used. The unit of analysis is an interior substate geography industry age sex cell kt. Substate geography in all cases is a county, whereas the industry classification is SIC. Analytical validity is obtained when the data display no bias and the additional dispersion due to the confidentiality protection system can be quantified so that statistical inferences can be adjusted to accommodate it. To analyze the impact on the time series properties of the distorted data,

0.09165 0.03697 0.00415 –0.02232 –0.08075 0.17239

0.13802 0.03965 –0.00010 –0.04169 –0.13951 0.27752

0.21484 0.07426 0.00002 –0.03475 –0.15197 0.36680

0.25999 0.04425 –0.00004 –0.00232 –0.17605 0.43603

90 75 50 25 10 P90–P10

90 75 50 25 10 P90–P10 0.25257 0.04218 –0.00004 –0.00232 –0.18858 0.44114

0.21274 0.07287 0.00000 –0.03696 –0.15879 0.37152

Unweighted

0.13494 0.04002 –0.00005 –0.04061 –0.13865 0.27359

0.09021 0.03709 0.00482 –0.02159 –0.08020 0.17040

Weighted

Full-quarter employment Weighted

0.14072 0.04312 –0.00449 –0.04274 –0.11827 0.25898 0.17390 0.03683 –0.00111 –0.07000 –0.20784 0.38173

Unweighted SIC Division 0.27966 0.09597 0.00001 –0.01610 –0.17476 0.45441 2-Digit SIC 0.30875 0.05200 –0.00004 –0.00075 –0.20977 0.51851

Accessions

0.29985 0.04938 –0.00004 –0.00075 –0.22191 0.52175

0.26821 0.09332 0.00000 –0.02105 –0.18000 0.44820

Unweighted

0.17117 0.03622 –0.00075 –0.06709 –0.19383 0.36500

0.13265 0.04043 –0.00414 –0.04015 –0.10717 0.23981

Weighted

Separations

0.36038 0.05296 –0.00002 –0.00127 –0.16911 0.52948

0.28185 0.09184 0.00262 –0.02998 –0.16044 0.44229

Unweighted

0.16118 0.04267 –0.00004 –0.04034 –0.14427 0.30544

0.11456 0.04159 0.00405 –0.02219 –0.08178 0.19633

Weighted

Total payroll

Notes: Estimated using Minnesota data. A statistic XTrue is calculated using job histories coded to the SEIN/SEINUNIT level; a statistic XImputed is calculated using job histories coded to the SEIN level with imputed SEINUNITs. See text for further details.

Weighted

Beginning-of-period employment

XImputed – XTrue Bias = XTrue

Distribution of proportional bias in unit-to-worker imputation

Unweighted

Table 5.8

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators Table 5.9

199

Distribution of the error in first order serial correlation of QWI Δr = r – r*

Percentile

Beginning-of-quarter employment

Accessions

Separations

Full-quarter employment

Net job flows

01 05 10 25 50 75 90 95 99

–0.085495 –0.047704 –0.034558 –0.015317 –0.000512 0.013438 0.030963 0.044796 0.080282

IL County × SIC Division –0.092455 –0.098770 –0.046665 –0.045208 –0.031767 –0.032898 –0.014197 –0.015077 –0.000997 –0.000707 0.011536 0.012457 0.027037 0.028835 0.037906 0.041862 0.079122 0.083824

–0.079205 –0.046830 –0.033607 –0.015533 –0.001000 0.011670 0.027970 0.040096 0.077419

–0.008447 –0.004959 –0.003186 –0.001189 –0.000049 0.000861 0.002489 0.004801 0.007537

01 05 10 25 50 75 90 95 99

–0.065342 –0.035974 –0.024174 –0.010393 0.000230 0.011382 0.025160 0.035176 0.060042

MD County × SIC Division –0.072899 –0.072959 –0.036995 –0.040314 –0.027689 –0.028577 –0.013686 –0.012505 –0.000542 0.000797 0.012628 0.013034 0.026325 0.025272 0.034114 0.034999 0.056477 0.055043

–0.058021 –0.030985 –0.021361 –0.009401 0.000279 0.009429 0.022027 0.030152 0.049213

–0.009081 –0.004540 –0.002823 –0.001243 –0.000025 0.001045 0.002799 0.004321 0.009208

Notes: Estimated from undistorted (r) and published data (r*). Unit of observation is a county × SIC division × age-group × sex cell for all private employment, interior cells only. For more details, see text and Abowd, Stephens, and Vilhuber (2006).

we estimated an AR(1) for the time series associated with each cell kt, using county-level data for all counties in each state. Two AR(1) coeﬃcients are estimated for each cell time series. The first order serial correlation coeﬃcient computed using undistorted data is denoted by r. The estimate computed using the distorted data is denoted by r∗. For each cell, the error r r – r∗ is computed. Table 5.9 shows the distribution of the errors r across SIC-division county cells, for B, A, S, F, and JF when comparing raw (confidential) data to published data, which excludes suppressed data items. The table shows that the time series properties of all variables analyzed remain largely unaﬀected by the distortion. The maximum bias (as measured by the median of this distribution) is never greater than 0.001. The error distribution is tight; the semi-interquartile range of the distortion for B in Maryland is 0.010, which is less than the precision with which estimated serial correlation coeﬃcients are normally displayed. The maximum semiinterquartile range for any variable in either of the two states is 0.012.26 26. Abowd, Stephens, and Vilhuber (2006) also report that the maximum semi-interquartile range for SIC2-based variables is 0.0241, and for SIC3-based variables, 0.0244.

200

J. Abowd et al.

Although the overall spread of the distribution is slightly higher when considering two-digit SIC county and three-digit SIC county cells, which are sparser than the SIC-division county cells, the general results hold in these cases as well (Abowd, Stephens, and Vilhuber 2006, tables 7 and 8). Abowd, Stephens, and Vilhuber (2006) thus conclude that the time series properties of the QWI data are unbiased with very little additional noise, which is, in general, economically meaningless. 5.8 Public-Use and Restricted-Access Files In this section, we briefly describe the public-use release files and those files available at Census Research Data Centers. We focus on how these files diﬀer from the corresponding internal files discussed in the rest of the article. 5.8.1 Public-Use Files Three public-use products, fully or partially based on QWI data, are currently available on a regular basis: the QWI distribution files, the Older Worker Reports, and OnTheMap.27 A subset of eight variables from the full public-use release is available at QWI Online (http://lehd.did .census.gov/). Additional variables are used in other applications accessible from the same Census Bureau web site. The complete set of QWI public-use variables is available from the Cornell Virtual Research Data Center (VirtualRDC) as of January 2008. The VirtualRDC is partially funded by grants from the National Science Foundation (NSF) and the National Institute on Aging. Computing resources to manipulate complete QWI are also available on the VirtualRDC for qualified researchers (http://www.vrdc.cornell.edu/). Other distribution options for the full QWI data may be available when this volume appears. Up-to-date information on all access options is posted at http://lehd.did.census.gov/. The public-use QWI data diﬀer from the Census-internal version because the public-use version has been subjected to the disclosure avoidance methods described in section 5.6. In order to preserve the integrity of these disclosure avoidance algorithms, all special tabulations released from the QWI must follow the same procedures. 5.8.2 Restricted-Access Files A larger set of files are available within the protected environment provided by the Census Research Data Centers (RDCs). The only information removed from RDC versions of QWI and LEHD Infrastructure files rela27. The Older Worker Reports are based entirely on the QWI public use files. OnTheMap uses the QWI micro data to produce a QWI report for the user-defined geographic area.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

201

tive to their internal-use counterparts is the information specifically used to do confidentiality protection of the QWI—the fuzz factors and the fuzzed data items. All of RDC-accessible LEHD files can be used for research purposes by submitting a research proposal to the Center for Economic Studies (CES) at the U.S. Census Bureau.28 ECF The version of the ECF available in the RDC environment is called the LEHD-ECF on the CES RDC documentation. It is identical to the one described in section 5.3.4 except for the removal of the QWI fuzz factors. Unit Flow Files: Establishment-Level QWI Data The SEINUNIT-level input files to the final aggregation step of the QWI, internally known as UFF_B, are available in the RDC environment under the name LEHD-QWI. These files are identical to final establishment level flow files documented in section 5.5 except that they contain only the unfuzzed raw aggregates. Establishment Crosswalks: Business Register Bridge The Census Bureau maintains lists of establishments to develop the frames for economic censuses and surveys. These lists are called the Employer and Nonemployer Business Registers (BR). The research version of the Employer BR is maintained by CES, which produces a new set of files annually. The BR contains very reliable information on business identifiers, business organizational structure, and business location. Unfortunately, the establishment identification system for the Business Register diﬀers from the LEHD establishment identifier (SEIN/SEINUNIT). As a consequence, there is no single best way to form linkages between these data sources. The LEHD Business Register Bridge (LEHD-BRB) that is available in the RDC network provides several ways to integrate the economic censuses and surveys with LEHD-provided data. The choice of linking strategy is left to researchers, who must determine the best definition of an entity on both side of the linkage, considering data sources and the stated research objectives. Available identifiers on the LEHD-BRB that are common to both the LEHD Infrastructure Files and the BR are the EIN, geographic information, and four-digit SIC. These variables may be used to construct pseudo-establishments that are aggregates of SEIN/SEINUNIT establishments at diﬀerent levels of aggregation. These identifiers can also be linked to sets of ALPHA/CFN establishments on the BR and other Census economic data products. A more detailed guide is available on the CES or LEHD web site (Chiang, Sandusky, and Vilhuber 2005). 28. Available at http://www.ces.census.gov

202

J. Abowd et al.

Household and Establishment Geocoding: GAL The GAL (Geocoded Address List) described in section 5.3.3, is available in the RDC environment under the reference LEHD-GAL. Access to the GAL is predicated on the project having permission to use business or residential address information from other RDC-available source files. Once that permission has been properly established, the researcher is granted access to the GAL for the purpose of obtaining a consistent set of geocodes. Wage Decomposition Data: Human Capital Files These files will contain employer-level distributions of human capital measures as initially developed in Abowd, Lengermann, and McKinney (2002). They are expected to become available in 2009. Remaining LEHD Infrastructure Files The remaining LEHD Infrastructure Files outlined in this chapter are available in the RDC environment as of the time of publication of this volume. In general, unless explicitly mentioned above, these files are provided to researchers as-is, and are subject to the same Title 13 use restrictions as all other data on the RDC network. The LEHD Infrastructure data are also subject to usage restriction in the MOU that governs the Census Bureau and state participation in the LED partnership. The most important of those restrictions is the one that requires the written consent of the state’s signatory oﬃcial on the MOU before state-specific results based on the LEHD Infrastructure Files may be released. Results may be released from analyses performed on multiple states. For up-to-date details, researchers should contact CES directly. 5.9 Concluding Remarks 5.9.1 Future Projects This section describes some of the ongoing eﬀorts to improve the LEHD Infrastructure Files. Planned Improvements to the ICF Currently, researchers at LEHD are developing an enhanced, longitudinal version of the ICF, internally named ICF version 4 because the current system was the third version of most of the infrastructure files. The improved ICF is the first national LEHD Infrastructure File system. Individuals appearing in any state, including those that have not yet joined the LED federal/state partnership have their ICF data on a set of annual records. Additional data sources will be integrated with this enhanced version of

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

203

the ICF using direct links. The statistical link to the 1990 Decennial Census will be replaced by a direct link to the 2000 Decennial Census, and additional links to the ACS will be incorporated. The existing education imputation will greatly benefit from this enhancement. The additional links, as well as improved links to currently integrated data, will also allow for additional time-invariant characteristics to be incorporated and completed, including information on race and ethnicity and additional time-varying characteristics such as Temporary Assistance for Needy Families (TANF) recipiency. Longitudinal residence information will be appended to the ICF based on the information available from the StARS. Where appropriate, residence will be imputed based on a change in residence imputation model and Bayesian methods for imputing geography at the block level, replacing the current residential address missing data imputation model. In fact, all imputation models will be based on the most up-to-date imputation engines developed at LEHD. Planned Improvements to the EHF The UI wage records in several states suﬀer from defects in the historical records. These defects can be detected automatically when they produce a big enough fluctuation in certain flow statistics, typically beginning of period employment as compared to total flow employment. Algorithms have been developed to detect the probable existence of missing wage records using the posterior predictive distribution of employment histories given the available data and an informative prior on certain patterns. Once detected, the missing wage records are imputed, again using appropriate Bayesian methods. The same imputation engines are also being used to impute top-coded UI wages. These improvements are in the testing stage and should be implemented within the next year. Planned Improvements to the ECF Two major enhancements to the ECF are in development. The first is a probabilistic record link to the Census Bureau’s Business Register in order to improve the physical addresses on the ECF. This enhancement is currently in the testing phase. The second major enhancement, which impacts not just the ECF, is the expansion of coverage to include entities so far not covered by the LEHD Infrastructure. Integration of Data from Missing Parts of the Universe Nonemployer data. The job universe currently used by all LEHD Infrastructure Files is legal employment with an employer that has mandatory reporting to the state UI wage record system. Nonemployer businesses are out-of-scope for this universe but are of intrinsic interest in the economic analysis of sources of labor income. In addition, the income to the sole pro-

204

J. Abowd et al.

prietor of an employer business is of interest as a source of labor income. The LEHD Program and CES are collaborating in developing enhancements to the Business Register to account for nonemployer income sources and to better track sole proprietor employers. The nonemployer enhancements will also aﬀect the LEHD Infrastructure Files because the information on the identity of the nonemployer, the identity of the nonemployer business, and the income from the nonemployer business provides a job record for this activity, which can then be integrated with the EHF, ICF, and ECF file systems. Federal government employment. The LEHD program has completed an MOU with the Oﬃce of Personnel Management in the federal government to obtain historical and ongoing information from the OPM databases that permits construction of LEHD Infrastructure File system records that correspond to job histories for federal employees in the EHF and employer-establishment records in the ECF. Records already exist for these individuals in the new ICF. Creation of Public-Use Synthetic Data As a part of a National Science Foundation Information Technology Research grant (SES-0427889), awarded to a consortium of Census Research Data Centers, researchers at LEHD and other parts of Census are collaborating with social scientists and statisticians working in the RDCs to create and validate synthetic micro data from the LEHD Infrastructure Files. Such synthetic micro data will be confidentiality protected so that they may be released for public use. They will also be inference valid—permitting the estimation of some statistical models with results comparable to those obtained on the confidential micro data. The First Twenty-First Century Statistical System The goal of the development of the Quarterly Workforce Indicators was to create a twenty-first century statistical system. Without increasing respondent burden, the LEHD infrastructure permits the creation of extremely detailed estimates that, for the first time in the United States, provide integrated demographic and economic information about the local labor market. The same techniques will work for other areas of interest— transportation dynamics and welfare-to-work dynamics, to name just two examples. The two essential features of twenty-first century statistical systems will be their heavy reliance on existing data instruments (surveys, censuses, and administrative records that are already in production) and their extensive use of data-intensive statistical modeling to enhance and summarize this information. In these regards, we think the LEHD infrastructure and the QWI system are worthy pioneers.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

205

Appendix Definitions of Fundamental LEHD Concepts A.1 A.1.1

Fundamental Concepts Dates

The QWI are a quarterly data system with calendar year timing. We use the notation yyyy:q to refer to a year and quarter combination. For example, 1999:4 refers to the fourth quarter of 1999, which includes the months October, November, and December. A.1.2

Employer

An employer in the QWI system consists of a single Unemployment Insurance (UI) account in a given state’s UI wage reporting system. For statistical purposes, the QWI system creates an employer identifier called a State Employer Identification Number (SEIN) recoded from the UI account number and information about the state (FIPS code). Thus, within the QWI system, the SEIN is a unique identifier within and across states but the entity to which it refers is a UI account. A.1.3

Establishment

For a given employer in the QWI system, a SEIN, each physical location within the state is assigned a unit number, called the SEINUNIT. This SEINUNIT is recoded from the reporting unit in the ES-202 files supplied by the states. All QWI statistics are produced by aggregating statistics calculated at the establishment level. Single-unit SEINs are UI accounts associated with a single reporting unit in the state. Thus, single-unit SEINs have only one associated SEINUNIT in every quarter. Multi-unit SEINs have two or more SEINUNITs associated for some quarters. Since the UI wage records are not coded down to the SEINUNIT, SEINUNITs are multiply imputed as described in section 5.4.2 on the unit-to-worker imputation. A feature of this imputation system is that it does not permit SEINUNIT to SEINUNIT movements within the same SEIN. Thus, for multi-unit SEINs, the following definitions produce the same flow estimates at the SEIN level whether the definition is applied to the SEIN or the SEINUNIT. A.1.4

Employee

Individual employees are identified by their Social Security Numbers (SSN) on the UI wage records that provide the input to the QWI. To protect the privacy of the SSN and the individual’s name, a diﬀerent branch of the Census Bureau removes the name and replaces the SSN with an internal Census identifier called a Protected Identification Key (PIK).

206

J. Abowd et al.

A.1.5 Job The QWI system definition of a job is the association of an individual (PIK) with an establishment (SEINUNIT) in a given year and quarter. The QWI system stores the entire history of every job that an individual holds. Estimates are based on the following definitions, which formalize how the QWI system estimates the start of a job (accession), employment status (beginning- and end-of-quarter employment), continuous employment (full-quarter employment), the end of a job (separation), and average earnings for diﬀerent groups. A.1.6 Unemployment Insurance Wage Records (the QWI System Universe) The Quarterly Workforce Indicators are built upon concepts that begin with the report of an individual’s UI-covered earnings by an employing entity (SEIN). An individual’s UI wage record enters the QWI system if at least one employer reports earnings of at least one dollar for that individual (PIK) during the quarter. Thus, the job must produce at least one dollar of UI-covered earnings during a given quarter to count in the QWI system. The presence of this valid UI wage record in the QWI system triggers the beginning of calculations that estimate whether that individual was employed at the beginning of the quarter, at the end of the quarter, and continuously throughout the quarter. These designations are discussed later. Once these point-in-time employment measures have been estimated for the individual, further analysis of the individual’s wage records results in estimates of full-quarter employment, accessions, separations (point-intime and full-quarter), job creations and destructions, and a variety of fullquarter average earnings measures. A.1.7

Employment at a Point in Time

Employment is estimated at two points in time during the quarter, corresponding to the first and last calendar days. An individual is defined as employed at the beginning of the quarter when that individual has valid UI wage records for the current quarter and the preceding quarter. Both records must apply to the same employer (SEIN). An individual is defined as employed at the end of the quarter when that individual has valid UI wage records for the current quarter and the subsequent quarter. Again, both records must show the same employer. The QWI system uses beginning and end-of-quarter employment as the basis for constructing worker and job flows. In addition, these measures are used to check the external consistency of the data, since a variety of employment estimates are available as point-in-time measures. Many federal statistics are based upon estimates of employment as of the twelfth day of particular months. The Census Bureau uses March 12 as the reference date for employment mea-

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

207

sures contained in its Business Register and on the Economic Censuses and Surveys. The BLS “Covered Employment and Wages (CEW)” series, which is based on the QCEW—formerly ES-202—data, use the twelfth of each month as the reference date for employment. The QWI system cannot use exactly the same reference date as these other systems because UI wage reports do not specify additional detail regarding the timing of these payments. The LEHD research has shown that the point-in-time definitions used to estimate beginning and end-of-quarter employment track the CEW month-one employment estimates well at the level of an employer (SEIN). For single-unit SEINs, there is no diﬀerence between an employer-based definition and an establishment-based definition of point-in-time employment. For multi-unit SEINs, the unit-to-worker imputation model assumes that unit-to-unit transitions within the same SEIN cannot occur. Therefore, point-in-time employment defined at either the SEIN or SEINUNIT level produces the same result. A.1.8

Employment for a Full Quarter

The concept of full-quarter employment estimates individuals who are likely to have been continuously employed throughout the quarter at a given employer. An individual is defined as full-quarter employed if that individual has valid UI wage records in the current quarter, the preceding quarter, and the subsequent quarter at the same employer (SEIN). That is, in terms of the point-in-time definitions, if the individual is employed at the same employer at both the beginning and end of the quarter, then the individual is considered full-quarter employed in the QWI system. Consider the following example. Suppose that an individual has valid UI wage records at employer A in 1999:2, 1999:3, and 1999:4. This individual does not have a valid UI wage record at employer A in 1999:1 or 2000:1. Then, according to the previous definitions, the individual is employed at the end of 1999:2, the beginning and end of 1999:3, and the beginning of 1999:4 at employer A. The QWI system treats this individual as a fullquarter employee in 1999:3 but not in 1999:2 or 1999:4. Full-quarter status is not defined for either the first or last quarter of available data. A.1.9

Point-in-Time Estimates of Accession and Separation

An accession occurs in the QWI system when it encounters the first valid UI wage record for a job (an individual [PIK]-employer [SEIN] pair). Accessions are not defined for the first quarter of available data from a given state. The QWI definition of an accession can be interpreted as an estimate of the number of new employees added to the payroll of the employer (SEIN) during the quarter. The individuals who acceded to a particular employer were not employed by that employer during the previous quarter, but received at least one dollar of UI-covered earnings during the quarter of accession.

208

J. Abowd et al.

A separation occurs in the current quarter of the QWI system when it encounters no valid UI wage record for an individual-employer pair in the subsequent quarter. This definition of separation can be interpreted as an estimate of the number of employees who left the employer during the current quarter. These individuals received UI-covered earnings during the current quarter but did not receive any UI-covered earnings in the next quarter from this employer. Separations are not defined for the last quarter of available data. A.1.10

Accession and Separation from Full-Quarter Employment

Full-quarter employment is not a point-in-time concept. Full-quarter accession refers to the quarter in which an individual first attains full-quarter employment status at a given employer. Full-quarter separation occurs in the last full-quarter that an individual worked for a given employer. As previously noted, full-quarter employment refers to an estimate of the number of employees who were employed at a given employer during the entire quarter. An accession to full-quarter employment, then, involves two additional conditions that are not relevant for ordinary accessions. First, the individual (PIK) must still be employed at the end of the quarter at the same employer (SEIN) for which the ordinary accession is defined. At this point (the end of the quarter where the accession occurred and the beginning of the next quarter) the individual has acceded to continuingquarter status. An accession to continuing-quarter status means that the individual acceded in the current quarter and is end-of-quarter employed. Next, the QWI system must check for the possibility that the individual becomes a full-quarter employee in the subsequent quarter. An accession to full-quarter status occurs if the individual acceded in the previous quarter, and is employed at both the beginning and end of the current quarter. Consider the following example. An individual’s first valid UI wage record with employer A occurs in 1999:2. Thus, the individual acceded in 1999:2. The same individual has a valid wage record with employer A in 1999:3. The QWI system treats this individual as end-of-quarter employed in 1999:2 and beginning-of-quarter employed in 1999:3. Thus, the individual acceded to continuing-quarter status in 1999:2. If the individual also has a valid UI wage record at employer A in 1999:4, then the individual is fullquarter employed in 1999:3. Since 1999:3 is the first quarter of full-quarter employment, the QWI system considers this individual an accession to full-quarter employment in 1999:3. Full-quarter separation works much the same way. One must be careful about the timing, however. If an individual separates in the current quarter, then the QWI system looks at the preceding quarter to determine if the individual was employed at the beginning of the current quarter. An individual who separates in a quarter in which that person was employed at the beginning of the quarter is a separation from continuing-quarter status in

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

209

the current quarter. Finally, the QWI system checks to see if the individual was a full-quarter employee in the preceding quarter. An individual who was a full quarter employee in the previous quarter is treated as a fullquarter separation in the quarter in which that person actually separates. Note, therefore, that the definition of full-quarter separation preserves the timing of the actual separation (current quarter) but restricts the estimate to those individuals who were full-quarter status in the preceding quarter. For example, suppose that an individual separates from employer A in 1999:3. This means that the individual had a valid UI wage record at employer A in 1999:3 but did not have a valid UI wage record at employer A in 1999:4. The separation is dated 1999:3. Suppose that the individual had a valid UI wage record at employer A in 1999:2. Then, a separation from continuing quarter status occurred in 1999:3. Finally, suppose that this individual had a valid UI wage record at employer A in 1999:1. Then, this individual was a full-quarter employee at employer A in 1999:2. The QWI system records a full-quarter separation in 1999:3. A.1.11

Point-in-Time Estimates of New Hires and Recalls

The QWI system refines the concept of accession into two subcategories: new hires and recalls. In order to do this, the QWI system looks at a full year of wage record history prior to the quarter in which an accession occurs. If there are no valid wage records for this job (PIK-SEIN) during the four quarters preceding an accession, then the accession is called a new hire; otherwise, the accession is called a recall. Thus, new hires and recalls sum to accessions. For example, suppose that an individual accedes to employer A in 1999:3. Recall that this means that there is a valid UI wage record for the individual 1 at employer A in 1999:3 but not in 1999:2. If there are also no valid UI wage records for individual 1 at employer A for 1999:1, 1998:4, and 1998:3, then the QWI system designates this accession as a new hire of individual 1 by employer A in 1999:3. Consider a second example in which individual 2 accedes to employer B in 2000:2. Once again, the accession implies that there is not a valid wage record for individual 2 at employer B in 2000:1. If there is a valid wage record for individual 2 at employer B in 1999:4, 1999:3, or 1999:2, then the QWI system designates the accession of individual 2 to employer B as a recall in 2000:2. New hire and recall data, because they depend upon having four quarters of historical data, only become available one year after the data required to estimate accessions become available. A.1.12

New Hires and Recalls to and from Full-Quarter Employment

Accessions to full-quarter status can also be decomposed into new hires and recalls. The QWI system accomplishes this decomposition by classifying all accessions to full-quarter status who were classified as new hires in

210

J. Abowd et al.

the previous quarter as new hires to full-quarter status in the current quarter. Otherwise, the accession to full-quarter status is classified as a recall to full-quarter status. For example, if individual 1 accedes to full-quarter status at employer A in 1999:4, then, according to the previous definitions, individual 1 acceded to employer A in 1999:3 and reached full-quarter status in 1999:4. Suppose that the accession to employer A in 1999:3 was classified as a new hire; then, the accession to full quarter status in 1999:4 is classified as a full-quarter new hire. For another example, consider individual 2, who accedes to full-quarter status at employer B in 2000:3. Suppose that the accession of individual 2 to employer B in 2000:2, which is implied by the full-quarter accession in 2000:3, was classified by the QWI system as a recall in 2000:2; then, the accession of individual 2 to full-quarter status at employer B in 2000:3 is classified as a recall to full-quarter status. A.1.13

Job Creations and Destructions

Job creations and destructions are defined at the employer (SEIN) level and not at the job (PIK-SEIN) level. For single-unit employers, there is never more than one SEINUNIT per quarter, so the definition at the employer level and the definition at the establishment level are equivalent. For multi-unit employers, the QWI system performs the calculations at the establishment level (SEINUNIT); however, the statistical model for imputing establishment described in section 5.4.2 does not permit establishmentto-establishment flows. Hence, although the statistics are estimated at the establishment level, the sum of job creations and destructions at a given employer in a given quarter across all establishments active that quarter is exactly equal to the measure of job creations that would have been estimated by using employer-level inputs (SEIN) directly. To construct an estimate of job creations and destructions, the QWI system totals beginning and ending employment for each quarter for every employer in the UI wage record universe; that is, for an employer who has at least one valid UI wage record during the quarter. The QWI system actually uses the Davis et al. (1996) formulas for job creation and destruction (see definitions in appendix A.2). Here, we use a simplified definition. If end-of-quarter employment is greater than beginning-of-quarter employment, then the employer has created jobs. The QWI system sets job creations in this case equal to end-of-quarter employment less beginning-ofquarter employment. The estimate of job destructions in this case is zero. On the other hand, if beginning-of-quarter employment exceeds end-ofquarter employment, then this employer has destroyed jobs. The QWI system computes job destructions in this case as beginning-of-period employment less end-of-period employment. The QWI system sets job creations to zero in this case. Notice that either job creations are positive or job destructions are positive, but not both. Job creations and job destructions can simultaneously be zero if beginning-of-quarter employment

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

211

equals end-of-quarter employment. There is an important subtlety regarding job creations and destructions when they are computed for diﬀerent sex and age groups within the same employer. There can be creation and destruction of jobs for certain demographic groups within the employer without job creation or job destruction occurring overall. That is, jobs can be created for some demographic groups and destroyed for others even at enterprises that have no change in employment as a whole. Here is a simple example. Suppose employer A has 250 employees at the beginning of 2000:3 and 280 employees at the end of 2000:3. Therefore, employer A has 30 job creations and zero job destructions in 2000:3. Now suppose that of the 250 employees, 100 are men and 150 are women at the beginning of 2000:3. At the end of the quarter suppose that there are 135 men and 145 women. Then, job creations for men are 35 and job destructions for men are 0 in 2000:3. For women in 2000:3 job creations are 0 and job destructions are 5. Notice that the sum of job creations for the employer by sex (35 0) is not equal to job creations for the employer as a whole (30) and that the sum of job destructions by sex (0 5) is not equal to job destructions for the employer as a whole. A.1.14

Net Job Flows

Net job flows are also only defined at the level of an employer (SEIN). Once again, the QWI system computes these statistics at the establishment level but does not allow establishment-to-establishment flows. Hence, the estimates for a given employer (SEIN) are the sum of the estimates for that employer’s establishments (SEINUNIT) that are active in the given quarter. Net job flows are the diﬀerence between job creations and job destructions. Thus, net job flows are always equal to end-of-quarter employment less beginning-of-quarter employment. If we return to the example in the description of job creations and destructions, employer A has 250 employees at the beginning of 2000:3 and 280 employees at the end of 2000:3. Net job flows are 30 (job creations less job destructions or beginning-of-quarter employment less end-of-quarter employment). Suppose, once again, that employment of men goes from 100 to 135 from the beginning to the end of 2000:3 and employment of women goes from 150 to 145. Notice that net job flows for men (35) plus net job flows for women (–5) equals net job flows for the employer as a whole (30). Net job flows are additive across demographic groups even though gross job flows (creations and destructions) are not. Some useful relations among the worker and job flows include: • Net job flows job creations – job destructions • Net job flows end-of-quarter employment – beginning-of-period employment • Net job flows accessions – separations

212

J. Abowd et al.

These relations hold for every demographic group and for the employer as a whole. Additional identities are shown in the second section of the appendix. A.1.15 Full-Quarter Job Creations, Job Destructions, and Net Job Flows The QWI system applies the same job flow concepts to full-quarter employment to generate estimates of full-quarter job creations, full-quarter job destructions, and full-quarter net job flows. Full-quarter employment in the current quarter is compared to full-quarter employment in the preceding quarter. If full-quarter employment has increased between the preceding quarter and the current quarter, then full-quarter job creations are equal to full-quarter employment in the current quarter less full-quarter employment in the preceding quarter. In this case full-quarter job destructions are zero. If full-quarter employment has decreased between the previous and current quarters, then full-quarter job destructions are equal to full-quarter employment in the preceding quarter minus full-quarter employment in the current quarter. In this case, full-quarter job destructions are zero. Full-quarter net job flows equal full-quarter job creations minus full-quarter job destructions. The same identities that hold for the regular job flow concepts hold for the full-quarter concepts. A.1.16

Average Earnings of End-of-Period Employees

The average earnings of end-of-period employees is estimated by first totaling the UI wage records for all individuals who are end-of-period employees at a given employer in a given quarter. Then, the total is divided by the number of end-of-period employees for that employer and quarter. A.1.17

Average Earnings of Full-Quarter Employees

Measuring earnings using UI wage records in the QWI system presents some interesting challenges. The earnings of end-of-quarter employees who are not present at the beginning of the quarter are the earnings of accessions during the quarter. The QWI system does not provide any information about how much of the quarter such individuals worked. The range of possibilities goes from one day to every day of the quarter. Hence, estimates of the average earnings of such individuals may not be comparable from quarter to quarter unless one assumes that the average accession works the same number of quarters regardless of other conditions in the economy. Similarly, the earnings of beginning-of-quarter workers who are not present at the end of the quarter represent the earnings of separations. These present the same comparison problems as the average earnings of accessions; namely, it is diﬃcult to model the number of weeks worked during the quarter. If we consider only those individuals employed at the employer in a given quarter who were neither accessions nor separations during that quarter, we are left, exactly, with the full-quarter employees.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

213

The QWI system measures the average earnings of full-quarter employees by summing the earnings on the UI wage records of all individuals at a given employer who have full-quarter status in a given quarter, then dividing by the number of full-quarter employees. For example, suppose that in 2000:2 employer A has ten full-quarter employees and that their total earnings are $300,000. Then, the average earnings of the full-quarter employees at A in 2000:2 is $30,000. Suppose, also, that six of these employees are men and that their total earnings are $150,000. So, the average earnings of full-quarter male employees is $25,000 in 2000:2 and the average earnings of female full-quarter employees is $37,500 ( $150,000/4). A.1.18

Average Earnings of Full-Quarter Accessions

As discussed previously, a full-quarter accession is an individual who acceded in the preceding quarter and achieved full-quarter status in the current quarter. The QWI system measures the average earnings of fullquarter accessions in a given quarter by summing the UI wage record earnings of all full-quarter accessions during the quarter and dividing by the number of full-quarter accessions in that quarter. A.1.19

Average Earnings of Full-Quarter New Hires

Full-quarter new hires are accessions to full-quarter status who were also new hires in the preceding quarter. The average earnings of full-quarter new hires are measured as the sum of UI wage records for a given employer for all full-quarter new hires in a given quarter divided by the number of fullquarter new hires in that quarter. A.1.20

Average Earnings of Full-Quarter Separations

Full-quarter separations are individuals who separate during the current quarter who were full-quarter employees in the previous quarter. The QWI system measures the average earnings of full-quarter separations by summing the earnings for all individuals who are full-quarter status in the current quarter and who separate in the subsequent quarter. This total is then divided by full-quarter separations in the subsequent quarter. Thus, the average earnings of full-quarter separations are the average earnings of fullquarter employees in the current quarter who separated in the next quarter. Note the dating of this variable. A.1.21 Average Periods of Nonemployment for Accessions, New Hires, and Recalls As noted previously, an accession occurs when a job starts; that is, on the first occurrence of a SEIN-PIK pair following the first quarter of available data. When the QWI system detects an accession, it measures the number of quarters (up to a maximum of four) that the individual spent nonemployed in the state prior to the accession. The QWI system estimates the

214

J. Abowd et al.

number of quarters spent nonemployed by looking for all other jobs held by the individual at any employer in the state in the preceding quarters up to a maximum of four. If the QWI system does not find any other valid UI wage records in a quarter preceding the accession, it augments the count of nonemployed quarters for the individual who acceded, up to a maximum of four. Total quarters of nonemployment for all accessions is divided by accessions to estimate average periods of nonemployment for accessions. Here is a detailed example. Suppose individual 1 and individual 2 accede to employer A in 2000:1. In 1999:4, individual A does not work for any other employers in the state. In 1999:1 through 1999:3 individual 1 worked for employer B. Individual 1 had one quarter of nonemployment preceding the accession to employer A in 2000:1. Individual 2 has no valid UI wage records for 1999:1 through 1999:4. Individual 2 has four quarters of nonemployment preceding the accession to employer A in 2000:1. The accessions to employer A in 2000:1 had an average of 2.5 quarters of nonemployment in the state prior to accession. Average periods of nonemployment for new hires and recalls are estimated using exactly analogous formulas except that the measures are estimated separately for accessions who are also new hires as compared with accession who are recalls. A.1.22

Average Number of Periods of Nonemployment for Separations

Analogous to the average number of periods of nonemployment for accessions prior to the accession, the QWI system measures the average number of periods of nonemployment in the state for individuals who separated in the current quarter, up to a maximum of four. When the QWI system detects a separation, it looks forward for up to four quarters to find valid UI wage records for the individual who separated among other employers in the state. Each quarter that it fails to detect any such jobs is counted as a period of nonemployment, up to a maximum of four. The average number of periods of nonemployment is estimated by dividing the total number of periods of nonemployment for separations in the current quarter by the number of separations in the quarter. A.1.23 Average Changes in Total Earnings for Accessions and Separations The QWI system measures the change in total earnings for individuals who accede or separate in a given quarter. For an individual accession in a given quarter, the QWI system computes total earnings from all valid wage records for all of the individual’s employers in the preceding quarter. The system then computes the total earnings for the same individual for all valid wage records and all employers in the current quarter. The acceding individual’s change in earnings is the diﬀerence between the cur-

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

215

rent quarter earnings from all employers and the preceding quarter earnings from all employers. The average change in earnings for all accessions is the total change in earnings for all accessions divided by the number of accessions. The QWI system computes the average change in earnings for separations in an analogous manner. The system computes total earnings from all employers for the separating individual in the current quarter and subtracts total earnings from all employers in the subsequent quarter. The average change in earnings for all separations is the total change in earnings for all separations divided by the number of separations. Here is an example for the average change in earnings of accessions. Suppose individual 1 accedes to employer A in 2000:3. Earnings for individual 1 at employer A in 2000:3 are $8,000. Individual 1 also worked for employer B in 2000:2 and 2000:3. Individual 1’s earnings at employer B were $7,000 and $3,000 in 2000:2 and 2000:3, respectively. Individual 1’s change in total earnings between 2000:3 and 2000:2 was $4,000 ( $8,000 $3,000 – $7,000). Individual 2 also acceded to employer A in 2000:3. Individual 2 earned $9,000 from employer A in 2000:3. Individual 2 had no other employers during 2000:2 or 2000:3. Individual 2’s change in total earnings is $9,000. The average change in earnings for all of employer A’s accessions is $6,500 ( [$4,000 $9,000] /2) , the average change in total earnings for individuals 1 and 2. A.2 A.2.1

Definitions of Job Flow, Worker Flow, and Earnings Statistics Overview and Basic Data Processing Conventions

For internal processing the variable t refers to the sequential quarter. The variable t runs from qmin to qmax, regardless of the state being processed. The quarters are numbered sequentially from 1 (1985:1) to the latest available quarter. These values are qmin 1 (1985:1) and qmax 88 (2006:4), as of December 2007. For publication, presentation, and internal data files, all dates are presented as (year:quarter) pairs (e.g., 1990:1) for first quarter 1990. The variable qfirst refers to the first available sequential quarter of data for a state (e.g., qfirst 21 for Illinois). The variable qlast refers to the last available sequential quarter of data for a state (e.g., qlast 88 for Illinois). Unless otherwise specified a variable is defined for qfirst t qlast. Statistics are produced for both sexes combined, as well as separately, for all age groups, ages fourteen to eighteen, nineteen to twentyone, twenty-two to twenty-four, twenty-five to thirty-four, thirty-five to forty-four, forty-five to fifty-four, fifty-five to sixty-four, sixty-five and over, and all combinations of these age groups and sexes. An individual’s age is measured as of the last day of the quarter.

216

J. Abowd et al.

A.2.2

Individual Concepts

Flow employment: (m): for qfirst t qlast, individual i employed (matched to a job) at some time during period t at establishment j (A1) mijt

0, otherwise.

1, if i has positive earnings at establishment j during quarter t

Beginning-of-quarter employment: (b): for q first t, individual i employed at the beginning of t (and the end of t – 1), bijt

(A2)

1, if mijt1 mijt 1 0, otherwise.

End-of-quarter employment: (e): for t qlast, individual i employed at j at the end of t (and the beginning of t 1), eijt

(A3)

1, if mijt mijt 1 1

0, otherwise.

Accessions: (a1): for qfirst t, individual i acceded to j during t a1ijt

(A4)

1, if mijt1 0 and mijt 1

0, otherwise.

Separations: (s1): for t qlast, individual i separated from j during t s1ijt

(A5)

1, if mijt 1 and mijt 0

0, otherwise.

Full-quarter employment: ( f ): for qfirst t qlast, individual i was employed at j at the beginning and end of quarter t (full-quarter job) (A6)

fijt

1, if mijt1 1 and mijt 1 and mijt 1 1

0, otherwise.

New hires: (h1): for qfirst 3 t, individual i was newly hired at j during period t (A7) h1ijt

1, if mijt4 0 and mijt3 0 and mijt2 0 and mijt1 0 and mijt 1

0, otherwise.

Recalls: (r1): for qfirst 3 t, individual i was recalled from layoﬀ at j during period t

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

(A8)

r1ijt

217

1, if mijt1 0 and mijt 1 and hijt 0

0, otherwise.

Accessions to consecutive-quarter status: (a2): for qfirst t qlast, individual i transited from accession to consecutive-quarter status at j at the end of t and the beginning of t 1 (accession in t and still employed at the end of the quarter) (A9)

a2ijt

1, if a1ijt 1 and mijt 1 1

0, otherwise.

Accessions to full-quarter status: (a3): for qfirst 1 t qlast, individual i transited from consecutive-quarter to full-quarter status at j during period t (accession in t – 1 and employed for the full quarter in t) (A10)

a3ijt

1, if a2ijt1 1 and mijt 1 1

0, otherwise.

New hires to consecutive-quarter status: (h2): for qfirst 3 t qlast, individual i transited from newly hired to consecutive-quarter hired status at j at the end of t and the beginning of t 1 (hired in t and still employed at the end of the quarter) (A11)

h2ijt

1, if h1ijt 1 and mijt 1 1 . 0, otherwise

New hires to full-quarter status: (a3): for qfirst 4 t qlast, individual i transited from consecutive-quarter hired to full-quarter hired status at j during period t (hired in t – 1 and full-quarter employed in t) (A12)

h3ijt

1, if h2ijt1 1 and mijt 1 1

0, otherwise.

Recalls to consecutive-quarter status: (r2): for qfirst 3 t qlast, individual i transited from recalled to consecutive-quarter recalled status at j at the end of t and beginning of t 1 (recalled in t and still employed at the end of the quarter) (A13)

r2ijt

1, if r1ijt 1 and mijt 1 1

0, otherwise.

Recalls to full-quarter status: (r3) for qfirst 4 t qlast, individual i transited from consecutive-quarter recalled to full-quarter recalled status at j during period t (recalled in t – 1 and full-quarter employed in t)

218

(A14)

J. Abowd et al.

r3ijt

1, if r2ijt1 1 and mijt 1 1

0, otherwise.

Separations from consecutive-quarter status: (s2): for qfirst t qlast, individual i separated from j during t with consecutive-quarter status at the start of t (A15)

s2ijt

1, if s1ijt 1 and mijt1 1

0, otherwise.

Separations from full-quarter status: (s3): for qfirst 1 t qlast, individual i separated from j during t with full-quarter status during t – 1 (A16)

s3ijt

1, if s2ijt 1 and mijt2 1

0, otherwise.

Total earnings during the quarter: (w1): for qfirst t qlast, earnings of individual i at establishment j during period t (A17)

w1ijt ∑ all UI-covered earnings by i at j during t.

Earnings of end-of-period employees: (w2): for qfirst t qlast, earnings of individual i at establishment j during period t (A18)

w2ijt

w1ijt, if eijt 1

undefined, otherwise.

Earnings of full-quarter individual: (w3): for qfirst t qlast, earnings of individual i at establishment j during period t (A19)

w3ijt

w1ijt, if fijt 1

undefined, otherwise.

Total earnings at all employers: (w1•): for qfirst t qlast, total earnings of individual i during period t (A20)

w1i•t

∑

w1ijt.

j employs i during t

Total earnings at all employers for of end-of-period employees: (w2•): for qfirst t qlast, total earnings of individual i during period t (A21)

w2i•t

w1i•t, if eijt 1

undefined, otherwise.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

219

Total earnings at all employers of full-quarter employees: (w3•): for qfirst t qlast, total earnings of individual i during period t (A22)

w3i•t

w1i•t, if fijt 1

undefined, otherwise.

Change in total earnings at all employers: (w1•): for qfirst t qlast, change in total earnings of individual i between periods t – 1 and t (A23)

w1i•t w1i•t w1i•t1.

Earnings of accessions: (wa1): for qfirst t qlast, earnings of individual i at employer j during period t (A24)

wa1ijt

w1ijt, if a1ijt 1

undefined, otherwise.

Earnings of consecutive-quarter accessions: (wa2): for qfirst t qlast, earnings of individual i at employer j during period t (A25)

wa2ijt

w1ijt, if a2ijt 1

undefined, otherwise.

Earnings of full-quarter accessions: (wa3): for qfirst 1 t qlast, earnings of individual i at employer j during period t (A26)

wa3ijt

w1ijt, if a3ijt 1

undefined, otherwise.

Earnings of full-quarter new hires: (wh3): for qfirst 4 t qlast, earnings of individual i at employer j during period t (A27)

wh3ijt

w1ijt, if h3ijt 1

undefined, otherwise.

Total earnings change for accessions: (wa1): for qfirst 1 t qlast, earnings change of individual i at employer j during period t (A28)

wa1ijt

w1i•t, if a1ijt 1

undefined, otherwise.

Total earnings change for full-quarter accessions: (wa3): for qfirst 2 t qlast, earnings change of individual i at employer j during period t

220

(A29)

J. Abowd et al.

w1i•t, if a3ijt 1

undefined, otherwise.

wa3ijt

Earnings of separations from establishment: (ws1): for t qlast, earnings of individual i separated from j during t (A30)

ws1ijt

w1ijt, if s1ijt 1

undefined, otherwise.

Earnings of full-quarter separations: (ws3): for qfirst 1 t qlast, individual i separated from j during t 1 with full-quarter status during t (A31)

ws3ijt

w1ijt, if s3ijt 1 1

undefined, otherwise.

Total earnings change for separations: (ws1): for t qlast, earnings change in period t 1 of individual i separated from j during t (A32)

ws1ijt

w1i•t 1, if s1ijt 1

undefined, otherwise.

Total earnings change for full-quarter separations: (ws3): for t qlast, earnings change in period t 1 of individual i full-quarter separated from j during t, last full-quarter employment was t – 1 (A33)

ws3ijt

w1i•t 1, if s3ijt 1

undefined, otherwise.

Periods of nonemployment prior to an accession: (na): for qfirst 3 t, periods of nonemployment during the previous four quarters by i prior to an accession at establishment j during t

(A34)

naijt

∑

nits, if a1ijt 1

1s4

undefined, otherwise.

where nit 1, if mijt 0 ∀ j. Periods of nonemployment prior to a new hire: (nh): for qfirst 3 t, periods of nonemployment during the previous four quarters by i prior to a new hire at establishment j during t

(A35)

nhijt

∑

nits, if h1ijt 1

1s4

undefined, otherwise.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

221

Periods of nonemployment prior to a recall: (nr): for qfirst 3 t, periods of nonemployment during the previous four quarters by i prior to a recall at establishment j during t

(A36)

nrijt

∑

nits, if r1ijt 1

1s4

undefined, otherwise.

Periods of nonemployment following a separation: (ns): for t qlast – 3, periods of nonemployment during the next four quarters by individual i separated from establishment j during t

(A37) A.2.3

nsijt

∑

nit s, if s1ijt 1

1s4

undefined, otherwise.

Establishment Concepts

For statistic xcijt denote the sum over i during period t as xc•jt. For example, beginning-of-period employment for firm j is written as: (A38)

b•jt ∑bijt. i

All individual statistics generate establishment totals according to the formula above. The key establishment statistic is the average employment growth rate for establishment j, the components of which are defined here. Beginning-of-period employment: (number of jobs) (A39)

Bjt b•jt.

End-of-period employment: (number of jobs) (A40)

Ejt e•jt.

Employment any time during the period: (number of jobs) (A41)

Mjt m•jt.

Full-quarter employment: (A42)

Fjt f•jt.

Net job flows: (change in employment) for establishment j during period t (A43)

JFjt Ejt Bjt.

Average employment: for establishment j between periods t – 1 and t (A44)

(Bjt Ejt) jt . E 2

222

J. Abowd et al.

Average employment growth rate: for establishment j between periods t – 1 and t JFjt (A45) Gjt . E jt Job creation: for establishment j between periods t – 1 and t (A46)

JCjt E jt max (0,Gjt).

Average job creation rate: for establishment j between periods t – 1 and t JCjt (A47) JCRjt . E jt Job destruction: for establishment j between periods t – 1 and t (A48)

JDjt E jt abs (min (0,Gjt)).

Average job destruction rate: for establishment j between periods t – 1 and t JDjt (A49) JDRjt . E jt Net change in full-quarter employment: for establishment j during period t (A50)

FJFjt Fjt Fjt1.

Average full-quarter employment: for establishment j during period t (A51)

Fjt1 Fjt F jt . 2

Average full-quarter employment growth rate: for establishment j between t – 1 and t (A52)

FJFjt FGjt . F jt

Full-quarter job creations: for establishment j between t – 1 and t (A53)

FJCjt F jt max (0, FGjt).

Average full-quarter job creation rate: for establishment j between t – 1 and t FJCjt (A54) FJCRjt . F jt Full-quarter job destruction: for establishment j between t – 1 and t (A55)

FJDjt F jt abs (min (0, FGjt)).

Average full-quarter job destruction rate: for establishment j between t – 1 and t

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

(A56)

223

FJDjt FJDRjt . F jt

Accessions: for establishment j during t (A57)

Ajt a1•jt.

Average accession rate: for establishment j during t Ajt (A58) ARjt . E jt Separations: for establishment j during t (A59)

Sjt s1•jt.

Average separation rate: for establishment j during t Sjt (A60) SRjt . E jt New hires: for establishment j during t (A61)

Hjt h1•jt.

Full-quarter new hires: for establishment j during t (A62)

H3jt h3•jt.

Recalls: for establishment j during t (A63)

Rjt r1•jt.

Flow into full-quarter employment: for establishment j during t (A64)

FAjt a3•jt.

New hires into full-quarter employment: for establishment j during t (A65)

FHjt h3•jt.

Average rate of flow into full-quarter employment: for establishment j during t FAjt (A66) FARjt . F jt Flow out of full-quarter employment: for establishment j during t (A67)

FSjt s3•jt.

Average rate of flow out of full-quarter employment: for establishment j during t (A68)

FSjt FSRjt . F jt

224

J. Abowd et al.

Flow into consecutive quarter employment: for establishment j during t CAjt a2•jt.

(A69)

Flow out of consecutive quarter employment: for establishment j during t CSjt s2•jt.

(A70) Total payroll of all employees:

W1jt w1•jt.

(A71)

Total payroll of end-of-period employees: W2jt w2•jt.

(A72)

Total payroll of full-quarter employees: W3jt w3•jt.

(A73) Total payroll of accessions: (A74)

WAjt wa1•jt.

Change in total earnings for accessions: (A75)

WAjt ∑

wa1ijt.

i∈{J(i,t)j}

Total payroll of transits to consecutive-quarter status: (A76)

WCAjt wa2•jt.

Total payroll of transits to full-quarter status: (A77)

WFAjt wa3•jt.

Total payroll of new hires to full-quarter status: (A78)

WFHjt wh3•jt.

Change in total earnings for transits to full-quarter status: (A79)

WFAjt ∑

wa3ijt.

i∈{J(i,t)j}

Total periods of nonemployment for accessions: (A80)

NAjt na•jt.

Total periods of nonemployment for new hires (last four quarters): (A81)

NHjt nh•jt.

Total periods of nonemployment for recalls (last four quarters): (A82)

NRjt nr•jt.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

225

Total earnings of separations: (A83)

WSjt ws1•jt.

Total change in total earnings for separations: (A84)

WSjt ∑

ws1ijt.

i∈{J(i,t)j}

Total earnings of separations from full-quarter status (most recent full quarter): (A85)

WFSjt ws3•jt.

Total change in total earnings for full-quarter separations: (A86)

WFSjt ∑

ws3ijt.

i∈{J(i,t)j}

Total periods of nonemployment for separations: NSjt ns•jt.

(A87)

Average earnings of end-of-period employees: (A88)

W2jt ZW2jt . Ejt

Average earnings of full-quarter employees: (A89)

W3jt ZW3jt . Fjt

Average earnings of accessions: (A90)

WAjt ZWAjt . Ajt

Average change in total earnings for accessions: (A91)

WAjt ZWAjt . Ajt

Average earnings of transits to full-quarter status: (A92)

WFAjt ZWFAjt . FAjt

Average earnings of new hires to full-quarter status: (A93)

WFHjt ZWFHjt . FHjt

226

J. Abowd et al.

Average change in total earnings for transits to full-quarter status: (A94)

WFAjt ZWFAjt . FAjt

Average periods of nonemployment for accessions: (A95)

NAjt ZNAjt . Ajt

Average periods of nonemployment for new hires (last four quarters): (A96)

NHjt ZNHjt . Hjt

Average periods of nonemployment for recalls (last four quarters): (A97)

NRjt ZNRjt . Rjt

Average earnings of separations: (A98)

WSjt ZWSjt . Sjt

Average change in total earnings for separations: (A99)

WSjt ZWSjt . Sjt

Average earnings of separations from full-quarter status (most recent full quarter): (A100)

WFSjt1 ZWFSjt1 . FSjt

Average change in total earnings for full-quarter separations: (A101)

WFSjt ZWFSjt . FSjt

Average periods of nonemployment for separations: (A102)

NSjt ZNSjt . Sjt

End-of-period employment (number of workers): [Aggregate concept not related to a business] (A103)

Nt n•t.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

A.2.4

227

Identities

The identities stated below hold at the establishment level for every age group and sex subcategory. These identities are preserved in the QWI processing. Definition 1: Employment at beginning of period t equals end of period t – 1 Bjt Ejt1. Definition 2: Evolution of end-of-period employment Ejt Bjt Ajt Sjt. Definition 3: Evolution of average employment (Ajt Sjt) jt Bjt . E 2 Definition 4: Job flow identity JFjt JCjt JDjt. Definition 5: Creation-destruction identity Ejt Bjt JCjt JDjt. Definition 6: Creation-destruction/accession-separation identity Ajt Sjt JCjt JDjt. Definition 7: Evolution of full-quarter employment Fjt Fjt1 FAjt FSjt. Definition 8: Full-quarter creation-destruction identity Fjt Fjt1 FJCjt FJDjt. Definition 9: Full-quarter job flow identity FJFjt FJCjt FJDjt. Definition 10: Full-quarter creation-destruction/accession-separation identity FAjt FSjt FJCjt FJDjt. Definition 11: Employment growth rate identity Gjt JCRjt JDRjt. Definition 12: Creation-destruction/accession-separation rate identity JCRjt JDRjt ARjt SRjt.

228

J. Abowd et al.

Definition 13: Full-quarter employment growth rate identity FGjt FJCRjt FJDRjt. Definition 14: Full-quarter creation-destruction/accession-separation rate identity FJCRjt FJDRjt FARjt FSRjt. Definition 15: Total payroll identity W1jt W2jt WSjt. Definition 16: Payroll identity for consecutive-quarter employees W2jt W1jt WCAjt WSjt. Definition 17: Full-quarter payroll identity W3jt W2jt WCAjt. Definition 18: New hires/recalls identity Ajt Hjt Rjt. Definition 19: Periods of nonemployment identity NAjt NHjt NRjt. Definition 20: Worker-jobs in period t are the sum of accessions and beginning of period employment Mjt Ajt Bjt. Definition 21: Worker-jobs in period t are the sum of accessions to consecutive quarter status, separations, and full-quarter workers Mjt CAjt Sjt Fjt. Definition 22: Consecutive quarter accessions in period t – 1 are the sum of consecutive quarter separations in period t and full quarter accessions in period t CAjt1 CSjt FAjt FSjt. A.2.5

Aggregation of Job Flows

The aggregation of job flows is performed using growth rates to facilitate confidentiality protection. The rate of growth JF for establishment j during period t is estimated by: (A104)

JFjt Gjt . E jt

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

229

For an arbitrary aggregate k (ownership state substate-geography industry age group sex) cell, we have:

∑ j∈{K( j)k} Ejt Gjt Gkt . E kt

(A105)

where the function K( j ) indicates the classification associated with firm j. We calculate the aggregate net job flow as JFkt

(A106)

∑

JFjt.

j∈{K( j)k}

Substitution yields JFkt ∑ (E jt Gjt) Gkt E kt

(A107)

j

so the aggregate job flow, as computed, is equivalent to the aggregate growth rate times aggregate employment. Gross job creation/destruction aggregates are formed from the job creation and destruction rates by analogous formulas substituting JC or JD, as appropriate, for JF (Davis et al. 1996, p. 189 for details). A.2.6

Measurement of Employment Churning

The QWI measure employment churning (also called turnover) using the ratio formula: (A108)

(FAkt FSkt)/2 FTkt Fkt

for an arbitrary aggregate k (ownership state substate-geography industry age group sex) cell. In the actual production of the QWI, the three components of this ratio are computed as separate estimates and are released.

References Abowd, J. M., J. C. Haltiwanger, and J. I. Lane. 2004. Integrated longitudinal employee-employer data for the United States. American Economic Review 94 (2): 224–29. Abowd, J. M., P. A. Lengermann, and K. L. McKinney. 2002. The measurement of human capital in the U.S. economy. Technical Paper TP-2002-09. Longitudinal Employer-Household Dynamics (LEHD), U.S. Census Bureau. Abowd, J. M., B. E. Stephens, and L. Vilhuber. 2006. Confidentiality protection in the Census Bureau’s Quarterly Workforce Indicators. Technical Paper TP-200602. Longitudinal Employer-Household Dynamics (LEHD), U.S. Census Bureau.

230

J. Abowd et al.

Abowd, J. M. and L. Vilhuber. 2005. The sensitivity of economic statistics to coding errors in personal identifiers. Journal of Business and Economic Statistics 23 (2): 133–52. Benedetto, G., J. Haltiwanger, J. Lane, and K. McKinney. 2007. Using worker flows in the analysis of the firm. Journal of Business and Economic Statistics 25 (3): 299–313. Bureau of Labor Statistics. 1997a. BLS Handbook of Methods. U.S. Bureau of Labor Statistics, Division of Information Services, Washington D.C. Available at http://www.bls.gov/opub/hom/ ———. 1997b. Quality improvement project: Unemployment insurance wage records. Report of the U.S. Department of Labor. Chiang, H., K. Sandusky, and L. Vilhuber. 2005. Longitudinal EmployerHousehold Dynamics (LEHD) Business Register Bridge technical documentation. Internal Document IP-LEHD-BRB. LEHD, U.S. Census Bureau. Davis, S. J., J. C. Haltiwanger, and S. Schuh. 1996. Job creation and destruction. Cambridge, MA: The MIT Press. Longitudinal Employer-Household Dynamics Program. 2002. The Longitudinal Employer-Household Dynamics program: Employment Dynamics Estimates Project version 2.2 and 2.3. Technical Paper TP-2002-05-rev1. LEHD, U.S. Census Bureau. Stephens, B. 2006. Firms, wage dispersion, and compensation policy: Assessment and implications. Ph.D. diss. University of Maryland, College Park, Maryland. Stevens, D. W. 2007. Employment that is not covered by state unemployment insurance laws. Technical Paper TP-2007-04. LEHD, U.S. Census Bureau.

Comment

Katharine G. Abraham

This chapter describes in some considerable detail the sources and methods used to construct the data files that underlie the new Quarterly Workforce Indicators (QWI) produced by the U.S. Census Bureau. This innovative program draws on a wide variety of data sources to produce county-level estimates of earnings, employment, and job flows, disaggregated by industry, age of worker, and sex of worker. The resulting estimates already have proven to be of considerable interest to local planners and policymakers, and it is easy to imagine additional uses for them. The chapter should be a valuable resource for users of the QWI data as well as for researchers who may be interested in working with the underlying data files. Unavoidably, given the ambitious nature of the exercise undertaken and the limitations of the underlying source data, development of the QWI has confronted a variety of data problems. The QWI files draw heavily on administrative records—including unemployment insurance (UI) wage Katharine G. Abraham is a professor in the Joint Program in Survey Methodology and a faculty associate of the Maryland Population Research Center at the University of Maryland, and a research associate of the National Bureau of Economic Research.

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

231

records, employer reports to state employment security agencies, and the Census Bureau’s Person Characteristics File based on the Social Security Administration’s Numident file—which were not developed for statistical purposes. Other information is drawn from large national surveys that have better statistical properties but cover only a fraction of the population. Much of the chapter is devoted to explaining the methods currently used to address the various shortcomings of the underlying source data, as well as improvements in those methods planned for the future. My comments review briefly some key issues that the QWI developers have had to confront. • Miscoding of individual identifiers. If not corrected, miscoding of individual identifiers will lead to overstatement of worker flows, misrepresentation of workers’ earnings trajectories, and misstatement of the earnings of both departing and newly hired workers. A 1997 study of UI wage records conducted by the Bureau of Labor Statistics found that approximately 7.8 percent of individual Social Security numbers were miscoded (U.S. Bureau of Labor Statistics 1997). Abowd and Vilhuber (2005) describe a clever automated method for identifying and correcting miscodes that may occur in the middle of an ongoing spell of employment, but this method cannot capture coding mistakes that are caught by reporters and permanently corrected, or coding mistakes that are never caught. By design, the Abowd and Vilhuber procedure is conservative, producing recodes for only about 0.5 percent of wage records. While they are somewhat dated, the larger figures from the BLS study suggest that there may be a substantial amount of miscoding in individual identifiers that the Abowd and Vilhuber procedure does not capture. Further research will be needed to determine the severity of individual identifier miscoding, and what it implies for various potential uses of the QWI and associated data files. • Failure to identify continuing firms or establishments with new identification numbers. Similar to the problems associated with miscoding of individual identifiers, treating continuing establishments as new businesses leads to overstatement of business births and business deaths, as well as to overstatement of worker flows. This is perhaps the moststudied of all of the various potential problems with the QWI source data, and the techniques employed to identify establishment matches in business register data have improved a great deal over the past ten years. A clever recent innovation pioneered in the course of developing the QWI is the use of information on flows of groups of workers across establishments to identify cases in which a firm that appears to be a new birth is really a reincarnation of an old firm. While there undoubtedly are remaining cases in which continuing businesses are not

232

J. Abowd et al.

identified as such, this has to be a less serious problem than it would have been even a few years ago. • Missing information on individual characteristics. In the QWI, missing information on individual characteristics is filled in using multiple imputation techniques. Information on individuals’ age and gender is derived from Social Security records and is missing for just 3 percent of QWI records. Place of residence is missing for about 10 percent of records. The only individual-level information on education presently available for use in building the QWI files is that derived from the Survey of Income and Program Participation (SIPP) and Current Population Survey (CPS), meaning that education is missing and must be imputed for most records. This is done based on the relationship of education to age, earnings and industry in the 1990 Census. The very high rates of imputation for education cannot help but make users of these data uneasy. The planned incorporation of direct information on education for the approximately one-sixth of the population that completed the 2000 Census Long Form will be a positive step, but the share of people for whom education must be imputed will remain large. • Missing information on employer characteristics. Employer-provided information contained in the business register files is used to assign NAICS codes and a geographic location to establishments, as well as to characterize the structure of the firms to which these establishments belong. Though specific percentages are not cited, a significant number of imputations must be performed to produce a complete data file (see Konigsberg et al. 2005, for a discussion of allocations and imputations in the Quarterly Census of Employment and Wages based on the same employer characteristic source data as the QWI). The best imputations likely are those that can be based on records for the same establishment from other time periods; such information, however, is not always available. As with the data for individuals, the use of imputed information on employer characteristics may be a problem for analytical uses of the data. • Missing information on the specific establishment in which each worker is employed. When a firm consists of just one physical establishment, there is no diﬃculty in determining where a person employed by that firm works; in cases where the firm has more than one establishment, however, the assignment of individual workers to specific establishments generally is not reported. Only in Minnesota do the UI wage records indicate which establishment of a multiple-establishment firm employs which workers. As described in the chapter, the data for Minnesota are used to develop a model for probabilistically assigning workers to specific worksites within their firm that is then applied to the information available for other states. Whether a model fit using

LEHD Infrastructure Files and Creating Quarterly Workforce Indicators

233

Minnesota data can reasonably be applied to other locations is, of course, very much an open question. One of the most intriguing uses of the QWI data files is to analyze the geography of economic development, looking, for example, at where people live, where they work, and the patterns of travel between those locations. Errors in the assignment of workers to establishments could be especially problematic for this sort of analysis. In addition to these data quality issues, the chapter also notes current limitations in the scope of the QWI data set. Two in particular seem important. First, it is not presently possible to track workers who move from one state to another. Second, the self-employed are presently excluded from the QWI universe. Depending on the question one was interested in answering, both of these exclusions could be substantively important. If, for example, significant numbers of displaced workers move into selfemployment, using the QWI data to study the earnings consequences of job loss could produce misleading conclusions. The chapter indicates that work is underway to address these current limitations of the QWI. A final point to note is that noise is added to the QWI records to protect the confidentiality of the underlying information. The designers of the process used to fuzz the QWI data pay attention to preserving their statistical properties, and the chapter suggests that the analytic validity of the files should not be adversely aﬀected. This can be asserted confidently, however, only with respect to the examination of relationships that were anticipated in the design of the fuzzing process. The preceding comments are in no way intended to be critical of the authors or to disparage the work that has been done to produce the Quarterly Workforce Indicators. As a practical matter, there is no real alternative to the use of administrative statistics to produce local labor market information at the level of detail contained in the QWI. Further, though they are sometimes discussed in a way that suggests they can be taken as truth, survey data also suﬀer from a variety of sampling and nonsampling errors. These are seldom as well documented as the potential errors in the QWI described in the chapter, but that does not mean they do not exist. Still, it is important to recognize and remember that a good deal of the information that underlies the Quarterly Workforce Indicators is imputed rather than measured directly. In some cases, this will not matter very much; in other cases, the use of imputed data could lead to results that are misleading. Given the complexity of the process used to construct the indicators, it is rather diﬃcult to know what degree of confidence to place in the picture they paint. Documenting the methods used to construct the data is an important first step and one the authors are to be commended for having taken. Further work will be required to develop a fuller under-

234

J. Abowd et al.

standing of the quality properties of the QWI estimates and data files, and of their suitability for diﬀerent analytic purposes. References Abowd, J. M., and L. Vilhuber. 2005. The sensitivity of economic statistics to coding errors in personal identifiers. Journal of Business and Economic Statistics 23 (2): 133–52. Konigsberg, S., M. Piazza, D. Talon, and R. Clayton. 2005. Quarterly Census of Employment and Wages (QCEW) Business Register Metrics. Paper presented at the Joint Statistical Meetings, August. Minneapolis, Minnesota. U.S. Bureau of Labor Statistics. 1997. Quality improvement project: Unemployment insurance wage records. Unpublished report, Washington, D.C.

III

Sector Studies of Producer Turnover

6 The Role of Retail Chains National, Regional, and Industry Results Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

6.1 Introduction The U.S. retail trade sector has undergone dramatic change in recent decades. The share of U.S. civilian employment associated with retail trade has increased from 12.6 percent in 1958 to 16.4 percent in 2000, and retail employment has more than doubled. In addition to this growth, the sector has been aﬀected in important ways by changes in technology and societal trends such as suburbanization and changes in consumer preferences. The structure of retail markets, aﬀected by all these forces, has been continuously evolving. A major feature of this evolution has been the growth of large national retail chains. This has been coupled with a dramatic decrease in the share of retail activity accounted for by small single location or mom-and-pop stores. In 1948, single location retail firms accounted for 70.4 percent of retail sales, but only 60.2 percent by 1967 (U.S. Census Bureau 1971). By 1997, this share had fallen further to 39 percent. In 1948, large retail firms with more than 100 establishments accounted for 12.3 percent of retail sales, but this number grew to 18.6 percent in 1967 (U.S.

Ronald S. Jarmin is chief economist and chief, Center for Economic Studies (CES) at the U.S. Census Bureau. Shawn D. Klimek is a senior economist at the Center for Economic Studies (CES) at the U.S. Census Bureau. Javier Miranda is an economist at the Center for Economic Studies (CES) at the U.S. Census Bureau. This chapter was written by Census Bureau staﬀ. It has undergone a more limited review than oﬃcial Census Bureau publications. Any views, findings, or opinions expressed in this chapter are those of the authors and do not necessarily reflect those of the Census Bureau. We would like to thank Jeﬀ Campbell, Tim Dunne, Mark Roberts, Brad Jensen, Emek Basker, and participants at the NBER-Conference on Research in Income and Wealth (CRIW) and the Business Data Linking Conference in Cardiﬀ, Wales for useful comments. Any remaining errors are solely the responsibility of the authors.

237

238

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

Census Bureau 1971). By 1997, these large retail firms account for 36.9 percent of all retail sales. Many observers have noted the dramatic changes in the structure of retail markets. Among the more important changes is the rise of big box national retail chains, such as Wal-Mart. However, the figures cited above indicate that the trend away from mom-and-pops towards national chains has been underway since long before the advent of the big box stores. The trend also predates the wide scale adoption of information technology by retailers. Rather, the rise of technologically sophisticated national retail chains like Wal-Mart, Toys-R-Us, and Home Depot is simply part of the larger trend—underway for some time—towards larger scale retail firms. What is clear is that the dynamics of the changes during the post-World War II era in the retail sector are not well documented. This is due, in part, to a lack of comprehensive firm level longitudinal data that would allow researchers to describe and analyze the structure of retail markets. In this chapter, we use a recently constructed Census Bureau data set, the Longitudinal Business Database (LBD), to examine local retail markets over the 1976 to 2000 period. We believe these are the best data available to study trends across the entire U.S. retail sector over a long time period. These data are not perfect, however, and we discuss several remaining data gaps and measurement issues. The chapter proceeds as follows. In section 6.2 we summarize some of the trends that have characterized the retail sector in the United States over the last several decades. We discuss data and measurement issues in section 6.3. We provide some basic but informative descriptions of diﬀerent types of firms in national and regional retail markets in section 6.4 and oﬀer conclusions and discuss future research in section 6.5. 6.2 Trends in the U.S. Retail Sector Like the rest of the U.S. economy, the retail trade sector has been undergoing significant structural changes in recent decades. However, since everyone is a consumer and interacts with businesses in the retail sector regularly, these changes have not come without controversy. The trend away from smaller scale mom-and-pop retailers and towards large national chains of big box stores is often blamed in the popular media for a host of social, economic, and environmental ills. Our purpose is not to participate in this debate, but to improve the tools researchers and policymakers have at their disposal to measure changes in the structure of the retail sector and to begin to understand the forces that underlie them. 6.2.1 Basic Features of the Recent Evolution of U.S. Retail Markets To lay the groundwork for the rest of the chapter, it is useful to review, from a more macro perspective, what has been going on in the retail sector

The Role of Retail Chains: National, Regional, and Industry Results

Fig. 6.1

239

U.S. retail employment and share of the employed civilian population

Sources: Statistical Abstract of the U.S., Economic Report of the President, and own calculations from the LBD.

over the last several decades. Figure 6.1 shows the growth of U.S. retail employment from 1958 to 2000. We see that, on the Standard Industrial Classification (SIC)1 basis, retail employment grew from just under 8 million in 1958 to over 22 million in 2000. The figure also shows that the share of retail in overall U.S. employment has gone up from 12.6 percent to 16.4 percent. Retail employment saw a dramatic increase of roughly 175 percent over the 1958 to 2000 period but, as shown in figure 6.2, the number of retail establishments increased by only a modest 17 percent. It is a striking feature of the evolution of retail markets that over the last four decades of the twentieth century, the U.S. population increased by just over 100 million persons (or 56 percent), but the number of retail establishments serving them grew at a much slower rate. Figure 6.2 also shows how the composition of the increase in retail establishments is accounted for by single location establishments (mom-and-pop stores) and establishments owned by multiple location retailers (chain stores). The figure shows that the number of single location retail establishments actually decreases slightly over the 1. We use a SIC definition of the retail sector in this chapter. The Census Bureau adopted NAICS in 1997, but maintained SIC codes on its business register until 2001. Given diﬃculties in reclassifying all historic retail establishment data in the LBD on the North American Industry Classification System (NAICS) basis (see Bayard and Klimek 2004), we decided to use SIC definitions.

240

Fig. 6.2

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

Number of retail establishments

Sources: Statistical abstract of the U.S. and own calculations from the LBD.

period while the number of chain store locations more than doubles. Retail establishments operated by multiple location chain retail firms accounted for 20.2 percent of all retail establishments in 1963 and increased to 35 percent by 2000. The ascendancy of chain stores is clearly one of the most important developments in the evolution of retail markets in the United States. Chain stores diﬀer in many ways from the single location mom-and-pop stores that once dominated retail. This has always been the case, but it has become more important over time. Figure 6.3 shows that until around 1980 single location retailers and chains had roughly equal shares of overall retail sector employment. Since 1980, the chain store share of employment has increased to almost two-thirds of total retail employment. Contrast this with figure 6.2, which shows that chain stores make up a relatively constant one-third of all retail establishments. Between 1976 and 2000, employment at single location retailers grew by roughly 2 million workers. Employment growth at the smaller number of chain store retailers, on the other hand, was slightly under 8 million. Thus, we see that all the growth in the number of retail outlets and most of the growth in retail employment has come from retail firms that operate multiple retail establishments. An obvious consequence of the faster growth of retail employment compared to retail establishments is that the average size of retail establishments has grown substantially over time. Figure 6.4 shows that the size of the average retail establishment has more than doubled between 1958 and 2000. Retail customers today are not shopping at the same kind of stores that existed forty years ago. They are far more likely to be patronizing large chain stores. Even the nature of the small single location, mom-and-pop

The Role of Retail Chains: National, Regional, and Industry Results

Fig. 6.3

241

Retail employment at single location and chain stores

Source: Own calculations from the LBD.

Fig. 6.4

Average retail establishment size

Source: Statistical Abstract of the U.S. and own calculations from the LBD.

stores has changed. In results discussed further in section 6.4, we see that single location retail firms have on average increased in size since 1976. This may be due to technological changes that increase optimal store sizes, or competitive pressure exerted by the growth of large chain retailers. 6.2.2 Analyses of the Evolution of Retail Markets Researchers have developed both theoretical and empirical models that attempt to explain many of the features of retail markets. However, researchers have been hampered by a lack of detailed and comprehensive data to produce a set of stylized facts about the structure of the retail sec-

242

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

tor. We hope that data sets such as the LBD will provide the tools researchers need to make more progress. The feature of retail markets that attracts the most attention in the academic literature is the emergence of dominant chain firms. Bagwell, Ramey, and Spulber (1997) show how firms can come to dominate retail markets through large investments in cost reduction and vigorous price competition. Holmes (2001) explains how investments in information technology can lead to lower inventories, more frequent deliveries, and larger store sizes. Doms, Jarmin, and Klimek (2004) estimate the impact of investments in information technology on retail firm performance. They find that large firms account for nearly all the investment in IT in the retail sector and that IT improves the productivity of large firms more than it does for small firms. However, as shown in the previous section, modern retail markets are marked by the simultaneous presence of large chain stores and small momand-pops. While the relative importance of the two classes of retailers has changed significantly over time, the chains have not yet driven out all the mom-and-pops. Dinlersoz (2004) and Ellickson (2005) have models that explain the simultaneous presence of dominant and fringe retailers. Basically, they view retail markets as segmented between large chain firms that invest in sunk costs, such as advertising, and small mom-and-pops that do not, but instead oﬀer other retail attributes, such as better customer service. These models predict that the number of chains operating in retail markets increases less than proportionately to increases in market size, and that the number of single location mom-and-pops grows roughly proportionately. Put diﬀerently, the average size of chain stores grows with market size and the average size of mom-and-pops does not. Also, Campbell and Hopenhayn (2005) show that models where margins decline with additional entry can explain observed market structures where the number of retailers decline with market size. Several observers have noted the important link between structural change in the retail sector and productivity growth. Sieling, Friedman, and Dumas (2001) and McKinsey (2002) both note that competitive pressure from technology-intensive chain stores such as Wal-Mart leads to productivity growth in the sector both by displacing less eﬃcient retailers and by stimulating productivity improvement at surviving retail firms. Foster, Haltiwanger, and Krizan (2006) use economic census data to decompose changes in aggregate productivity. They show that net entry accounts for nearly all the productivity growth in the retail sector. The entry of establishments owned by chains is especially important as they are typically more productive than even the surviving incumbents. In a detailed analysis of the displacement of existing establishments induced by the entry of a Wal-Mart, Basker (2005) shows that in the short run, Wal-Mart entry boosts county retail employment by several hundred.

The Role of Retail Chains: National, Regional, and Industry Results

243

She uses a data set of the entry of Wal-Marts into counties, and uses publicly available County Business Patterns (CBP) data to examine the ex post change in the employment and number of producers. Although the short run impact is positive, county retail employment eventually falls as smaller retailers exit the market. The end result is that retail employment is actually larger (by about fifty jobs) than it was prior to Wal-Mart entering the county, while the number of establishments falls. However, she also finds an adverse aﬀect on the wholesale sector, which loses about twenty jobs. Many of the empirical findings for retail are limited by the quality of available data. Campbell and Hopenhayn (2005) and Basker (2005), for example, both use publicly available CBP data. These data are annual with a long time series, but cannot be used to measure the dynamics other than the net entry of establishments and firms. Other studies are limited to particular states or industries. Most do not have the industry coverage and detailed geography to describe changes in local retail markets. The goal of this chapter is to use the rich establishment-level microdata contained in the LBD to construct a set of stylized facts about the dynamics of the retail sector. The data allow us to examine results for the national and county markets, diﬀerent categories of firms, establishments and firms, urban and rural counties, across diﬀerent industries for the universe of retailers with paid employees. Even though much of our analysis does not use microdata, most of our measures could not be constructed without it. 6.3 Data and Measurement Issues The discussion in the previous section helps us consider the data requirements for analyzing the dynamic structure of retail markets. The concept of producer dynamics as described in economics textbooks is pretty straightforward. Producer dynamics capture the entry and exit of sellers in an abstract market for a good or service. Theoretical models describing the behavior of buyers and sellers in various market settings show that the structure (e.g., the number and/or size distribution of sellers) and the presence (or absence) of barriers to entry (e.g., sunk costs) are important factors in determining how eﬃciently markets operate. In this context it is critical we understand what defines a market. The theoretical literature abstracts away from the definition of a market, but this definition is at the very heart of empirical work. Empirical analyses of markets ideally require data at the firm-product level where product refers to some bundle of characteristics that would include price, location, and other product characteristics. However, such detailed data are rarely available. Thus, most empirical analyses of producer dynamics do not precisely measure the concepts that are so important for understanding market outcomes. The detailed geographic codes and firm ownership information in the LBD allows us to consider some of these issues.

244

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

6.3.1 Using the Longitudinal Business Database to Study the Evolution of Retail Markets The Census Bureau’s Longitudinal Business Database (LBD)2 is being developed by CES as part of its mandate to construct, maintain, and use longitudinal research data sets. While falling short of the ideal dataset, several unique features of the LBD make it a powerful tool for studying producer dynamics and the evolution of retail markets. These include: • Establishment (store) level data for the universe of retailers with paid employees • Information for each establishment on the following: • Longitudinal linkages • Firm aﬃliation (i.e., firm structure and ownership changes) • Location • Year of birth (provides age for continuers) • Year of death • Detailed industry codes (SIC and/or NAICS) • Size (based on payroll and employment) • The LBD can be linked to Economic Census and survey data at the establishment and firm levels to provide more detailed data on inputs and outputs not available from administrative sources. • Long time series These features allow researchers to flexibly define markets and track changes in their structure over time. Linked to data on demand conditions and other unique features of particular markets, the LBD can be an extremely useful tool to researchers interested in producer dynamics. Following we discuss how we use these features of the LBD to examine the evolution of retail markets. We also point out remaining data gaps and measurement issues. First we provide a brief description of the basic features of the LBD. The LBD is based on the Census Bureau’s Business Register (BR)3 and contains longitudinally linked establishment data for all sectors of the economy. Currently, it covers the period between 1975 and 2001. For this chapter, the main advantage is that longitudinally linked data are available annually for all retail establishments in the United States. The quality of these links is critical to constructing accurate measures of establishment entry and exit, so a few additional points about its construction are useful (a detailed description can be found in Jarmin and Miranda [2002]). 2. The LBD contains confidential data under Titles 13 and 26 United States Code (U.S.C.). However, it can be accessed by researchers with approved projects at Census Bureau Research Data Center (RDC). Information on accessing these and other confidential Census Bureau microdata can be found at www.ces.census.gov 3. Formerly known as the Standard Statistical Establishment List (SSEL).

The Role of Retail Chains: National, Regional, and Industry Results

245

The LBD is created by linking annual snapshots of the BR files. For this purpose the BR contains a number of numeric establishment and firm identifiers that can be used to track establishments over time. In particular, the Permanent Plant Number (PPN) was introduced in 1981 to facilitate longitudinal analysis. It is the only numeric establishment identifier on the BR that remains fixed as long as the establishment remains in business at the same location. However, research using the Longitudinal Research Database (LRD)—a manufacturing sector precursor to the LBD—showed that there are breaks in PPN linkages leading to spurious establishment births and deaths. Other numeric identifiers can change over time with various changes in the status of an establishment (e.g., ownership changes). For these reasons, name and address matching was used to augment the numeric identifiers to create the longitudinal linkages for the LBD. Successive years of the BR were first linked using numeric identifiers. The matches (i.e., numerically identified continuers) were set aside and the residuals were submitted to name and address matching using sophisticated statistical record linkage software. The improved establishment level identifier allows us to create the most accurate measures of establishment entry and exit for any Census Bureau data set. Establishment and firm identifiers in the LBD combined with precise location information allow us to examine the entry and exit behavior of firms and establishments within specific geographic markets. The length and frequency are especially useful for these purposes, particularly for a sector as dynamic as retail trade. No other data source provides annual coverage for the universe of employer establishments and firms for as long a period as the LBD. Other data sources share some, but not all, of these characteristics. For example, the Census of Retail Trade also covers the universe of establishments, but only occurs every five years. This implies that entry and exit of retail establishments and firms between Census years would be missed. The Annual and Monthly Surveys of Retail Trade occur more frequently, allowing the measurement of changes at the annual or even monthly level, but these data only collect information from a relatively small sample of firms. This means that we no longer have universal coverage of the sector, and the entry and exit of nonsampled firms would be missed. The Bureau of Labor Statistics (BLS) also has a longitudinally linked version of their business register, but they only have information for a taxpaying unit within a state. This means that the BLS data could not be used to address questions about the role of regional or national firms, as we discuss in the following section. Finally, it is important to stress that the LBD gives us the ability to match establishments with their parent firm. This allows us to analyze both establishment-and firm-based measures of market structure. The relationship between the two measures is not obvious. On the one hand, firm dynamics omit relevant information regarding the entry and exit of estab-

246

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

lishments, as firms already producing in the market expand the number of establishments in the market. This information is vital to understanding how firms expand their operations. On the other hand, establishment dynamics will miss vital information on the ownership and control of establishments, which may be an important determinant of establishment behavior. Given the very diﬀerent nature of these alternative measures and the implications on aggregate statistics, we compute statistics for both establishments and firms. 6.3.2 Measurement Strategy and Issues The ability to identify retail firms and to locate them in specific geographic markets is critical to our study. Firms are not homogeneous entities; some firms are large, have more resources, and may have experience in multiple markets. These diﬀerences are likely to drive diﬀerences in firm behavior and outcomes. Along these lines, there has been much popular attention regarding the displacement of small mom-and-pop stores by large national chains. In this section, we describe measurement issues related to our identification of firms and the markets in which they operate. We use the information in the LBD to identify and distinguish between four types of retail firms in much of the analysis that follows. Our classification is based on the number of states a firm operates in similar to Foster, Haltiwanger, and Krizan (2006). First, single store retailers are defined as one type, which we also consider to be representative of mom-and-pop stores. Second, we classify multi-unit firms into three types of chain firms: local, regional, and national. A firm is a local chain if it operates multiple establishments in only one state. A firm is a regional chain if it operates in at least two states but no more than ten states. Finally, a firm is a national chain if it operates in more than ten states.4 We use detailed information in the LBD to analyze the changes taking place in small geographic areas. This apparently simple task presents us with several challenges. Ideally we would like to define markets based on some measure of the geographic clustering of retailers and the population that they serve. However, county is the smallest reliable geographic unit of analysis that is available in the LBD. Coding to finer levels is less of a priority for the Census Bureau since few economic statistics are published for geographic units smaller than the county level, and as a result these measures are not as reliable.5 With these constraints we define local markets 4. We also explored an alternate definition using a measure of distance for all establishments within a firm. We find that this measure does diﬀer somewhat from a number of states based definition. We decide to stay with the literature. 5. Depending on the availability and quality of a physical street address, the Census Bureau can, and does, assign more detailed geography codes. Depending on the year, between 60 and 75 percent of establishments have Census Block and Tract codes. In Jarmin and Miranda (forthcoming), we have assigned many of these establishments latitude and longitude coordinates.

The Role of Retail Chains: National, Regional, and Industry Results

247

based on the administrative definition of a county. This has both advantages and disadvantages. On the one hand, defining local markets in this fashion is clearly arbitrary. A local retail market can encompass multiple counties, particularly in metropolitan areas. At the same time one county can encompass multiple local markets, as is often the case in physically large or densely populated counties. On the other hand, even though the county unit is a relatively crude way to define retail markets, an advantage is that there is a large amount of county level information (e.g., population) that researchers have available to control for market characteristics. One market characteristic that receives a lot of attention in the literature is size. We have a wide variety of options available in measuring the size of a county market. For this chapter, we use a parsimonious and accessible measure of market size. In addition, for the statistics we generate and report, we do not want individual counties to change market type over the period under study. Thus, we classify counties as metropolitan, micropolitan, or rural based on their 2000 Core Base Statistical Area (CBSA) code.6 Even at this crude level of geography we find that about 4 percent of establishments in the LBD have inconsistent county codes. Census assigns these codes every year based on their physical or mail address. As a consequence it is not unusual in our data to see establishments that border county lines switching back and forth. This is primarily an artifact of updates to the census files that map street names to counties. In our empirical analysis, we assign a unique county code to establishments observed switching county codes.7 We assign the county coded during the latest census year when possible; otherwise, we assign the modal county for the establishment. Our eventual goal is to use variation in many dimensions at the county level to control for diﬀerences in market characteristics including demographic composition, population density, tax structure, communications infrastructure, and proximity to other population centers. There are 1,083 counties classified as metropolitan areas, 682 counties classified as micropolitan areas, and 1,336 counties classified as nonmetro areas based on CBSA codes. We refer to these nonmetro areas as rural areas. We exclude from our computations the states of Alaska and Hawaii as well as outlying U.S. territories. Table 6.1 shows that most of the 2000 U.S. population of individuals and firms is located in metropolitan areas. Approximately 17.3 percent of the population of individuals and 13.5 percent of the population of establishments is located in rural or micropolitan areas. On average, rural areas are less than 7 percent the size (in popula6. Detailed information on these new geographic definitions can be found in Oﬃce of Management and Budget (2000). 7. Miranda (2001) documents that approximately 4 percent of establishments show changes in county codes.

248

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

Table 6.1

Metro Micro Rural Metro Micro Rural

U.S. Retail markets by CBSA and rural areas in 2000 Counties

Population

1,083 682 1,336

229,783,293 29,023,781 19,229,414 212,173 42,619 14,404

Firms

Establishments

Employment

Payroll

Totals by market type 961,264 1,223,079 159,969 176,701 120,242 129,161

18,660,642 2,187,425 1,256,810

319,571,179 31,296,137 17,625,669

Averages by market type 888 1,129 235 259 90 96

17,231 3,212 939

295,080 45,956 13,163

Source: Own calculations from the LBD. Note: This number represents the number of firms operating in a CBSA type. Chain firms can operate in counties of, potentially, all three types. Thus, there is double counting of firms in the table. The number of retail firms operating in the U.S. in 2000 was 1,066,510.

tion terms) of metropolitan areas. The average micropolitan area is about 20 percent the size of the average metropolitan area. The decision to open (or close) an establishment in a particular market is made at the firm level. In this sense, the ability to identify firm dynamics in small geographic areas is critical for understanding firm behavior as well as their response to market changes. The detailed establishment-level data in the LBD allow us to identify when a firm first enters a county, when it exits a county, and whether it has a presence in other county markets. We can also identify firm expansions or contractions in a particular market, and whether it does so by adjusting employment at existing establishments or by adjusting the number of establishments. Note that as a result of our focus on local markets, a firm can be an entrant simultaneously into multiple markets and also account for one or more market exits in diﬀerent locations. Similarly, an establishment entry is not necessarily a firm entry if the firm was already present in that market. Finally, the closure of an establishment does not necessarily generate a firm exit if the firm remains operational in the county. In the chapter we restrict our analysis to retail firms. The quality of the industry codes available on the LBD is critical to the construction of a retail sector micro data set. New establishments, especially those that begin operations between census years (i.e., those ending in two or seven) often have missing or poor quality industry codes. Between 1 percent and 10 percent of records have missing codes in the BR depending on the year and whether it is a single-unit or multi-unit establishment. Valid and improved codes are eventually obtained from direct Census Bureau collections or other sources and incorporated into the BR. These clean-up activities are concentrated in particular years, usually in preparation for an economic

The Role of Retail Chains: National, Regional, and Industry Results

249

census. To maximize the quality of industry codes on the LBD, we choose the best code available for each establishment and take advantage of codes obtained from various sources and at diﬀerent times. In particular, we use census or survey collected data whenever possible, but we may use an administrative code if no other data is available.8 Industry codes are subject to change for particular establishments over time. This occurs for about 4.5 percent of the establishments classified as retail at some point in their operational existence. There are two possible reasons for this. First, establishment may legitimately decide to change its type of activity. Second, errors in the data are possible. We address both issues by assigning each establishment in our data a unique two-digit SIC that remains fixed over the establishment’s entire history. When possible, we use industry codes collected in surveys or the economic census for the unique SIC. Alternatively, we assign the unique SIC using the most recent SIC available on the file. A current limitation of the LBD is that it is based primarily on a SIC basis. From 1976 to 1996, the SIC industry codes were the basis for all Census Bureau publications. From 1997 onward, data have been published on a NAICS basis. The Census Bureau continued to maintain SIC industry codes on the BR through 2001. Since 2002, the Census Bureau maintains only NAICS industry codes on the BR, resulting in a potential time series break in the LBD data. In addition, it is possible that the quality of the SIC codes declined between 1998 and 2001. The LBD contains information on two important measures of establishment size, payroll and employment. Revenue information contained in the BR is not currently on the LBD since it is only available for single-unit firms and at the employer identification number (EIN) level for multi-unit firms. While payroll and employment are clearly two important measures of economic activity at the establishment, they only measure inputs to retail production. Success or failure of an establishment or firm should depend on profits. This means that researchers wishing to use detailed data on establishment and firm profits (or productivity) must rely on Census Bureau censuses and surveys. Finally, the LBD covers a relatively long period. It extends back to 1975, and covers the recessions of the early 1980s and early 1990s and spans a period of significant technical change and innovation. However, this may not be long enough to actually witness much of the structural change in the retail sector. As figure 6.3 shows, employment by chain stores surpassed that of single-establishment firms in 1977. It is likely that in order to observe the long run changes in the retail sector we would need a data set that extended 8. Industry codes are obtained from multiple sources and these can change depending on the year. The most reliable code is obtained from survey forms in Census years. Other sources include administrative data from the Internal Revenue Service (IRS), Social Security Administration (SSA), and the Bureau of Labor Statistics (BLS).

250

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

back to the 1940s or 1950s, when we would expect to find relatively few chain stores and the dominance of mom-and-pop stores. As we show in the following section, diﬀerent types of geographic markets might be at diﬀerent stages in this process, and we focus on the long run diﬀerences from 1976 to 2000. 6.4 Results In an average year, there are over 1.4 million retail establishments associated with over 1 million firms. The database used in this section consists of all retail establishments from 1976 to 2000. Data elements available for the period include industry, geography, payroll, and employment. In 2000, these firms employed more than 22 million workers and generated over $368.5 billion in payroll. The section is organized in the following manner. First, we examine the trends in the national market for our four types of firms: mom-and-pops, and local, regional, and national chains. Next, we look at similar patterns, but disaggregated by the three types of county markets: metropolitan, micropolitan, and rural. Finally, we summarize the results at the two-digit SIC industry level. 6.4.1 National Market, by Type of Firm In this subsection, we analyze some basic trends in the structure of retail markets averaged across all county markets. We first look at trends in the number and size of retail establishments (i.e., stores) by retail firm type. We then look at the basic establishment entry and exit statistics, also by retail firm type. Basic Results on Retail Market Structure: Trends in the Number of Size of Retail Establishments Figure 6.5 shows the mean number of retail establishments per 1,000 county residents over the 1976 to 2000 period broken out by the four types of firms. Overall, the mean number of retail establishments drops from 7.44 to 5.88 establishments for all counties. The only type of firm that experiences a decline in the number of establishments per capita over the period is the mom-and-pops. The number of mom-and-pop stores falls from 6.2 to 4.25 stores (or 31.4 percent) during this period. All three types of chains see the number of establishments increase during this period. Overall, chain stores increase from 1.32 to 1.76 establishments, or a 36.6 percent increase. On average, the composition of firm types in these markets is shifting from mom-and-pops to chains. Figure 6.6 combines the number of establishments and employment data to examine the shift in establishment size within these types of firms. We find that all types of firms grow on average, even the mom-and-pop stores. Mom-and-pops grow on average since their employment remains

The Role of Retail Chains: National, Regional, and Industry Results

Fig. 6.5

Mean number of establishments per 1,000 residents—all counties

Source: Own calculations from the LBD.

Fig. 6.6

Mean establishment size by firm type—all counties

Source: Own calculations from the LBD.

251

252

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

relatively constant, but the number of establishments on average declines during this period. However, they only grow from about five employees to about seven employees. We find that firms of all types have larger store sizes during this period, with the largest increase coming from national chains. Local chain stores increase employment from roughly nine to fifteen employees, regional chains from roughly twelve to nineteen, and national chains from roughly fifteen to twenty-five. Basic Results on Retail Market Structure: Establishment Entry and Exit The firm entry, exit, and continuer rates in tables 6.2, 6.4, and 6.5 are defined as in Dunne, Roberts, and Samuelson (1988). We define Nft–1 as the number of establishments owned by retail firms of type f in period t – 1, Xft as the number of establishments owned by firms of type f that were active in period t – 1 but are no longer active in period t, and Eft as the number of establishments owned by firms of type f that were not active in period t – 1, but are active in period t. Finally, we define Cft as the number of establishments owned by firms of type f that were active in both period t – 1 and t. Entry, exit and continuer rates are: Entry Rate:

ERft Eft/Nft1,

Exit Rate:

XRft Xft/Nft1

Continuer Rate:

CRft Cft/Nft1.

where f is in {single-unit, local chain, regional chain, national chain}. All rates are relative to the number of firms operating in the prior period, implying that XR CR 1 for each type of firm. We can also weight by employment to construct the entrant, exit, and continuer employment share.9 Table 6.2 reports these rates averaged across all years by retail firm type. The top panel shows unweighted entry, exit, and continuer rates. Recall that figure 6.5 shows a relatively large decline in the number of single-unit establishments per capita and slight increases in the number of establishments per capita for chains. The top panel of table 6.2 confirms that single location firms have higher rates of exit than entry and, thus, on average experience net exit each year. As we move to the diﬀerent types of chains, the larger the chain, the lower the rates of both entry and exit (except for a slightly higher entry rate for regional chains). The overall eﬀect is that net entry (ER-XR) is positive for all types of chains, and larger chains have higher rates of net entry. In the bottom panel of table 6.2, we present entry, exit, and continuer 9. In Dunne, Roberts, and Samuelson (1988), the entrant share of employment (ESH) is divided by the period t employment, but in this chapter we divide by period t 1 employment. The exit share of employment (XSH) is constructed the same way, dividing by the period t employment.

The Role of Retail Chains: National, Regional, and Industry Results Table 6.2

253

Establishment entry and exit rates for the U.S. retail sector (National rates averaged across all years, 1976–2000)

Unweighted Entry Rate (ER) Exit Rate (XR) Continuer Rate (CR) Weighted by employment Entrant Share (ESH) Exit Share (XSH) Continuer Share (CSH)

Single

Local

Regional

National

0.149 0.151 0.849

0.092 0.085 0.915

0.093 0.076 0.924

0.088 0.069 0.931

0.078 0.108 0.892

0.078 0.056 0.944

0.065 0.046 0.954

0.055 0.043 0.957

Source: The LBD.

rates weighted by employment. Across all firm types, entrants and exits tend to be smaller than continuing firms, thus the weighted entry and exit rates are lower than their unweighted counterparts. The results on employment-weighted shares show that the net entry of employment for chains is actually highest for local, regional, and then national chains on average during the period. 6.4.2 Results by Market and Firm Type In the previous section, we examined the national retail market; however, we have already shown that there are considerable diﬀerences across county types. In this section, we examine changes in market structure and dynamics across the three county market types and by firm type. We start by summarizing the changing nature of the distribution of the number of retail establishments and firms operating in county markets and retail employment by county type. We then look at firm entry into and exit from these county markets by county type. We focus on firm entry since the firm is the relevant decision maker in the market. Table 6.3 describes the distribution of establishments, firms, and employment per capita. It reports the mean number of establishments, firms, and employees per 1,000 county residents within each county type for both 1976 and 2000. We also report the standard deviation to provide a sense of the variation across counties within each type of county. We see a number of important diﬀerences between the three types of county markets. At the beginning of the period, rural counties have on average two more establishments per capita than do metropolitan counties, but they also have two more firms and nine fewer employees per capita. This implies that we observe a larger number (on a per capita basis) of smaller firms in rural areas. Micropolitan counties also have more establishments and firms per capita than metropolitan counties, but not as many as rural counties. In terms of employment, micropolitan and metropolitan

254

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

Table 6.3

County retail market structure: Number of establishments, firms, and employees by market type (based on per capita county level aggregates) Mean

Standard deviation

Year

Market Type

Establishments

Firms

Employment

Establishments

Firms

Employment

1976 1976 1976 2000 2000 2000

Metro Micro Rural Metro Micro Rural

6.3 7.5 8.3 5.2 6.0 6.4

5.8 7.2 8.1 4.5 5.6 6.1

47.8 48.4 38.2 70.9 71.2 52.7

1.9 2.0 2.8 1.9 2.0 2.9

1.8 1.9 2.8 1.7 1.8 2.9

25.5 18.6 18.1 37.6 28.2 28.2

Source: The LBD.

counties have roughly the same number of retail employees per capita. From 1976 to 2000, there is a significant decline in the number of establishments and firms in all types of county markets. At the same time, we observe a significant increase in the retail employment across all types of counties. Metropolitan and micropolitan counties continue to have roughly the same levels of retail employment, and rural counties are still significantly smaller. The overall eﬀect is that the average size of an establishment has grown in each type of region. Finally, we see that the variance of the establishment and firm distributions did not change over time, but that the variance of the employment distribution increased over the period from 1976 to 2000. In table 6.4, we present average firm entry, exit, and continuer rates by metropolitan, micropolitan, and rural county types. As in table 6.2, we show the annual rates averaged over the entire period of 1976 to 2000. Like the results for establishments in table 6.2, we see that single-unit firms have higher entry and exit rate across all market types. Local chains have slightly higher rates of entry and exit than do regional and national chains. Table 6.4 shows only small diﬀerences between regional and national chains. Table 6.4 reveals that average net entry rates for single-unit retailers are negative for all market types. This is similar to figure 6.5, which showed the drop in the average number of single-unit establishments per capita across all counties. In contrast, net entry rates are nonnegative for chain retailers. Firm turnover rates are computed as the sum of the entry and exit rates (ER XR). These are a measure of churning within retail markets. We see from table 6.4 that single-unit retailers experience more churning that do chain stores. More interesting perhaps is the finding that turnover rates increase with market size. Metropolitan counties, in particular, experience more turnover across all types of retail firms than do micropolitan or rural markets. The diﬀerence in retail firm turnover between metropolitan and rural county market types is 0.006, 0.017, 0.038, and 0.019 for single units,

The Role of Retail Chains: National, Regional, and Industry Results Table 6.4

Entry Rate (ER) Rural Micro Metro Exit Rate (XR) Rural Micro Metro Continuer Rate (CR) Rural Micro Metro

255

Firm entry and exit rates for the U.S. retail sector (Rates by market type averaged across all years, 1976–2000) Single

Local

Regional

National

0.143 0.144 0.151

0.085 0.087 0.094

0.077 0.082 0.097

0.077 0.077 0.089

0.153 0.150 0.151

0.078 0.077 0.087

0.061 0.065 0.079

0.064 0.063 0.070

0.847 0.850 0.849

0.922 0.923 0.913

0.939 0.935 0.921

0.936 0.937 0.930

Source: The LBD.

local, regional, and national retail chains, respectively. Thus, we see that large metropolitan retail markets are characterized by fewer competitors per capita than rural and micropolitan county markets, but that competition in metropolitan markets is marked by higher firm turnover, and that this higher turnover is more pronounced among chain store retailers. Further, our firm turnover measure may understate the degree of volatility in county markets since retail chains can change their scale of activity in county markets by opening or closing stores. Our measure does not capture when firms expand or contract the number of stores in a county, as long as they continue to operate at least one store in the county. Table 6.5 shows employment-weighted entry, exit, and continuer rates. As before, we see that entrants and deaths tend to be smaller than continuing firms as reflected by the lower weighted entry and exit rates. This result is true across market types. Also note the net gain in employment from entry and exit of retail stores across market types for all retail chains. This is not the case for mom-and-pops, which show the highest losses in metropolitan areas. 6.4.3 Industry Diﬀerences In this section, we look at diﬀerences in producer dynamics and the role of chain stores across two-digit retail industries. First, we compare the number of county markets served by the four firm types in 1977 and 2000. We are trying to understand how the role of these firm types within county retail markets has changed over time and determine if there are systematic diﬀerences in these changes across diﬀerent retail industries. The results of this exercise are reported in table 6.6. One important thing to note in table 6.6 is that many county markets are

256

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

not served by all retail firm types. Expectedly, most of the 3,101 U.S. counties (excluding Alaska and Hawaii), are served by single-unit firms in most two-digit SIC retail industries. However, the situation is quite diﬀerent when looking at the diﬀerent chain types. Indeed, it is often the case that the majority of U.S. counties are not served by one or more chain types within these broad two-digit SIC industries. From table 6.1 we know that rural counties are the dominant county market type numerically, are quite small, and may not oﬀer suﬃcient demand to justify the scale of many

Table 6.5

Employment-weighted firm entry and exit rates for the U.S. retail sector (Mean by market type, 1976–2000)

Entrant Share (ESH) Rural Micro Metro Exit Share (XSH) Rural Micro Metro Continuer Share (CSH) Rural Micro Metro

Single

Local

Regional

National

0.078 0.078 0.078

0.072 0.078 0.078

0.055 0.058 0.067

0.060 0.051 0.055

0.107 0.107 0.109

0.053 0.052 0.057

0.039 0.040 0.047

0.040 0.040 0.043

0.893 0.893 0.891

0.947 0.949 0.943

0.961 0.960 0.953

0.960 0.960 0.957

Source: The LBD.

Table 6.6

Number of county markets served by different retail firm types (1977 and 2000, by two-digit SIC) Single

SIC 52 Building Materials and Hardware 53 General Merchandise 54 Food Stores 55 Auto Dealers and Gas Stations 56 Apparel and Accessories 57 Home Furnishing and Equipment 58 Eating and Drinking Places 59 Miscellaneous Retail Source: The LBD.

Local

Regional

National

1977

2000

1977

2000

1977

2000

1977

2000

3,005 2,835 3,089

2,960 2,138 3,072

1,909 1,485 2,327

1,765 629 2,352

1,484 1,886 1,891

1,380 843 2,277

1,157 2,087 1,770

1,490 2,673 1,806

3,096

3,066

2,441

2,504

1,954

2,407

1,770

2,039

2,904

2,518

1,865

1,092

1,544

1,180

1,852

1,763

2,848

2,792

1,666

1,429

1,020

1,035

954

1,393

3,095 3,067

3,088 3,060

2,062 2,480

2,384 2,224

1,603 1,631

2,275 1,804

1,465 2,101

2,010 2,204

The Role of Retail Chains: National, Regional, and Industry Results

257

chain retailers. Nevertheless, some retailers such as Wal-Mart have declared intentions for substantial expansion of the next several years.10 It will be interesting to see whether chains will continue to expand into new markets. The changes over the period in the number of county markets served by the diﬀerent firm types are quite striking. We see that the number of counties served by at least one mom-and-pop retailer actually falls in every industry. The fall is not dramatic, but that fact that we observe a decline is surprising given the ubiquity of small retailers. On the other side, we find that the number of markets being served by a national chain is increasing for all industries and that some of the increases are dramatic. Results for local and regional chains vary across the diﬀerent industries. General Merchandise firms show a very interesting trend. As expected, given the rise of stores such as Wal-Mart and Target and the consolidation of once-regional department stores, we see that the number of county markets served by national retail chains has grown substantially over the period. This growth is accompanied by dramatic reductions in the number of markets served by single-unit, local chains, and regional chains of general merchandise firms. The trends in the number of county markets served by the various firm types diﬀer substantially across retail industries. In Eating and Drinking Places, there is only a small reduction in the number of markets served by single-unit producers and there are large increases in the number of markets served by all types of chains. Contrast that with the trends in Apparel and Accessories, where we see that the number of markets served by all firm types decreases as the industry shrinks. While changes in the number of markets served by the diﬀerent types of firms are interesting, we also focus on how entry and exit rates (establishment and firm) diﬀer across industries. We construct a more detailed data set with entry and exit rates defined within the county, year, two-digit SIC, and chain type. While more detailed industries at the six-digit level are potentially available in the LBD, we already have a significant number of industries at the two-digit level where we cannot construct an entry or exit rate (since Nft–1 0). We mitigate this problem by computing entry and exit rates as in Davis, Haltiwanger, and Schuh (1996): Efct ERfct [(Nfct Nfct1)/2] Xfct XRfct [(Nfct Nfct1)/2] 10. Wal-Mart’s 2005 annual report indicates that it plans to open 1,000 new supercenters in the United States over the next five years and Lee Scott, Wal-Mart’s CEO says there is room in the United States for 4,000 more Wal-Mart supercenters.

258

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

We summarize industry diﬀerences in entry and exit rates using a series of simple regressions. We include dummies for both firm and county market type. We also include a series of dummies for each five-year period from 1976 through 2000. The omitted group is mom-and-pop stores in rural markets during the period of 1996 to 2000. We present entry rate results for both establishments and firms in table 6.7. Looking at the intercept terms, we see that the industry with the highest establishment and firm entry rates is SIC 58, Eating and Drinking Establishments (this still holds if one uses the other coeﬃcients to calculate entry rates for chains in nonrural counties). The industry with the lowest establishment and firm entry rates is SIC 52, Building Materials and Hardware.

Table 6.7 SIC

Intercept Time period 1976–1980 1981–1985 1986–1990 1991–1995 Market type Metro Micro Firm type National chain Regional chain Local chain Intercept Time period 1976–1980 1981–1985 1986–1990 1991–1995 Market type Metro Micro Firm type National chain Regional chain Local chain

Establishment and firm entry rate regressions 52

0.083

53

54

55

56

Panel A: Establishment entry rates 0.140 0.133 0.089 0.106

57

58

59

0.106

0.154

0.114

0.032 0.023 0.007 –0.011

–0.017 –0.026 –0.013 –0.008

0.010 0.017 0.000 –0.013

0.046 0.043 0.022 0.000

0.034 0.030 0.026 0.004

0.033 0.046 0.026 0.009

0.045 0.042 0.030 0.009

0.031 0.034 0.018 0.009

0.024 0.015

0.017 0.008

0.021 0.013

0.010 0.002

0.022 0.007

0.018 –0.005

–0.003 –0.002

0.013 0.001

–0.035 –0.040 –0.047

–0.078 –0.073 –0.073

–0.082 –0.090 –0.075

–0.041 –0.022 –0.029

–0.050 –0.038 –0.063

–0.053 –0.034 –0.052

–0.071 –0.068 –0.059

–0.046 –0.038 –0.062

0.088

0.141

Panel B: Firm entry rates 0.133 0.099 0.120

0.101

0.166

0.121

0.032 0.024 0.010 –0.012

–0.016 –0.010 0.010 –0.006

0.013 0.024 0.010 –0.008

0.043 0.041 0.014 –0.006

0.024 0.020 0.019 0.002

0.039 0.048 0.026 0.007

0.045 0.041 0.028 0.007

0.030 0.037 0.021 0.009

0.026 0.014

0.009 0.005

0.028 0.009

0.011 0.004

0.020 0.008

0.031 0.007

–0.001 0.001

0.013 0.004

–0.030 –0.028 –0.059

–0.091 –0.056 –0.083

–0.074 –0.062 –0.080

–0.029 –0.021 –0.047

–0.054 –0.036 –0.078

–0.029 –0.034 –0.073

–0.072 –0.052 –0.078

–0.031 –0.033 –0.075

Source: Own calculations from the LBD. Notes: Unit of Observation is a {county, year, firm type} cell. Regressions are run by 2-digit SIC with controls for time period, market type, and firm type. All coefficients are significant at the 5 percent level.

The Role of Retail Chains: National, Regional, and Industry Results

259

The pattern of estimated time period dummies generally show that entry rates are declining over time. We observe monotonic declines in the time period dummies in several industries. Only with SIC 53, General Merchandise Stores, do we observe a lower entry rate in the initial period than we do in the final period. This finding holds for both establishment and firm entry rates. With the exception of Eating and Drinking Places, SIC 58, entry rates are highest in metropolitan markets and slightly higher in micropolitan markets. This is similar to results for the entire retail sector shown in table 6.4. We find mixed results for the chain type dummies. The negative coeﬃcients imply that the mom-and-pop stores have the largest entry rates, regardless of industry or unit of measure (establishment or firm). We find exit rate results for both establishments and firms similar to those for the entry rate. The results, presented in table 6.8, again show that SIC 58 has the highest establishment and firm exit rates and SIC 52 has the lowest. We also find that exit rates are declining over time, with the eﬀect being monotonic in about half the industries. We generally find that exit rates are highest in metropolitan markets and slightly higher in micropolitan markets than in rural markets. We find mixed results for the diﬀerent types of chains. The negative coeﬃcients imply that the mom-and-pop stores have the largest exit rates, regardless of industry or unit of measure (establishment or firm). We also find that firm exit rates are next highest for regional chains for all industries, with no pattern for local and national chains across the industries. This pattern does not hold for establishment exit rates. 6.5 Conclusion This chapter provides a rich set of stylized facts describing the evolution of U.S. retail markets over the last thirty years. We use the Longitudinal Business Database, which oﬀers a long time series of longitudinal data covering all retail establishments with paid employees. Detailed information on establishment location and firm ownership allows us to examine changes in market structure and producer dynamics, focusing on the role of retail chains. These data allow us to corroborate several important trends already described by other empirical work, as well as document some new findings. We document the steady ascendance of retail chains in terms of both their share of employment and establishments, as well as the decline of relatively small mom-and-pops. Customers shop at much larger stores today than they did thirty years ago. Interestingly, we find there are fewer establishments per 1,000 residents, but they are significantly larger. The absolute growth in the size of the national chain store is particularly striking in this regard. However, we also observe that single location mom-and-pop stores

260

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

Table 6.8 SIC

Establishment and firm exit rate regressions 52

Intercept Time period 1976–1980 1981–1985 1986–1990 1991–1995 Market type Metro Micro Firm type National chain Regional chain Local chain Intercept Time period 1976–1980 1981–1985 1986–1990 1991–1995 Market type Metro Micro Firm type National chain Regional chain Local chain

0.095

53

54

55

56

Panel A: Establishment exit rates 0.147 0.130 0.103 0.164

57

58

59

0.114

0.164

0.130

0.030 0.015 0.006 0.003

0.000 0.004 –0.005 0.003

0.028 0.016 0.014 0.005

0.088 0.046 0.028 0.004

–0.017 –0.022 –0.021 –0.003

0.031 0.019 0.002 –0.003

0.043 0.022 0.011 –0.003

0.029 0.010 0.000 0.000

0.018 0.008

0.015 0.006

0.019 0.009

0.010 0.002

0.017 0.006

0.021 0.002

–0.011 –0.006

0.005 –0.001

–0.072 –0.056 –0.057

–0.114 –0.083 – 0.072

–0.084 –0.095 –0.079

–0.067 –0.041 –0.055

–0.098 –0.062 –0.056

–0.082 –0.042 –0.050

–0.083 –0.100 –0.083

–0.073 –0.055 0.061

0.097

0.158

0.162

0.119

0.173

0.139

0.032 0.018 0.012 0.006

–0.014 0.000 0.000 –0.015

0.020 0.017 0.017 –0.004

0.083 0.044 0.017 –0.002

–0.005 –0.009 –0.012 0.005

0.037 0.021 0.006 –0.004

0.045 0.027 0.013 –0.009

0.029 0.011 0.003 –0.005

0.019 0.007

0.014 0.004

0.027 0.007

0.010 0.001

0.017 0.010

0.021 0.002

–0.007 –0.004

0.004 0.001

–0.058 –0.036 –0.061

–0.120 –0.053 –0.069

–0.089 –0.061 –0.074

–0.056 –0.035 –0.057

–0.100 –0.055 –0.067

–0.071 –0.038 –0.061

–0.081 –0.069 –0.081

–0.058 –0.036 –0.065

Panel B: Firm exit rates 0.136 0.115

Source: Own calculations from the LBD. Notes: Unit of Observation is a {county, year, firm type} cell. Regressions are run by 2-digit SIC with controls for time period, market type, and firm type. All coefficients are significant at the 5 percent level.

have grown larger over time, perhaps as a response to competitive pressures from chain stores. Our analysis by county market type shows that rural markets are still served by a relatively large number of small mom-and-pop stores. These areas are experiencing net losses of this type of store. Our regional analysis shows that there are fewer competitors in larger markets, but competition in these markets is marked by higher firm turnover across all firm types. The chapter also shows interesting diﬀerences across broad retail industries. Chain stores and mom-and-pop stores appear to be able to coexist in some industries better than others. Independent general merchandise stores and apparel and accessories store owners are disappearing from

The Role of Retail Chains: National, Regional, and Industry Results

261

many markets while independent eating and drinking places can still be found in most markets. In future work, we will delve deeper into the relationship between market size and market structure. How does the mix of ownership types change as market size changes? How does firm turnover change as market size changes? Asplund and Nocke (2006) develop a model with predictions regarding firm turnover and market size. They argue that turnover should be higher in larger markets. The LBD is ideal to look at this issue. How does firm size change in response to changes in market size? We can examine over a long period of time the relationship between establishment/firm size and how it varies across firm type.

References Asplund, M., and V. Nocke. 2006. Firm turnover in imperfectly competitive markets. Review of Economic Studies 73 (2): 295–327. Bagwell, K., G. Ramey, and D. F. Spulber. 1997. Dynamic retail price and investment competition. RAND Journal of Economics 28 (2): 207–27. Basker, E. 2005. Job creation or destruction? Labor-market eﬀects of Wal-Mart expansion. Review of Economics and Statistics 87 (1): 174–83. Bayard, K., and S. D. Klimek. 2004. Creating a historical bridge for manufacturing between the Standard Industrial System and the North American Industry Classification System. 2003 Proceedings of the American Statistical Association, Business and Economic Statistics Section (CD-ROM): 478–84. Campbell, J., and H. Hopenhayn. 2005. Market size matters. Journal of Industrial Economics 53 (1): 1–25. Davis, S. J., J. C. Haltiwanger, and S. Schuh. 1996. Job creation and destruction. Cambridge, MA: MIT Press. Dinlersoz, E. M. 2004. Firm organization and the structure of retail markets. Journal of Economics and Management Strategy 13 (2): 207–40. Doms, M. E., R. S. Jarmin, and S. D. Klimek. 2004. Information technology investment and firm performance in U.S. retail trade. Economics of Innovation and New Technology 13 (7): 595–613. Dunne, T., M. J. Roberts, and L. Samuelson. 1988. Patterns of Firm Entry and Exit in U.S. Manufacturing Industries. RAND Journal of Economics 19 (4): 495–515. Ellickson, P. B. 2005. Supermarkets as a natural oligopoly. Duke University Department of Economics. Working paper 05-04. Foster, L. S., J. Haltiwanger, and C. J. Krizan. 2006. Market selection, reallocation and restructuring in the U.S. retail trade sector in the 1990s. Review of Economics and Statistics 88 (4): 748–58. Holmes, T. J. 2001. Bar codes lead to frequent deliveries and superstores. RAND Journal of Economics 34 (4): 708–25. Jarmin, R. S., and J. Miranda. 2002. The longitudinal research database. Center for Economic Studies. CES Working paper CES-WP-02-17. ———. Forthcoming. The impact of Hurricane Katrina on business establishments. Journal of Business Valuation and Economic Loss Analysis.

262

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

McKinsey Global Institute. 2002. How IT Enables Productivity Growth MGI Reports, November. Available at www.mckinsey.com/knowledge/mgi/IT/ Miranda, J. 2001. LBD documentation: Geography. Center for Economic Studies, U.S. Census Bureau, Technical Notes CES-TN-2001-02, January. Oﬃce of Management and Budget. 2000. Standards for defining metropolitan and micropolitan statistical areas. Federal Register 65 (249): 82228–38. Sieling, M., B. Friedman, and M. Dumas. 2001. Labor productivity in the retail trade industry 1987–99. Monthly Labor Review Online 124 (12): 3–14. U.S. Bureau of the Census. 1971. Census of business, 1967; Vol. I, Retail tradesubject reports. Washington, D.C.: U.S. Government Printing Oﬃce. ———. 1994. Statistical abstract of the United States: 1994 (114th edition) Washington, D.C.: U.S. Government Printing Oﬃce.

Comment

Jeﬀrey R. Campbell

Technology introduction takes place firm-by-firm and establishment-byestablishment. Even a good idea that falls from the sky (the classic neutral technology shock) must be read and incorporated into a production plan. For this reason, the analysis of individual producers’ birth, growth, and death occupies a central place in productivity analysis. The Longitudinal Research Database provided the first observations of this process for the United States’ Manufacturing sector, and its analysis by Dunne, Roberts, and Samuelson (1988), Bartelsman and Dhrymes (1998), and others created a new appreciation of creative destruction’s contribution to productivity growth. Of course, these empirical developments would have been impossible without the contributions of Jovanovic (1982) and Hopenhayn (1992) to the theory of industry dynamics. Manufacturing led U.S. economic growth through the 1960s, but Retail Trade and Services have worn the yellow jersey since then. Further progress relating productivity growth to industry dynamics therefore requires our empirical and theoretical work to catch up to this new leading sector. Jarmin, Klimek, and Miranda have given us a substantial push in this direction. Although they are not the first to examine producer-level data from Retail Trade, they are the first (to my knowledge) to do so in light of that sector’s central economic fact: the replacement of stand-alone momand-pop stores by large chain stores with low prices. Today, Wal-Mart’s rise occupies the headlines, but regional and nation chain growth inspired the anti-chain-store movement of the 1920s and 1930s. The specific players and their tactics have changed, but the issues at hand remain the same: do

Jeﬀrey R. Campbell is a senior economist at the Federal Reserve Bank of Chicago and a faculty research fellow of the National Bureau of Economic Research. E-mail: jcampbell @frbchi.org

The Role of Retail Chains: National, Regional, and Industry Results

263

new low-cost retailers improve welfare by lowering prices or retard it by lowering wages and displacing other competitors? Previous work on the Manufacturing sector has left us ill prepared for these questions, because the dominant approach to that sector presumed some form of atomistic competition (either the price-taking perfect variety or the price-setting monopolistic variety) in which strategic interactions are absent. Evidence in Campbell and Hopenhayn (2005), Campbell (2006), and Yeap (2005) shows that atomistic competition cannot even rationalize basic features of the data like the dependence of establishment size, prices, and turnover on market size. The first step in understanding industry dynamics and productivity growth in the retail and service sectors is to confront the strategic aspects of their market interactions. Jarmin, Klimek, and Miranda contribute to this by delineating the important players in any retail market and by reporting useful stylized facts about trend rates of displacement and turnover. In this discussion, I wish to complement their contribution with a relatively simple model of dynamic retail competition with both chain stores and independents. Retail has great potential for strategic complexity. Firms operate several distinct technologies and diﬀerentiate themselves geographically. It is not diﬃcult to specify a model that embodies all of these features. However, such a model’s complexity precludes its analytic characterization. There might be one equilibrium or many, and they do not lend themselves to local comparative statics results like those Hopenhayn (1992) develops for a competitive industry. With this in mind, the model I develop vastly simplifies the spatial aspects of competition so that we can learn something about competition between dominant chain producers and a fringe of high-cost independent producers. A Model To build a model with nontrivial dynamics and strategic interaction, I draw on my previous work with Jaap Abbring (Abbring and Campbell 2006). We develop a model of Markov-perfect duopoly dynamics with stochastic demand, sunk costs of entry, and irreversible exit. Firms make their continuation decisions oldest first, and we focus on the unique equilibrium in which firms’ exits follow a last-in first-out pattern. In this discussion, I construct a symmetric equilibrium in a similar model with a fringe of monopolistically-competitive independent producers. Primitives Consider a region with a central city and a large number L of outlying villages. The city’s name is 0, and the villages names are j 1, . . . , L. The villages are arranged in a circle with the city at its center. A single road connects each village to the city. We denote the population of location j in

264

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

year t with C jt. These follow independent (across locations) Markov chains. With probability , C jt C jt–1. With the complementary probability, Cjt is a draw from a uniform distribution on [Cˆv /L, Cˇv /L] for a village and [Cˆc, Cˇc ] for the city. Consumers have identical incomes measured in money ( y) and they allocate their purchases across an outside good available everywhere at a price of 1 and the good of interest. This latter good is not necessarily available at the same price everywhere. If a consumer purchases q units of this good at a price of p in her home location, then her utility level is q

0 D

(x)dx y pq.

1

If she has to travel to make the same purchase, she must pay a transportation cost T (in units of money). For simplicity, assume that a villager may only travel to the city and ignore the possibility of a city dweller shopping in a village. Consumers’ travel costs are random. The c.d.f (x) Pr[T x] governs their distribution There are two production technologies. One has higher fixed costs and lower variable costs than the other. I refer to these as the big-box and independent technologies. Each village has one potential entrant per period. This firm must choose between entering at that location with the independent technology or remaining out of the market. This opportunity always goes to a new firm, so the decision to remain inactive is irreversible. The city has two potential entrants each period, and each of them has access to only one of the technologies. As with the villages’ potential entrants, they cannot delay their entry decisions. The sunk cost of entering with the big-box technology is b. Producing with this technology in any period after entry requires paying the fixed cost

b. The only way of avoiding this fixed cost is to exit irreversibly. This technology’s constant marginal cost of production is b. Entering with the independent technology requires no sunk cost and a per-period fixed cost of

i b. A higher marginal cost i b and a shorter life span oﬀset these advantages. A firm entering with the independent technology can produce for only one period. Relaxing this extreme assumption in future versions of this model is clearly desirable. Firms with the big-box technology discount future profits with the constant discount factor 1. The model has two physical state variables, the number of firms which produced in→the city in the previous period, N 0t , and the vector of market populations Ct (C 0t , C1t , . . . ,C Lt ). Each period, the sequence of actions proceeds as follows. 1. All potential entrants and incumbent firms (in the city) observe the → realization of Ct. 2. Any incumbent firms make their continuation decisions simultaneously.

The Role of Retail Chains: National, Regional, and Industry Results

265

3. The city’s big-box potential entrant decides whether or not to enter. 4. The city’s independent potential entrant decides whether or not to enter. 5. The villages’ potential entrants make their entry decisions simultaneously. 6. With observations of their travel costs and all firms’ entry and continuation decisions, consumers select their shopping locations. 7. After the consumers arrive at their shopping locations, firms simultaneously choose quantities. An auctioneer then sets prices to clear the locations’ markets. Equilibrium With the model’s primitives in place, we seek a Markov-perfect equilibrium. We first characterize the static parts of the model, which correspond to stages 5 through 7 in the previous list. With these solved, I build on results from Abbring and Campbell (2006) to characterize the dynamics of the big-box sector. To simplify the analysis, I proceed under two assumptions: entering as the third big-box producer and entering the city as an independent with a big-box firm committed to production are dominated strategies. With the proposed equilibrium in place, finding conditions that guarantee this will be the case is not hard. Static Play Begin with the firms’ quantity decisions. All firms’ profits are linear in the number of customers shopping at their locations, so we can consider their choices of quantity per customer. By construction, at most one firm serves each village. Its producer surplus per customer is [D–1(q) – i]q. Denote the profit-maximizing choice of q with qi∗ and the resulting per customer surplus with ∗i [D–1(q∗i ) – i]q∗i . For a firm exclusively operating the big-box technology in the city, the choice of q is similar. The resulting per customer profit and its associated quantity are ∗b(1) and q∗b(1). If there are two big-box firms operating, then their quantity decisions correspond to the standard Cournot solution. Denote the per customer duopoly profit for a firm with marginal cost w facing a rival with ∗b(2) and the per customer duopoly quantity (summed across both firms) with q∗b(2). The quantity choices and the resulting prices determine villagers’ choices of shopping locations. A resident of a village with no producer chooses to shop in the city if her utility gain from doing so exceeds her travel cost. That is, if q∗b(Nt0)

0

D1(x)dx D1[q∗b(N 0t )]q∗b(N 0t ) T.

The last term on the left-hand side is the total purchase cost of the q∗b(N 0t ) units of the good. If we call the left-hand side of this inequality Wc(N 0t ),

266

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

then the fraction of such villagers choosing to shop in the city is [Wc(N0t )]. For residents of villages with producers, purchasing from the local producer is the alternative to shopping in the city. The utility gain from shopping locally (compared with consuming the entire budget in the outside good) is Wv(N 0t )

q∗

i

0

D1(x)dx y D1(q∗i )q∗i .

Clearly, the local producer’s profit maximization guarantees that this is positive, so the fraction of consumers choosing to shop locally is 1 – [Wc(N 0t )– Wv(N 0t )]. The remaining consumers shop in the city. Given the number of firms serving the city, a village’s potential entrant rationally forecasts Wc(N t0) and consumers’ travel decisions and decides to enter only if the corresponding profit is nonnegative. That is, if C jt{1 [Wc(N 0t ) Wv (N 0t )]}∗i κi 0. Clearly, there is a threshold value of population C i(N 0t ), which sets this j profit to zero. Entry into village j is profitable if C t C i(N 0t ). Because L is large and the villages’ populations are statistically independent, we can apply a law of large numbers to show that the number of villagers traveling to the city is a nonstochastic function of only Nt0, the number of competitors in the city. In any given period, the number of residents of villages with no local producer equals 1/2{[C i (N 0t )] 2 – (Cˆv)2}/(Cˇv – Cˆv). The remaining villagers have the option of purchasing from a local producer. Putting these together, we get that the number of villagers shopping in the city equals [C i(N 0t )]2 (Cˆv)2 M(N 0t ) [Wc(N 0t )] 2(Cˇv Cˆv) i(N 0t )]2 (Cˇv)2 [C . [Wc(N 0t ) Wv(N 0t )] 2(Cˇv Cˆv) Dynamic Big-Box Competition The sunk costs of entry and incumbents’ priority in serving a market make the problem of an entrant using the big-box technology dynamic. To characterize the evolution of big-box competition, consider the dynamic game with only the big-box firms as players and payoﬀs given by the outcome of the static competition described above. I construct a very simple Markov-perfect equilibrium for this game. It is symmetric in the sense that duopolists’ continuation decisions follow the same mixed strategy. The equilibrium construction begins with the problem of a pessimistic duopolist who believes (irrationally) that the rival firm will never exit. Its current profit is [C 0t M(2)](2)/2 – b. It will earn this until the next time that C 0t changes, at which point the new demand value will be statistically

The Role of Retail Chains: National, Regional, and Industry Results

267

independent of its current value. The conjecture that the rival will never exit allows us to show that the following piecewise-linear function of C t0 gives this duopolist’s value.

[C t0 M(2)](2) 1 b v˜(2) 2 v(C 0t , 2) 1 (1 )

0

if C 0t C2 otherwise.

Here, C 2 is the largest value of C that sets v(C,2) to zero and v˜(2)

v(C, 2) dC. (Cˇ Cˆ ) Cˇ

Cˆ

Let C 2 be the unique value of C which sets v(C, 2) to ϕb. If C 0t exceeds this threshold, then creating a duopoly through entry is rational given the pessimistic expectation that the incumbent will never exit. The next step is to consider the problem of an incumbent monopolist that expects • the potential entrant will actually enter if and only if C 0t C 2, and • the potential entrant will never exit following entry. With these expectations, the value of such a monopolist is also piecewise linear in C.

if C 0t C 2

v(C 0t , 2)

1 v(C 0t , 1) [C 0t M(1)(1) b v˜ (1)] if C1 C 0t C 2 1 (1 ) 0

otherwise.

In parallel with the case of the pessimistic duopolist, C1 is the largest value of C that sets v(C, 1) to zero and Cˇv v(C, 1) v˜(1) dC. Cˆv (Cˇv Cˆv)

Entry places this incumbent into the position of the pessimistic duopolist, so v(C, 1) v(C, 2) if C C 2. Otherwise, this incumbent expects to earn the monopoly profit until either C decreases below C1 or increases above C 2. The players in this game are any initial incumbents and the entire sequence of potential entrants. A Markovian strategy for a player is a pair of functions As(N, C ) and AE (N, C ) which give probabilities of survival and entry as a function of the number of incumbent firms and the current demand state. A strategy forms a symmetric Markov-perfect equilibrium if any action it prescribes with positive probability yields a weakly higher

268

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

payoﬀ than any other action given that all other players follow the same strategy. Consider the following strategy built from the value functions v(C, 1) and v(C, 2).

0 1 A (1, C ) 0 1 A (1, C) 0

AE(0, C )

1

E

S

1 if C C otherwise, 2 if C C otherwise, if C C1 otherwise,

1

AS(2, C) p(C) 0

if C C2 if C C2 , 1 C otherwise

where v(C, 1) p(C) . v(C, 1) {[C M(2)](2)/2 κb v˜(2)} Verifying that this strategy forms a symmetric Markov-perfect equilibrium begins by showing that v(C, 1) and v(C, 2) give the values of a monopolist and duopolist when all firms follow this strategy. The key to this is to note that the mixed strategy p(C ) yields an expected payoﬀ of zero to a firm that chooses not to exit. Such a firm trades oﬀ low duopoly profits (partially oﬀset by the probability of a favorable later realization of C ) with the possibility of outlasting the rival and becoming a monopolist. With this established, deviations from the given strategy cannot improve either firm’s payoﬀ by construction. Equilibrium Summary How would data generated by this equilibrium appear to an econometrician? The big-box sector will be either empty, a monopoly, or a duopoly at any given moment. Changes in demand will shift it between those three states. If we think of each village’s independent producer as an establishment, then the econometrician observes entry when the village acquires a producer and exit when the village’s producer exits. These changes will arise from idiosyncratic village-level demand shocks and in response to changes in the big-box sector. Specifically, an increase in C 0t can induce bigbox entry, thereby lowering prices and drawing villagers with low transportation costs to the city. This lowers the profitability of operating the in-

The Role of Retail Chains: National, Regional, and Industry Results

269

dependent technology in a village of any given size, so the expansion of the big-box sector comes at the expense of the independent producers. Accordingly, the number of independent producers shrinks. These dynamics mimic the salient facts Jarmin, Klimek, and Miranda document: big-box and independent retailers compete for the same customers, and the entry and exit rates of both types of firms are positive. What is to be Done? The present model helps us see Jarmin, Klimek, and Miranda’s findings in the context of a single market outcome. While that in itself could be helpful and might inspire the creation of new stylized facts, it is only one small step towards quantifying the welfare and productivity contributions of chain retailers. Although the model has some obvious shortcomings, addressing all of them is not the most obvious high marginal product task at hand. I would like to focus my conclusion on one task that is central: understanding the possibilities for technological change and diﬀusion in the retail trade sector. Big-box retailers (and before them chain retailers) have a well-deserved reputation for deploying new technology. The macroeconomic consequences of this are large, as documented by Basu, Fernald, Oulton, and Srinivasan (2003). Nevertheless, there exists no consensus view on the constraints and possibilities for developing retail technology. Are most innovations accidental or the outcome of deliberate research? How do leadingedge technologies diﬀuse from their origin to the industry as a whole? How important is true innovation relative to imitation of other industries’ practices? Without answers to these questions, it will be hard to judge how impeding chain store development changes growth and welfare. I expect answers to come from theory, case studies, and further econometric work on large enterprise data sets. References Abbring, J. H., and J. R. Campbell. 2006. Last-in first-out oligopoly dynamics. Federal Reserve Bank of Chicago Working paper 2006-28. Bartelsman, E. J., and P. J. Dhrymes. 1998. Productivity dynamics: U.S. manufacturing plants, 1972–1986. Journal of Productivity Analysis 9 (1): 5–34. Basu, S., J. G. Fernald, N. Oulton, and S. Srinivasan. 2003. The case of the missing productivity growth, or does information technology explain why productivity accelerated in the United States but not in the United Kingdom? In NBER Macroeconomics Annual 2003, ed. M. Gertler and K. Rogoﬀ, 9–63. Cambridge, MA: MIT Press. Campbell, J. R. 2006. Competition in large markets. Federal Reserve Bank of Chicago Working Paper 2005-16. Campbell, J. R., and H. A. Hopenhayn. 2005. Market size matters. Journal of Industrial Economics 53 (1): 1–25.

270

Ronald S. Jarmin, Shawn D. Klimek, and Javier Miranda

Dunne, T., M. J. Roberts, and L. Samuelson. 1988. Patterns of firm entry and exit in U.S. manufacturing industries. RAND Journal of Economics 19 (4): 495–515. Hopenhayn, H. A. 1992. Entry, exit, and firm dynamics in long run equilibrium. Econometrica 60 (5): 1127–50. Jovanovic, B. 1982. Selection and the evolution of industry. Econometrica 50 (3): 649–70. Yeap, C. 2005. Competition and market structure in the food services industry: Changes in firm size when market size expands. University of Minnesota. Unpublished Manuscript.

7 Entry, Exit, and Labor Productivity in U. K. Retailing Evidence from Micro Data Jonathan Haskel and Raﬀaella Sadun

Introduction The retail sector has gradually become one of the most prominent industries of the U.K. economy, absorbing approximatively 20 percent of total employment in 2004 and experiencing average annual employment growth rates of about 1 percent per annum over the last decade (EUKLEMS 2008). The expansion of the sector does not seem to be matched by an equally impressive productivity performance. As documented by Basu et al. (2003), while retail trade, hotels, and catering account for about three-quarters of the U. S. Total Factor Productivity (TFP) acceleration between 1995 and 2003 (Domar-weighted industry TFP growth), the same sector seems to account for about a third of the U. K. TFP deceleration. These stylized facts have made the retail industry an area of both policy and academic interest. The purpose of this chapter is to inform the recent debate surrounding the productivity of the U. K. retail sector with new evidence arising from Jonathan Haskel is head of the economics department at Queen Mary, University of London. Raﬀaella Sadun is a research oﬃcer at the Centre for Economic Performance, London School of Economics. Financial support for this research comes from the ESRC/EPSRC Advanced Institute of Management Research, grant number RES-331-25-0030, and is carried out at CeRiBA, the Centre for Research into Business Activity, at the Business Data Linking Branch at the ONS; we are grateful to all institutions concerned for their support. This work contains statistical data from ONS which is crown copyright and reproduced with the permission of the controller HMSO and Queen’s Printer for Scotland. The use of the ONS statistical data in this work does not imply the endorsement of the ONS in relation to the interpretation or analysis of the statistical data. We thank Ralf Martin for helpful discussions and Felix Ritchie (ONS) for help on the data. We also thank participants at the CRIW conference and our discussant David Audretsch. Any errors are our own.

271

272

Jonathan Haskel and Raﬀaella Sadun

previously unexplored micro data sources. The chapter investigates the U.K. retail sector using store- and firm-level data between 1998 and 2003. First, we present the first—to the best of our knowledge—exhaustive description of the U. K. retail sector using micro data sources.1 Second, in the spirit of Foster, Haltiwanger, and Krizan (2006), we look at the contributions of firm entry and exit for the productivity growth of the sector. Third, we provide some new evidence of the recent shift of large U. K. retailers toward smaller retail formats (also documented by Griﬃth and Harmgart [2005]), which followed the introduction of new and more restrictive planning constraints for the opening of large retail stores. Based on a companion work (Haskel and Sadun 2007), we suggest that this change in the store configurations of the major U. K. retailers might be one of the factors behind the recent TFP slowdown experienced by the industry in the United Kingdom.2 The plan of the chapter is as follows. In section 7.2 we document the data sources, then describe, in section 7.3, entry and exit. Section 7.4 looks at productivity levels and growth and regulations that might have aﬀected it, and section 7.5 concludes. 7.2 Data 7.2.1 Time Period and Industries The data in this chapter comes from the Annual Respondents Database (ARD). This is a comprehensive business database that is based on the Annual Business Inquiry (ABI) performed by the Oﬃce for National Statistics (ONS). Regarding time period, the data available to us is annual from 1997. As we shall see, however, the 1997 data is not accurate, therefore in practice our analysis starts in 1998. At the time of writing the 2003 data was the final period available.3 As for industries, the ARD database covers almost all firms with Standard Industrial Classification (SIC) codes from 2010 to 93050. The retailing sector is covered by SIC92 codes from 52111 to 52740 (i.e., all codes beginning with 52). Retailing is then split into seven broad categories, as listed in table 7.1.

1. With the exception of Haskel and Khawaja (2003), an early version of this chapter. The main diﬀerence between this chapter and the previous version is that this one uses an extra year of data, and computes numbers using a diﬀerent employment measure. The latter turns out to make a substantial diﬀerence since the earlier employment measure was available only for a subset of firms, causing many firms to be dropped. This aﬀects the productivity decompositions. 2. See Haskel and Sadun (2007), and Haskel et al. (2007). 3. We particularly thank Felix Ritchie for helping in the timely provision of the 2002 and 2003 data.

Entry, Exit, and Labor Productivity in U.K. Retailing Table 7.1

273

Industries covered in UK ARD retailing data

SIC code

Industry

Notes

521

Retail sales in nonspecialized covering food, beverages, or tobacco (for example)

Includes supermarkets and department stores

522

Food, beverages, tobacco in specialized stores

523

Pharmaceutical and medical goods, cosmetic, and toilet articles

Includes chemists

524

Other retail sales of new goods in specialized stores

Includes sales of textiles, clothing, shoes, furniture, electrical appliances, hardware, books, newspapers and stationary, cameras, office supplies, computers. Clothing is the biggest area

525

Secondhand

Mostly secondhand books, secondhand goods, and antiques

526

Not in stores

Mostly mail order and stalls and markets

527

Repair

Repair of personal goods, boots and shoes, watches and clocks

7.2.2 Units of Analysis A crucial issue in what follows will be whether the analysis is by store, chain of stores, or chain of chain of stores. This section sets out in some detail what data are available to us.4 To summarize: 1. Employment, entry, and exit data are available at the store level. The store is defined as a Local Unit (LU). 2. Productivity data are available at the firm level. The firm is defined as a Reporting Unit (RU). Business Structure: Enterprises, Enterprise groups, and Local Units The fundamental business data set in the United Kingdom is the Interdepartmental Business Register (IDBR). This business register is compiled using a combination of tax records on Value Added Tax (VAT) and PayAs-You-Earn (PAYE), information lodged at Companies House, Dun and Bradstreet data, and data from other surveys. The IDBR has been operating since 1994 (before that the IDBR register information was rather uncoordinated across diﬀerent government departments). The IDBR tries to 4. It follows closely Criscuolo, Haskel, and Martin (2003).

274

Jonathan Haskel and Raﬀaella Sadun

capture the structure of ownership and control of firms and plants or business sites that make up the U. K. economy using three aggregation categories: local units, enterprises, and enterprise groups. Their meaning is best illustrated by means of an example set out in figure 7.1. Consider the left hand panel. Suppose that Brown is a single business, operating in a single location, producing goods for a single industry. Now consider the right side of the panel. Smith and Jones Holdings are a holding company, registered in London. In turn, they own two businesses, Smith and Jones, who are involved in separate industrial activities. Smith has four shops (or more generally plants/business sites, that is, a particular geographic location where trade occurs): Smith North, Smith South, Smith East, and Smith West. Jones has a shop, Jones North and a Research and Development lab, Jones R&D. Brown, being responsible for a single business activity, is an enterprise. Smith and Jones Holdings, owing businesses with distinct business activities, is called an enterprise group.5 Smith and Jones are two enterprises. All business sites, a business entity at a single mailing address, are called local units. Consequently, if Jones R&D is located at a diﬀerent site than Jones North the enterprise Jones would consist of two local units. If Jones R&D was located at the same site as Jones North the two would form one local unit for the IDBR.6 (The diagram also refers to reporting units; this will be explained later.) Maintaining Information on Business Structure: Enterprise Groups, Enterprises and Local Units The Annual Register Inquiry (ARI) is designed to maintain the business structure information on the IDBR (Jones 2000). It began operation in July 1999 and is sent to large enterprises (over 100 employees) every year, to enterprises with twenty to ninety-nine employees every four years, and to smaller enterprises on an ad hoc basis. The ARI currently covers around 68,000 enterprises, consisting of about 400,000 local units. It asks each enterprise for employment, industry activity, and the structure of the enterprise. This is straightforward for the Brown enterprise in our example. A multisite enterprise such as Smith receives a form and is asked to report on its overall activity and employment. It will also be sent four extra forms to report the same for each local unit. If Smith has closed a local unit it must report this on the form. If a local unit has opened it has to fill out extra forms, which are obtained from ONS by an automated procedure. Returns from the ARI update the IDBR in the summer of each year. 5. A holding company responsible for a number of enterprise groups is called an apex enterprise. 6. The two could nevertheless be separate local units depending on the survey. If, for example, an R&D survey which collects data just for the R&D part of the business was undertaken, this would identify them as distinct. Thus, some care has to be taken in matching business using diﬀerent surveys.

Entry, Exit, and Labor Productivity in U.K. Retailing

Fig. 7.1

275

Plants and firms in the IDBR

Maintaining Information on Employment, Turnover and Other Data As well as the structure of business information, the IDBR holds other data, such as address and SIC code. However, since the IDBR is based mostly on tax data (plus old records from previous inquiries), it also sometimes contains other data. Output information on the IDBR comes from VAT records if the original source of business information was VAT data. Employment information comes from PAYE data if that is the source of the original inclusion. Thus, as long as the single-local unit enterprise Brown is large enough to pay VAT (the threshold was £52,000 in 2000/01), it would have turnover information at the enterprise and local unit level. On the other hand, if Brown does not operate a PAYE scheme, it will have no employment information. However, employment data is required to construct sampling frames and hence is interpolated from turnover data. For the multi-local unit enterprise Smith, no turnover information will be available for Smith’s local units, since most multi-local unit enterprises do not pay VAT at the local unit level. If the PAYE scheme is operated at the local unit level, it would have independent employment data. 7.2.3 The ABI and the ARD While the IDBR holds much useful information, more data is required on outputs and other inputs in order to calculate GDP. Thus, the ONS con-

276

Jonathan Haskel and Raﬀaella Sadun

ducts a business survey based on the IDBR called the Annual Business Inquiry (ABI). The ABI covers production, construction, and some service sectors, but not public services, defense, and agriculture.7 The ARD consists of the panel micro-level information obtained from successive crosssections of the ABI. The questions asked on the ABI for retailing vary somewhat. They are required to provide details on turnover (total and broken down in retail and nonretail components, and by commodity sold), expenditures (employment costs, total materials, and taxes), items defined as work in progress, and capital expenditures (separately for acquisitions and disposals). They also have to answer sections related to import or export of services and on the use of e-commerce and employment, with further data on parttimers. However, the survey form can be sent in a long or in a short format. The main diﬀerence between the two types of formats is that in long format firms are required to provide a finer detail of the broad sections defined previously. For instance, in the long format firms break down their disposals and acquisitions information about twenty diﬀerent items, whereas in the short format they only report the aggregate values. Also, in the long format, firms answer on questions such as the total number of sites and the amount of squared meters they consist of. Reporting Units, Selected and Nonselected Data The ABI is covered by the Statistics of Trade Act (1947); therefore, the firms are obliged by law to provide data if they get a form.8 To reduce compliance costs, however, the ABI is not a census of all local units. This is in two regards: aggregation and partial sampling. Regarding aggregation, en7. The ABI replaces Annual Employment Survey, Annual Census of Production and Construction (ACOP/ACOC), and the six following Annual Inquiries: wholesale, retail, motor trades, catering, property, service trades. In Catering and Allied Trades, between 1960 and 1979 there was a benchmark inquiry into catering roughly every four years or so, but from 1979 the inquiry became annual. There has been a property inquiry since the mid-1950s, but until 1994 data was only collected on capital expenditure. From 1995, the range of data was extended to bring the inquiry in line with the other DS inquiries. The first major inquiry into Wholesaling and Dealing was carried out in respect of 1950, as part of the Census of Distribution. Subsequently, periodic large-scale detailed inquiries were conducted in respect of 1959, 1965, 1974, and 1990, but simpler annual inquiries were conducted for most intervening years and for all years since 1991. The first major inquiry into motor trades was carried out in 1950 as part of the Census of Distribution. Subsequently, periodic large-scale inquiries were conducted in respect of 1962, 1967, and 1972, although simple annual inquiries were carried out in most intervening years. By 1977 the annual inquiry was collecting detailed information on turnover and purchases. Regarding retailing, from 1950 periodic Censuses of Distribution were conducted, the last of which was in 1971. Full-scale inquiries covering every retail business and every retail outlet were taken for 1950, 1961, and 1971, with large-scale inquiries for 1957 and 1966. The first annual retailing inquiry was conducted in respect of 1976 with a sample of 30,000 units. Throughout the late 1970s and 1980s the inquiry varied from year to year in terms of both sample size and the amount of information collected. From 1991 to 1997 the sample remained reasonably constant at around 12,000. 8. Companies who have to fill out a form can refer to http://www.statistics.gov.uk/about/ business_surveys/abi/default.asp for help and information.

Entry, Exit, and Labor Productivity in U.K. Retailing

277

terprises normally report on all their local units jointly. There are two major exceptions. First, if the enterprise has local units in both Britain and Northern Ireland, there is a legal requirement for the ONS to keep data for these two areas separate, and therefore enterprises are required to report data separately in this case. Second, there is separate reporting on LUs if a business explicitly requests such a split. So, for example, Smith may decide to report on North and South combined and East and West separately. Returned data is at what is called the reporting unit (RU) level. Some examples of the possible RU structures are shown for our example at the bottom of figure 7.1. Brown forms one RU (A) only, whereas Smith has two RUs (comprising of Smith North and Smith South, and Smith East and Smith West). Jones has one RU, comprising Jones North and Jones R&D.9 Thus, these RUs are the fundamental unit for reported data on the ARD. It is worth noting at this point that the RU and LU distinction is crucial for our analysis. For example, entry and exit at the LU level might look very diﬀerent to that at the RU level. Regional issues are also important here; looking at RU data when an RU reports on a number of LUs where the LUs are based in diﬀerent regions may give a very diﬀerent picture to looking at LUs. Regarding sampling, to reduce costs, only reporting units above a certain employment threshold (currently 25010) are all sent an ABI form every year. Smaller reporting units are sampled by size-region-industry bands.11 In the ARD, all data returned from reporting units is held on what is called the selected file. Other data is held on the nonselected file. Since the nonselected RUs are not sent a form, the nonselected data is of course the IDBR data. 7.2.4 Firms (RU) and Stores (LU) in UK Retailing We now document some basic facts regarding the number of retail firms (RU) and stores (LU) operating in the United Kingdom. Table 7.2 sets out some of the relevant data for 2003, the most recent period available. First, in column 1, top panel, there were 196,286 RUs in all retailing in 2003 and 285,291 LUs. Recall that RUs can report on one or more LUs, so the higher number of LUs is to be expected. Many of these RUs and LUs, by number, are in “Other Retail,” “Food, Beverages, Tobacco,” and “Nonspecialized Stores.” The remainder of the top panel shows data on the numbers of LUs that RUs report on. Column 3 shows that 10,745 RUs report on more than 9. On other surveys the RU structure might be slightly diﬀerent, for example, on the R&D survey Jones might report on Jones R&D only that would be its RU for that survey. This matters when matching surveys. 10. The threshold was lower in the past. See Barnes and Martin (2002) for more details. 11. The employment size bands are 1–9, 10–19, 20–49, 50–99, and 100–249; the regions are England and Wales combined, Scotland, and Northern Ireland (NI). Within England and Wales industries are stratified at 4-digit level, NI is at two-digit level, and Scotland is at a hybrid 2/3/4-digit level (oversampling in Scotland and NI is by arrangement with local executives). See Partington (2001).

278

Jonathan Haskel and Raﬀaella Sadun

one LU. Thus, as column 4 shows, 185,541 RUs, the bulk of the LUs, just report on one LU (i.e., these are stand-alone firms). The remaining columns sum up to 10,745 in column 3. So, for example, the final column shows that only 171 RUs report on more than 100 LUs. In sum, approximately two-thirds of retailing outlets were accounted for by stand-alone businesses (185,541/285,291). Looking at the individual sectors, the distribution of units is the same in all seven. These data are just numbers of RUs and LUs. The lower panel shows the average employment that these units account for. Here the picture, not surprisingly, is rather diﬀerent. Columns 1 and 2 of the lower panel show mean employment in RU and LU (headcount, not FTE) is 14.14 and 9.73 in all retailing, respectively. Mean employment for Reporting Units with a single Local Unit is 3.66. But looking at the last column, the RU who reports on more than 100 LUs has average employment per RU of over 9,000. This figure suggests a very high concentration of employment across few retail firms, especially in Nonspecialized Retail. Table 7.2 suggests there are many LUs and RUs by number and considerable concentration of employment. Table 7.3 gives some more details on this. Consider the top left panel, which shows data for all industries. The first number, 185,541, is the same as in table 7.2, column 4, top cell, namely, the number of RUs who are stand-alone. As the second column shows, this group accounts for 94.4 percent of the total number of RUs. Reading further across the table, however, total employment in these LUs is 678,496, which accounts for 24.4 percent of all employment. By contrast, looking at the bottom row of the top left panel, those reporting on more than 100 local units (171 RUs, just 0.1 percent of total numbers of RUs), account for 56.7 percent of employment in all retailing. For “Nonspecialized Stores” (mostly supermarkets), 77.2 percent of employment is accounted for by just 37 RUs, who are below 1 percent of the total number of RUs. Likewise in “Pharmaceuticals” and “Other,” the largest group accounts for a very small number of RUs by number, but 47.5 and 47.9 percent of total employment. By contrast, secondhand stores are concentrated by both number and size in small groups, and so is, to a lesser extent, “Food, Beverages, and Tobacco.”12 The concentration of employment is also shown in table 7.4, which reports the percentage of the sector’s employment in the top 5 and 10 RUs and LUs. Looking at the RU data, in nonspecialized stores just ten stores account for over half of total employment.13 12. One issue for us is whether significant RUs change industry over time (e.g., for many retailers are wholesalers as well and could be classified in diﬀerent industries over time). To check this, we looked at the six largest supermarkets in the data set and found that they were consistently classified to one industry (SIC52119). Evidently, we do not have this problem in the data set for these companies. 13. The previous data has shown the relation between RUs and LUs. Above RUs are of course enterprise groups; in unreported tables we computed that most enterprise groups consist of one RU (i.e., the mean number of RUs that each enterprise group consists of is 1.01 in all sectors).

9.73 22.15 4.34 7.58 7.64 2.62 6.05 3.61

10,745 915 1,653 768 6,841 161 266 141

# of RU with more than 1 LU (3) 9,425 749 1,478 667 6,020 <150 239 <150

RU with 2 to 5 LU (5)

3.66 3.75 3.36 5.08 3.87 2.57 2.99 2.91

RU with 1 LU

17.93 36.14 15.42 16.25 14.64 10.17 <80 10.76

RU with 2 to 5 LU

Mean employment

185,541 34,503 33,492 5,405 87,656 5,389 12,611 6,485

RU with 1 LU (4)

Frequencies

83.03 171.30 49.78 60.02 69.19 23.00 <500 <50

RU with 6 to 10 LU

610 61 98 <60 375 <10 <20 <10

RU with 6 to 10 LU (6)

RU with more than 100 LU 9,201.63 25,332.27 2,653.00 <5000 <5000 1,090.50 1,414.00 <2000

564.61 1,610.22 334.52 <250 <450 70.08 <1500 <1500

171 37 <20 <10 106 <10 <20 <10

RU with more than 100 LU (8)

RU with 11 to 100 LU

539 68 <100 <50 340 <10 <20 <10

RU with 11 to 100 LU (7)

Source: Authors’ calculations from ARD. Note: Some of the cells have been suppressed for disclosure reasons. Columns 4, 5, 6, 7, and 8 add up to column 1. Columns 5, 6, 7 and 8 add to column 3.

14.14 34.26 5.64 15.56 12.06 3.33 6.09 4.71

Mean LU employment

Mean RU employment

52. All sectors 521. Nonspecialized 522. Food, beverages, tobacco 523. Pharmaceutical 524. Other retail 525. Secondhand 526. Not in stores 527. Repair

52. All sectors 521. Nonspecialized 522. Food, beverages, tobacco 523. Pharmaceutical 524. Other retail 525. Secondhand 526. Not in stores 527. Repair

285,291 54,678 45,219 12,556 143,932 6,987 13,759 8,160

196,286 35,418 35,145 6,173 94,497 5,550 12,877 6,626

Sectors

Total # of LU (2)

Reporting Unit and Local Unit numbers (Year 2003)

Total # of RU (1)

Table 7.2

Table 7.3

Employment in largest firms, by sector, 2003

Sector 52. All sectors

521. Nonspecialized

522. Food, beverages, tobacco

523. Pharmaceutical

524. Other retail

525. Secondhand

526. Not in stores

527. Repair

# of LU belonging to RU

Freq.

0– 2– 6– 11– 101– 0– 2– 6– 11– 101– 0– 2– 6– 11– 101– 0– 2– 6– 11– 101– 0– 2– 6– 11– 101– 0– 2– 6– 11– 101– 0– 2– 6– 11– 101– 0– 2– 6– 11– 101–

185,541 9,425 610 539 171 34,503 749 61 68 37 33,492 1,478 98 63 14 5,405 667 <60 <50 <10 87,656 6,020 375 340 106 5,389 <150 <10 <10 <10 12,611 239 <20 <20 <10 6,485 133 <10 <10 <10

% of total # of RU 94.5 4.8 0.3 0.3 0.1 97.4 2.1 0.2 0.2 0.1 95.3 4.2 0.3 0.2 0.0 87.6 10.8

92.8 6.4 0.4 0.4 0.1 97.1 2.5

97.9 1.9

97.9

Total emp.

% of . total emp.

Mean emp.

678,496 169,037 50,649 304,326 1,573,478 129,234 27,067 10,449 109,495 937,294 112,507 22,791 4,878 21,075 37,142 27,468 10,836 3,181 8,917 45,677 338,828 88,105 25,948 141,017 545,993 13,873 1,413 184 841 2,181 48.07 22.19 7.60 20.34 1.80 60.59 4.59 0.17 22.54 12.11

24.4 6.1 1.8 11.0 56.7 10.6 2.2 0.9 9.0 77.2 56.7 11.5 2.5 10.6 18.7 28.6 11.3

3.7 17.9 83.0 564.6 9,201.6 3.7 36.1 171.3 1,610.2 25,332.3 3.4 15.4 49.8 334.5 2,653.0 5.1 16.2 60.0 228.6 5,075.2 3.9 14.6 69.2 414.8 5,150.9 2.6 10.2 23.0 70.1 1,090.5 3.0 72.8 425.4 1,329.1 1,414.0 2.9 10.8 53.0 1,406.4 1,888.5

Source: Authors’ calculations from ARD. Note: Some of the cells have been suppressed for disclosure reasons.

47.5 29.7 7.7 2.3 12.4 47.9 75.0

11.8 36.5 12.6

66.5

Entry, Exit, and Labor Productivity in U.K. Retailing Table 7.4

281

Firm concentration of employment by industry 5 and 10 firm concentration ratios, 2003

Reporting Units industry 52. All sectors 521. Nonspecialized 522. Food, beverages, tobacco 523. Pharmaceutical 524. Other retail 525. Secondhand 526. Not in stores 527. Repair

cr5

cr10

Number of RU

22.03 49.60 13.05 42.14 15.08 14.97 25.63 35.04

29.67 65.20 17.04 48.62 23.04 18.09 33.76 36.96

196,286 35,418 35,145 6,173 94,497 5,550 12,877 6,626

Source: Authors’ calculations from ARD. Note: cr = concentration ratios.

Table 7.5

Mean employment, by region, 2003, all industries

Region South East (G) East Anglia (F) South West (J) West Midlands (E) East Midlands (C) Yorkshire and Humberside (C) North West (B) North (A) Wales (W) Scotland (X)

Employment per Reporting Unit

Employment per Local Unit (using RU regional identifier)

Employment per Local Unit (using LU regional identifier)

RU Frequency

22.38 6.27 9.80 6.50 8.05 10.60 17.25 6.39 5.98 11.61

12.95 5.33 7.40 5.07 6.30 8.13 11.75 5.06 4.91 8.69

10.31 9.87 9.61 9.01 9.66 9.44 9.69 9.64 9.20 9.16

65,518 7,380 18,029 17,368 14,118 17,516 21,276 9,011 9,426 16,644

Source: Authors’ calculations from ARD.

Thus far we have looked at employment by industry. Table 7.5 shows mean employment by region. Consider first the average employment per RU in column 1. This is 22 in the Southeast, larger than elsewhere. There are two issues here. First, an RU might actually consist of multiple LUs and hence column 2 shows employment per LU; this number is a bit smaller. Second, RUs might report on a number of LUs, and if the RU is the head oﬃce (located in London, for example), this might be a misleading number for the average size of the actual store. Thus, column 3 shows average size by LU using the regional identifier by the LU rather than by the RU. This shows smaller numbers than in columns 1 and 2 and the numbers are now much closer together.

282

Jonathan Haskel and Raﬀaella Sadun

7.2.5 Section Summary We find that: 1. In 2003 there were 285,291 stores in U.K. retailing and 196,286 firms/ chains. 2. Average store employment is 9.73 employees (not FTEs). 3. 171 chains accounted for 56.7 percent of total employment. 7.3 Entry and Exit This section looks at exit and entry defined as: 1. 2. 3. 4.

Entrant: Present in t and not present in t – 1. Exitor: Present in t and not present in t 1. 1-year: Present in t and not present in either t – 1 or t 1. Stayer: None of the above three.

We look separately at RUs and LUs to provide as full as information as possible. Using these definitions, the basic data for the whole retailing sector covering the period 1998 to 2001 is set out in table 7.6. The total numbers of RUs and LUs are as shown in the total column, and the numbers in the left-hand four columns add up to this number. As it shows, the bulk of the RUs and LUs are stayers with entry and exit rates (i.e., entry and exit numbers as shares of the total number of LUs that year) of around 10 to 20 percent, depending a bit on RU or LU status. Note the apparently high entry rate in 1998 by LU and RU, which might have to do with register problems in 1997. Table 7.7 shows data on entry and exit rates by industry (regional entry and exit rates were quite similar). By industry, entry and exit rates for LUs (lower panel of table 7.5) look quite similar, with 19.8 percent in “Not in Stores” and 7.55 percent in “Pharmaceutical” being the maximum and minimum exit rates, and 14.26 and 7.26 being likewise the maximum and minimum exit rates.14 7.3.1 Section Summary We find that: 1. Entry/exit/one-year/stayers are fairly stable fractions of all stores, being about 11 percent,11 percent, 5 percent, and 63 percent. 2. Entry and exit rates are lowest in “Pharmaceuticals” and highest in “Not in Stores.” 14. We looked at whether entry and exit diﬀered statistically significantly by region and/or industry, using an analysis of variance approach. We found, however, (results available on request) that it did not diﬀer significantly by region, but did do so by industry. Note that the Competition Commission (2000) states that planning policy is national, so the extent that entry and exit rates might be aﬀected by planning might not expect them to diﬀer by industry.

144,314 163,399 164,055 162,989 154,971

203,585 223,803 212,879 208,524 208,374

1998 1999 2000 2001 2002

1998 1999 2000 2001 2002

53,038 31,687 32,392 34,177 35,177

39,947 21,531 20,217 17,337 17,253

Entrants

Entry and exit

Source: Authors’ calculations from ARD.

Stayers

Year

Table 7.6

28,469 32,820 42,611 36,747 34,327

18,316 20,862 20,875 21,283 25,355

Exitors

Frequencies Total

10,076 7,510 8,980 14,017 13,151

Local Units 295,168 295,820 296,862 293,465 291,029

Reporting Units, All sectors 6,855 209,432 4,774 210,566 4,436 209,583 3,593 205,202 4,033 201,612

1 year

17.97 10.71 10.91 11.65 12.09

19.07 10.23 9.65 8.45 8.56

Entry rate

9.65 11.09 14.35 12.52 11.80

8.75 9.91 9.96 10.37 12.58

Exit rate

Rates

68.97 75.66 71.71 71.06 71.60

68.91 77.60 78.28 79.43 76.87

Stay rate

3.41 2.54 3.02 4.78 4.52

3.27 2.27 2.12 1.75 2.00

1 year rate

154,971 27,162 30,075 5,285 74,241 4,729 9,114 4,365

208,374 36,387 34,585 9,649 104,813 5,210 9,293 4,687

52. All sectors 521. Nonspecialized 522. Food, beverages, tobacco 523. Pharmaceutical 524. Other retail 525. Secondhand 526. Not in stores 527. Repair

52. All sectors 521. Nonspecialized 522. Food, beverages, tobacco 523. Pharmaceutical 524. Other retail 525. Secondhand 526. Not in stores 527. Repair

Source: Authors’ calculations from ARD.

Stayers

35,177 8,010 5,960 915 16,467 960 1,620 924

17,253 3,582 2,294 366 8,397 339 1,409 866

Entrants

Entry and exit rates by industry

Sectors

Table 7.7

12.09 14.66 12.42 7.26 11.58 13.20 11.42 12.79

Entry rate

Local Units (Year 2001) 34,327 13,151 291,029 7,083 3,160 54,640 6,549 910 48,004 952 1,087 12,603 14,629 6,334 142,243 922 181 7,273 2,808 464 14,185 918 698 7,227

Total

8.56 9.82 5.99 5.75 8.86 5.74 10.45 13.70

1 year

Reporting Units (Year 2002) 25,355 4,033 201,612 5,010 732 36,486 5,318 629 38,316 651 68 6,370 10,265 1,819 94,722 702 139 5,909 2,591 374 13,488 818 272 6,321

Exitors

Frequencies

11.80 12.96 13.64 7.55 10.28 12.68 19.80 12.70

12.58 13.73 13.88 10.22 10.84 11.88 19.21 12.94

Exit rate

71.60 66.59 72.05 76.56 73.69 71.63 65.51 64.85

76.87 74.44 78.49 82.97 78.38 80.03 67.57 69.06

Stay rate

Rates

4.52 5.78 1.90 8.62 4.45 2.49 3.27 9.66

2.00 2.01 1.64 1.07 1.92 2.35 2.77 4.30

1 year rate

Entry, Exit, and Labor Productivity in U.K. Retailing

285

7.4 Productivity 7.4.1 What Productivity Data is Available? As discussed in the previous section, data is at essentially two levels, RU and LU. Reporting Unit (RU) data is returned data (i.e., it relies on data actually reported by firms). Local Unit (LU) data is a mix of data that is from the ARI, and so is reported by firms and from other sources (e.g., taxes, which is inferred). Given that the LUs correspond to stores, this would seem to be the most desirable for a number of cases, especially since a number of retailers consist of many stores. Unfortunately, there are some issues surrounding the use of productivity data at the store level, especially for stores that belong to retail chains (i.e., that are not stand-alone), which force us to use firm-level (instead of store-level) productivity data.15 7.4.2 Data Available on Outputs and Inputs As described previously, the only reliable input and output data is that available for RUs. Table 7.8 sets out some of the basic data available for all retailing sectors. Each observation in the data represents one RU. The top rows show data on sales, gross value added, and gross output. Following the ONS, gross value added at factor cost is calculated as equal to Turnover (exc. VAT) Net addition to stocks Work of Capital Nature by Own Staﬀ Insurance Claims Received – Purchases. Gross output, on the other hand, is equal to Turnover (exc. VAT) Work in Progress Stocks Bought for Resale Work of Capital Nature by Own Staﬀ. The main diﬀerence between the two is the purchases figure, which is deducted in the calculation of gross value added. The rest of the table shows some summary statistics for each variable; not surprisingly, purchases are the largest element after sales. One interesting point is that we have data on employment and the fraction of employees 15. The productivity data for LUs that do not correspond to single-unit RUs comes from the IDBR database, which is derived either from the IDBR administrative sources (i.e., the VAT or PAYE), or other data that brought the business onto the register in the first place, or the ARI. First, as discussed in section 7.3, some of the input data is interpolated from sales data, and vice versa. An additional problem arises from the fact that—according to ONS (2001)—when a business first arrives on the register, its employment, if present, is frozen at its first reported point until updated, and the updating process seems to be particularly slow. Updating is done from the results of the ARI, or before the ARI was introduced, if the firm was in one of the Annual Employment Surveys (AES). According to Partington (2001), in 2000 8.5 percent of total employment had not been updated since 1993, the year when there was last a Census of Employment. The updating problem seems to be concentrated in the smallest enterprises. In enterprises of size 0–9 28.7 percent of employment and 40.2 percent of employment in enterprises of size 10–19 had not been updated since 1993. Of enterprises of size 0–9 and 10–19, 56.9 and 21.8 have never been sent an ARI form or included in the AES. By contrast, larger enterprises are updated more frequently. An additional problem is that the ONS (2001) also state that even larger enterprises in the ARI or AES may not have fully reported on their local units.

286

Jonathan Haskel and Raﬀaella Sadun

Table 7.8

Basic data available for selected firms (year = 2003) Nonmissing observations

Mean

Median

Standard deviation

6,071 6,042 6,042 6,071 6,042 6,042 6,071 5,725 6,074

24,697.01 5,758.63 24,903.89 78.64 9.41 5.10 19,058.97 326.03 186.36

330.00 83.00 335.83 0.00 0.00 0.00 227.00 5.00 2.00

395,078.80 84,984.59 396,577.00 2,095.73 277.86 134.73 313,330.65 4,332.64 2,646.73

Sales Gross value added Gross output Net addition to stocks Work of capital nature by own staff Insurance claims received Purchases of materials and fuel Employment Part-timers Source: Authors’ calculations from ARD.

who are part-time. We do not know, however, what proportion of the full week such employees work, so we allocated them to 50 percent of the work week to calculate FTEs. In what follows, we present productivity data by employment and by FTE employment. 7.4.3 Deflators We use price deflators provided by ONS for four-digit industries, which, for retailing, are mostly disaggregated indices from the retail price index. Therefore, they are consumer price indices. No deflators are available for retailing for materials and fuel purchased, and so value added is single deflated. 7.4.4 Productivity in Retailing: Definitions An important problem of measuring retailing productivity surrounds the diﬃculty of defining what output a retailer provides. This is important in considering the argument that, for example, retailers have raised their productivity by simply shifting costs to either consumers (the growth of self-service stores) or onto producers (the allegation that Wal-Mart gets it suppliers to do more of the work in delivering the item to the shelf, for example, by supplying in shelf-ready packets, Bosworth and Triplett [2003]). It is also important in considering that measured sales of electronic stores have risen by 15 percent per year from 1987 to 2001 in the United States (Bosworth and Triplett 2003). Oi (1993) emphasizes that the output of a retail firm is a bundle of services surrounding the product sold. Betancourt and Gautschi (1993) suggest they can be put into five categories: convenience, assortment, assurance of delivery in the desired form at the desired time, information, and ambience. Consider, then, a self-service supermarket selling fruit and packed meat as against a grocer selling fruit and a butcher selling fresh meat. By

Entry, Exit, and Labor Productivity in U.K. Retailing

287

making consumers serve themselves, the supermarket has shifted costs to consumers and so this should be deducted from retail output. Against this, the supermarket is providing the service of convenience to consumers (having the food available under one roof), which should be added to retail output. Betancourt and Malanoski (1999) thus model retailer’s transformation function as consisting of both the output of retail items and also the output of distribution services (in this case the convenience of items under one roof and the input of the shopper’s time). In turn, the output of distribution services is an input into the household production function, which then determines the demand for the supermarket’s physical goods. Thus, growth of self-service stores represents the increasing provision of distribution services (all items under one roof, that is, increased service provision) along with substitution between in-store labor (who used to serve every customer) to consumer labor (i.e., reduced service provision). Therefore, to measure retail output we should have to subtract from measured sales values the net valuation of these changes in services, which is of course a very hard task. What about shifting to suppliers? Here is a case of substitution not between final consumers and in-store labor, but between bought-in materials and in-store labor. This cautions against using margins (sales less costs of bought-in goods) as a measure of output since this is only valid if there is no substitution between bought-in materials and other inputs (just as in the literature on raw materials and productivity, see Bruno [1978]). Instead, it would seem to be more appropriate to use double-deflated value added. Finally, the increase in real sales in electronic stores is surely due to the fall in prices of underlying goods, and not the increased eﬀorts of the staﬀ in electronic stores. Thus, it would not seem appropriate to use sales per person as a productivity measure. In conclusion, the argument is essentially one of interpretation. Just as in conventional production functions output per person might be very high in capital-intensive industries, so it is that output per person might be very high in retailing sectors where customers do all the work and/or input prices are very low. The important contribution of this theory is that it helps list the key inputs that account for measured sales per person. In the case of retailing, this is the important insight that retailers produce both sales of physical goods but also distribution services; some of the latter can be shifted onto consumers. Given the problems of measuring consumer services, in what follows we use productivity with the numerator measured using both sales and value added. Thus, it should be emphasized that crosssection comparisons might not be a good guide to the bundle of sales and distribution services that are more appropriate measures of retail output. There are at least two other issues that might or might not be more important than the failure to adjust for distribution services in making crosssection comparisons. First, diﬀerent stores sell diﬀerent baskets of goods. Second, retailing employs many part-time workers. To deal with the latter,

288

Jonathan Haskel and Raﬀaella Sadun

we use both employees and full-time equivalent employees in the denominator of the productivity calculation. 7.4.5 Weights Since we use the selected file, we deal not with the whole industry, but a sample. Thus, we need to develop weights to use where appropriate. To do this we use both the selected and nonselected file, but with a robustness check as follows. We combined both files to make a grand file of selected RUs and nonselected LUs. We then split the sample into six sizebands (0– 9, 10–20, 21–50, 51–100, 101–250, and 250). We then calculated weights as the sum of selected and nonselected employment divided by selected employment in each sizeband. So for example, if a firm falls into sizeband 50 to 99 and total selected employment was 1,000, but total selected and nonselected employment was 2,000, the weight for that sizeband would be 2. For robustness, we checked to see that no weight was abnormally large. 7.4.6 Productivity Findings Table 7.9 looks at productivity levels by size of RU, with productivity measured by log GVA per full-time equivalent, with the left panel showing all sectors and the right the nonspecialized industry (supermarkets). As the table shows, productivity levels rise by size of RU. It is interesting that the size advantage of the largest firms is 34 percent (2.99 to 2.65) when using FTEs but 21 percent when using all employees, suggesting that part-timers are more heavily represented in larger RUs. Note, too, that the productivity advantage is 48 percent in nonspecialized stores. The lower panels of table 7.9 show growth rates by size. It is notable that the smaller reporting units have grown faster than the larger ones, thus narrowing the gap be-

Table 7.9

Log GVA per full-time equivalent, by employment size, 2003 All retailing

Sizeband Employment FTE Frequency Log growth rates (1998–2003) Log growth rates (1998–2002) Log growth rates (1998–2001)

Nonspecialized (521)

0–9

10–100

100–500

500+

0–9

10–100

100–500

500+

2.45 2.65 3,088

2.67 2.92 1,599

2.69 2.94 330

2.66 2.99 283

2.16 2.39 497

2.14 2.55 209

2.42 2.76 67

2.49 2.87 75

5.67

6.16

3.03

4.31

8.19

5.64

1.71

1.36

5.38

6.33

3.81

4.32

8.75

4.76

1.92

0.43

6.87

5.95

2.59

4.46

12.16

5.19

–0.03

–0.53

Source: Authors’ calculations from ARD.

Entry, Exit, and Labor Productivity in U.K. Retailing

289

tween large and small RUs. This is particularly marked in the nonspecialized sector. Table 7.10 contains data on productivity spreads. Foster, Haltiwanger, and Krizan (2006) for the United States, using data on stores, quote a standard deviation and interquartile range of 0.5 for hours-weighted log gross output per head in after taking deviations from four-digit means. We use data on log gross output per FTE in after taking deviations from threedigit means. As the table shows, we find a slightly higher standard deviation and interquartile range than they do. Note the spreads are not too much aﬀected by whether FTE or not. 7.5 The Sources of Productivity Growth What is the contribution of entry and exit to productivity growth in services? We employ the decomposition of Foster, Haltiwanger, and Krizan (FHK 2006). We start by writing manufacturing-wide productivity in year t, Pt as: Pt ∑ it pit

(1)

i

where i is the share of establishment i (employment share) and pit is ln productivity. Foster, Haltiwanger, and Krizan (2006) (FHK) suggest a decomposition to the change in manufacturing-wide labor productivity or ln TFP between t – k and t, Pt as (2)

Pt ∑ i,tkpit ∑ it( pi,tk Ptk) ∑ itpit i∈S

i∈S

i∈S

∑ it ( pit Ptk) ∑ i,tk ( pi,tk Ptk) i∈N

FHK

i∈X

where S, N, and X denotes the establishments that survive, enter, and exit respectively between t and t – k. The first term in (2) shows the contribu-

Table 7.10

Productivity spread, 2003 Variable GVA per head GVA per FTE GO per head GO per FTE Frequency

Standard deviation

IQR

0.91 0.90 0.74 0.73 5,300

0.95 0.88 0.85 0.81

Note: All data are transformed first into deviations from three-digit industry means. GVA means gross value added, GO gross output. IQR means interquartile range. Source: Authors’ calculations from ARD.

290

Jonathan Haskel and Raﬀaella Sadun

tion to productivity growth of growth within the surviving establishments; the second term shows the contribution of changes in shares of the survivors weighted by start period productivity relative to the average; the third term is an additional covariance term that is positive when market share increases (falls) for establishments with growing (falling) productivity; the fourth and fifth terms show the contribution of entry and exit.16 They are positive when there is entry (exit) of above- (below) average productivity establishments. To calculate this we proceed as follows. First, we performed the decomposition 1998 to 2001, 2 and 3. It is quite plausible that over diﬀerent year spans there might be diﬀerent fractions of productivity growth accounted for by diﬀerent components of the decomposition. Second, we undertake this investigation by RU and so drop all LUs since we have no productivity information for them (but recall there are many, by number, single LU and RUs who we retain since they have productivity information). Recall that RUs can exit and enter from the selected file if they are not sampled. In this case, they have moved to the nonselected file and so we use the selected and nonselected data to identify true exitors and entrants.17 But we drop an RU if it exited from the selected data into the nonselected data (or entered from nonselected into selected) since although they are a stayer, we have no productivity data for them in at least one period. Fourth, we calculate two sets of weights: employment (FTE) weights for i in (1) and (2) and also employment (FTE) weights taking into account sampling.18 Fifth, we perform these calculations by three-digit industry, that is, the P in (2) is the threedigit average industry productivity level and the is the share of each RU in three-digit industry employment. Thus, the number for all industries is constructed as a weighted sum of the numbers for the individual industries, where, following Foster, Haltiwanger, and Krizan (2006) the weights are the share of gross value added (since we use value added as our productivity measure in the decomposition) in each industry averaged over the start and end period. Sixth, the data are deflated by prices from the Retail Sales Inquiry values. The results are shown in table 7.11. The table takes up three main issues. First, the decomposition is for diﬀerent years. Second, we use both simple weights and stratified weights. The latter should upweight the smaller firms (who are more likely to enter and exit and so increase this category). Third, panel B drops the top five companies (i.e., those with the largest weights). 16. With industry data one can decompose Pt into the within and between terms, but cannot account for net entry. 17. An RU might disappear via a takeover if the taking over firm amalgamates its RUs into one or more existing RU structures. It might not disappear if it keeps the RU number. Practice on this seems to vary across firms. 18. The former are straightforward, being employment for unit i divided by employment in all i units in the industry. The latter is employment in plant i times the weight that plant has, divided by the sum of thus weighted employment in the industry.

Entry, Exit, and Labor Productivity in U.K. Retailing Table 7.11

Year

1998–2003 1998–2002 1998–2001

1998–2003 1998–2002 1998–2001

291

FHK decomposition, all retailing sectors gross value added per FTE

Weights

Simple Stratification Simple Stratification Simple Stratification

Productivity growth A. All RUs 0.14 0.19 0.02 0.06 0.04 0.06

B. Dropping top 5 RUs by weight Simple 0.23 Stratification 0.28 Simple 0.10 Stratification 0.14 Simple 0.06 Stratification 0.08

Stayers’ share

Entry/exit share

0.92 0.65 1.04 0.24 0.71 0.42

0.08 0.35 –0.04 0.76 0.29 0.58

0.76 0.56 0.54 0.32 0.74 0.46

0.24 0.44 0.46 0.68 0.26 0.54

Source: Authors’ calculations from ARD. Note: Productivity is calculated as gross value added per FTE. Numbers in “Stayers” and “Entry/Exit” columns are shares of total productivity growth in the “Productivity growth” columns. These shares are the shares for each three-digit industry, the weights are the share of gross value added in each three-digit industry averaged over the start and end period.

As we have seen, retailing is very concentrated and thus a few large RUs dominate the market. It therefore seems sensible to examine the decompositions with and without their contributions as a matter of robustness. Table 7.11 uses gross value added per FTE as the productivity measure. The first row shows that between 1998 and 2003, productivity growth was 0.14 percent over the whole period, with 92 percent accounted for by stayers (the sum of the first three terms of [2]) and the remaining 8 percent accounted for by net entry (the last two terms of [2]). The second row shows that taking account of stratification via the weights changes the proportion to 65 percent and 35 percent. The other rows in panel A do the same, but for diﬀerent time periods. The stayers’ share is generally in the majority with the exception of the stratified results for 1998 to 2002, where it is much smaller. Before drawing some overall conclusions, consider panel B, which removes the top five firms by weight. Here the picture is a little less volatile, although diﬀerences still do occur due to stratification. Using the stratification weights generally raises the contribution of entrants and exitors, as would be expected. Dropping the top five RUs lowers overall productivity growth and, in 1998 to 2003 and 1998 to 2002 the contribution of stayers (the 1998 to 2001 data are little aﬀected one way or another). Upon further investigation, this turned out to be due to one large RU with the third largest weight, who had a very large fall in productivity growth and in market share. The fall in productivity growth was suﬃcient to reduce average productivity growth as shown in panels A and B. The

292

Jonathan Haskel and Raﬀaella Sadun

stayers’ contribution rose, however, since this firm had a large and positive covariance term (productivity growth was falling but market share was falling too). Finally, both aggregate productivity growth and the share of entrants looks somewhat diﬀerent in 1998 to 2003, there appearing to be a burst of productivity growth over that year accounted for by entrants. So which results are the most reliable? First, regarding stratification, the sampling weights that we use are designed to take account of the fact that the large firms are sampled always and the small firms only with a certain probability (around 50 percent, depending on sizeband). However, by ONS rules, firms with less than ten employees are excluded from business surveys for three years after filling in a form. Hence these firms are, by definition, dropped from the decomposition since even if they would have been on the selected sample they cannot be so sampled. Thus, a weight of the inverse of the sizeband sampling probability will not allow for this since the weight has also to be time dependent. Assuming that these firms are likely to exit this means that exitors in the sample we use are likely to be too large relative to the underlying population of exitors (since they have to be observed in the base year) and so might be more productive than would otherwise be the case, making it seem as if good firms exit, thus lowering the exit contribution. Since a small RU observed in 1998 cannot appear until at least 2002, this cautions against using the shorter decompositions. A second point about stratification relates to continuers bias (Martin 2004). Recall we are using a sample of firms rather than taking a census. Consider, then, a group of firms observed in the base year. Firms who are truly stayers are more likely to be dropped since they are not likely to be observed in the final year, whereas exitors will be recorded as exitors. Thus, the weight on stayers is too low and hence their contribution is too small. There is an additional complication, however; sampling is in fact by sizeband, with firms over 250 employees subject to 100 percent sampling and under 250 subject to partial sampling. Thus, we have to drop initially large stayers who migrate to smaller sizebands (since we observe them in the base period but possibly not in the final period) and also have to drop initially small stayers who migrate to larger sizebands (since we observe them in the final period but not necessarily in the base period). The eﬀect on the contribution of the remaining stayers depends on whether transitions across sizebands are symmetrically related to productivity growth; one hopes either that the productivity growth of the firms who do not migrate across sizebands is representative of stayers as a whole, and/or that the dropping of firms who migrate down to a smaller sizeband is outweighed by the dropping of those who migrate up to a larger sizeband.19 19. Consider, then, an initially large firm. Let us suppose they have above-average productivity, which large firms tend to do. A transition to a lower sizeband means they have lost market share, so the second term in the decomposition is negative. If they have had falling productivity with falling market share then the cross term is positive (although downsizing might

Entry, Exit, and Labor Productivity in U.K. Retailing

293

Third, we should note that the 1998 to 2002 results in the top panel for nonstratified weighting are based on an overall productivity growth of 2 percent. Thus, the shares of stayers’ and net entry are formed by dividing the values of the contributions by 2 percent, which is a small number. These shares might then not be so reliable, although of course, since they are accounting decompositions they are exact descriptions of how the 2 percent growth is accounted for. Overall then, the 1998 to 2003 numbers might be regarded as the most reliable for all these reasons, and they suggest that stayers account for most of the productivity growth over that period. What, then, can we say about the possible impact of planning on productivity growth over the period? This depends upon what one would have expected the shares to be without planning restrictions. One way to look at this is using the pre-1996 data, but that is not available to us at the moment. Another way is to use the United States as a yardstick. These show that, using stores and not firms as we do here, almost 100 percent of productivity growth between 1987 and 1997 (and subperiods) to be due to entry and exit.20 However, table 7.13 of FHK also provides data on the fraction of entry and exit due to expansion and closure of stores within existing firms. This shows that 40 percent of all productivity growth is due to this source. Thus, a U.S. decomposition using firms would show that 40 percent of productivity growth is due to within-firm eﬀects and 60 percent is due to entry and exit. This seems to give a larger eﬀect for entrants than the eﬀect here. 7.5.1 Section Summary We find that: 1. Productivity is best measured at the RU level. 2. The variation in labor productivity across retailers is somewhat larger than in the U.S. 3. If anything, the contribution of entry and exit to productivity growth is somewhat smaller than in the U.S. have meant falling employment and so market share and rising productivity). The results for the initially smaller firm are the opposite. Note, however, there is a fundamental asymmetry since large firms are always sampled. Hence, of the firms moving up the size distribution, one is always more likely to drop the initially smaller since the initially larger who get larger are included both in the initial and final period. But of the firms moving down, one is likely to drop even numbers of them as they move to lower sizebands. Hence, relative to the full census, one is always being forced to drop from stayers more small firms who get larger. Overall, then, uneven sampling by size understates the role of small firms who get bigger. If they are initially low productivity and their productivity grows as they get bigger, then the eﬀect on the decomposition of their omission is that the first term is understated, the second term is overstated, and the third term understated. If their productivity falls as they get bigger, the first term is overstated, the second term is overstated still, and the third term overstated. Therefore, which way the bias goes depends on the particular industry at the time. 20. Our data is a shorter subperiod than the five years that FHK use, but the subperiods they use still show the same fraction of productivity growth due to entry and exit in the longer period.

294

Jonathan Haskel and Raﬀaella Sadun

7.6 Planning, Store Size, and the U. K. Retail Productivity Performance As discussed in the previous section, U.K. retailing productivity growth has lagged behind that of the United States. A widespread hypothesis is that the United Kingdom slowdown is somehow linked to the introduction of regulations constraining the entry of large retail formats (e.g., planning), which is generally dated to 1996. In this section we review data on U.K. regulations and planning. Building a new supermarket in the United Kingdom requires planning permission from local authorities.21 An application is made public to allow objections to be tabled. The local authority decides the application in the light of national planning rules (see the following for a description of these). In some cases the secretary of state is required to be notified; in retailing this is for things such as large proposals (more than 5,000 square meters of gross retail floor space) or substantial changes of use. If the secretary of state decides to call in an inquiry then a public inquiry is held. In the case of a refusal of planning permission an appeal can be lodged, with appeals regarding large retail sites as the subject of a public inquiry. Otherwise the decision is made by a local inspector at a hearing, but in certain cases (e.g., retail development over 9,280 square meters of gross floor space) the inspector makes his recommendation in his report to the minister and the decision is made at this level. What are the broad parameters of planning policy? This is nicely summarized in the Competition Commission (CC 2000, paragraphs 2.162ﬀ and appendix 12.4). They begin by summarizing the position before 1996. “Over recent years there has been a marked change in emphasis of the policy on land use planning for retail development. The planning guidance for England is set out in Planning Policy Note PPG6: Town Centres and Retail Developments. The first version of PPG6, issued in January 1988, did not contain advice on specific locations for retail development. A proliferation of large superstores followed, often on Greenfield sites, and sometimes as part of a far larger mixed retail park development” (paragraph 2.162). In 1996, however, this changed when a sequential approach to planning was adopted. The details of such a change are set out in the CC (2000, Appendix 12.3). The PPG6, issued in 1996, states that that city, town, and district centers should be the preferred locations for all developments that attract many trips; that is, for leisure and commercial and public oﬃce 21. The Competition Commission (2000, Appendix 12.2) reviews the rules: “Before any development is carried out it must have planning permission (in the case of some forms of development it is not necessary to apply for planning permission because permission is granted automatically by virtue of ‘permitted development’ arrangements). A development is defined as either the carrying out of specified ‘operations’ or a material change of use. One of the specified ‘operations’ is the erection of a building. Generally, a change of use will not be material unless it is of such a character that it is significant with regard to the objectives of planning control.”

Entry, Exit, and Labor Productivity in U.K. Retailing

295

development as well as retail development. However, there was a particular focus on supermarkets, who were viewed acting as an anchor for smaller city centers in particular. The PPG6 outlined a sequential approach to identifying additional sites for both retail development and other key town center uses that attract many people, including commercial and public oﬃces, entertainment, and leisure. This gives first preference for town center sites, followed by edge-of-center sites, and only then out-of-center sites. Developers proposing new supermarkets outside town, district, or local centers should demonstrate that: “. . . there is a ‘need’ for the retail floor space proposed and no more central sites that are suitable or available for developing such a store, after having been flexible about format, scale, design and amount of car parking required, tailoring these to fit the local circumstances.” (CC 2000). The issue of need was also taken up in a document issued in February 1999 by the planning minister (CC 2000): “Need should not be regarded as being fulfilled simply by showing that there is capacity (in physical terms) or demand (in terms of available expenditure in the catchment area) for the proposed development.” It stated that, while the existence of capacity or demand may form part of the demonstration of need, the significance in any particular case of the factors that may show need will be a matter for the decision-maker. The CC suggests that this document, which was designed to clarify need, served mostly to obscure it. Overall, the costs of planning regulation for businesses are nonnegligible. The U.K. Competition Commission (CC 2000) documents that it takes on average eleven to twenty-four months to obtain a planning decision for a large retail store. Moreover, the average cost per project of getting planning permission was £50,000 (approximately $90,000). 7.6.1 Impact What has been the impact of the planning policy? The CC draws some conclusions with respect to supermarkets: “The policy change, with its emphasis on revitalising town centres, has had a major impact on the store development plans of some of the larger multiples, in particular Asda and Morrison (CC 2000, paragraph 2.168). They continued (paragraph 2.203): “Multiple retailers such as Asda and Morrison, whose existing store formats are at the upper end of the size range, will have been most aﬀected by the restrictions imposed by the new planning guidelines because sites for such stores will rarely be available in town centres. Because Asda and Morrison in particular have maintained their policy of building only very large stores, they will also be the least well placed to adapt to smaller formats. . . . Tesco has also already diversified into smaller town centre formats.” In a recent paper Griﬃth and Harmgart (2005) study this further. They use data from the Institute of Grocery Distribution (IGD), which provides store-level data for all large grocery chains, all co-ops, and around 80 per-

296

Jonathan Haskel and Raﬀaella Sadun

cent of grocery retailers. What they show is a very substantial move by the big four supermarkets (who dominate the supermarket sector)22 towards opening small store formats, often in the centers of towns and often via takeover of existing small store retailers. Their figure 3, for example, shows that in 1996 the big four supermarkets opened around twenty-five high street/neighborhood stores, out of ninety total new openings. By 2004 they were opening 125 high street or neighborhood stores out of 160 total openings. Strategy also varies by supermarket. Tesco, in 2003, opened around 120 out of 140 local stores (local Tesco stores are around 2,000 square feet, a Tesco supermarket is around 27,000 square feet or Tesco Extra hypermarket of 69,000 square feet). But Asda (who were taken over by Wal-Mart in 1999 and had a strategy of building big-box stores) have tended to stick to this strategy and not open smaller stores. What does our ONS data set show? We do not have, at time of writing, data pre-1996. Thus, we cannot do a before/after comparison in this chapter. But we can document some changes from 1997. Consider first, as background, the move from independents to chains, as documented in the United States by Jarmin, Klimek, and Miranda. Table 7.12 shows the business shares of chain stores versus independents, with chains split up into regional chains (who are in one region) and national chains (who are in two or more regions). The top panel shows the data for all retailing and the bottom for SIC 521 nonspecialized stores (supermarkets), with the data showing the shares of numbers of stores on the left and shares of employment on the right. As the data show, there has been a shift in all, not so much away from independents, but away from regional to national chains (see columns 2 and 3 in the top panel). The lower panel suggests that supermarkets have seen a decline in both regional chains and in stand-alone shops with an accompanying growth in national chains. How has this occurred? It is often said that planning makes entry diﬃcult in U.K. retailing and hence the only way to expand is by merger. Table 7.13 gives some information on this, showing, again for all retailing (top panel) and supermarkets (bottom panel) the share of entrants and share of employment accounted for by entry and takeovers. The takeover data are quite volatile, reflecting the fact that a single large takeover can aﬀect the data substantially, but overall takeover and entry shares do match each other in at least some years. Table 7.14 gives a further perspective on this, by computing the shares of the stores of all entrants and exitors due to stand-alone (independent shops), regional, chains and national chains. As the top two rows show, in 1998 stand-alone shops accounted for 75 percent of entry and exit, but by 22. They are Tesco, Sainsbury, Morrisons, and ASDA. The first three are long-established U.K. companies, with Morrisons predominant in the north of England. Morrisons took over Safeway in 2004. The ASDA chain was taken over by Wal-Mart in 1999.

Entry, Exit, and Labor Productivity in U.K. Retailing Table 7.12

297

Business share of chains and independents Frequency shares

Stand-alone shops

Regional chains (1 region)

1997 2003

67.83 66.43

10.57 8.56

1997 2003

74.40 64.93

5.19 4.59

Employment shares

National chains (> 2 regions)

Regional chains (1 region)

National chains (> 2 regions)

28.34 25.82

8.81 6.56

62.85 67.61

SIC521—Nonspecialized 20.41 13.90 30.48 11.41

4.74 2.89

81.36 85.71

Stand-alone shops

All retailing 21.60 25.01

Source: Authors’ calculations from ARD.

Table 7.13

Takeovers versus entry

Year

Frequency share of entry

1998 1999 2000 2001 2002 2003

0.18 0.11 0.11 0.12 0.12 —

1998 1999 2000 2001 2002 2003

0.09 0.08 0.09 0.14 0.23 —

Frequency share of takeovers

Employment share of entry

Employment share of takeovers

0.11 0.08 0.09 0.18 0.14 —

0.06 0.05 0.04 0.04 0.02 0.04

0.05 0.06 0.05 0.28 0.18 —

0.11 0.04 0.03 0.05 0.00 0.07

All retailing 0.02 0.03 0.02 0.01 0.01 0.02 SIC 521—Nonspecialized 0.11 0.09 0.08 0.05 0.00 0.10

Source: Authors’ calculations from ARD.

2002 this was down to 56 percent of entry. The diﬀerence was made up by national chains, whose share of entry was 17 percent in 1998 but 42 percent in 2002 (see rows 5 and 6). Similar data show up for the supermarket sector. So far we have seen a tendency towards dominance by national chains. What has this done to their store profile? Table 7.15 sets out some evidence for this. First, we rank all firms by their size and split them into quantiles. Note we choose firms here so that small stand-alone shops are a firm, but a large chain is one firm. Thus, the large chains are at the top of this distribution. Then we calculate, for each quantile, the fraction of shops in that

298

Jonathan Haskel and Raﬀaella Sadun

Table 7.14

Shares by number of stores of entry and exit accounted for by stand-alone, regional, and national chains

Type

All retailers entrants (%)

All retailers exitors (%)

Nonspecialized entrants (%)

Nonspecialized exitors (%)

1998 2002

Stand-alone shops Stand-alone shops

74.10 56.13

70.84 82.61

71.12 51.42

68.15 80.54

1998 2002

Regional chains Regional chains

8.66 2.38

8.62 6.98

3.88 1.84

2.81 3.24

1998 2002

National chains National chains

17.25 41.48

20.54 10.41

25.00 46.74

29.04 16.21

Source: Authors’ calculations from ARD.

quantile who are small, where small is defined as the shops below the 1997 median of three-digit industry employment of the shop (or local unit). Thus, the top cell of table 7.15, 93.12, says that 93.12 percent of firm employment of firms in the 1st (lowest) quantile in 1997 is in shops who are small. Looking down the table at the 3rd quantile, who are the biggest firms, the fraction of shops in their firms who are small is 48 percent in 1997 and 47 percent in 2003. The last rows in the panel show averages for firms in the 75th to 94th and 95th to 100th percentile of the distribution, and again, these changes are small. This seems like a small change in the light of the changes that we saw previously in the focus on small stores. Thus, we explored this further by looking at SIC521, the results of which are set out in the lower panel. The results for the 3rd quantile are most interesting. In 1997, 70.69 percent of employment in the largest quantile by firm was in small shops, whereas in 2003, 80.3 percent of employment was. Both mean and median employment have dropped only slightly, but recall that since many of the smaller shops are so small relative to the large scale stores the diﬀerence in average employment is not likely to show up very dramatically. Looking at the lower panel, lower two rows, we can see that the move to large firms taking over small shops is apparent in the 95th to 100th percentile, where mean and media shop employment have fallen and the standard deviation of employment has risen. Thus, we can see in these data an increase in the fraction of small shops in large chains, in agreement with the data in Griﬃth and Harmgart (2005). We have seen, then, a rise in the proportion of small stores in larger firms. What possible impact on productivity levels might this have? First, consider the gap to be explained. Figure 10 of Baily (1993) estimates the 1987 productivity levels gap at 18 percent for the United Kingdom versus the United States, using value added per employee, at Purchasing Power Parity (PPP). But Baily points out that the productivity diﬀerence in like

Entry, Exit, and Labor Productivity in U.K. Retailing

Table 7.15

299

Share of small shops

Year

Position in the distribution

Employment share

1997 2003 1997 2003 1997 2003 1997 2003 1997 2003

1st quantile 1st quantile 2nd quantile 2nd quantile 3nd quantile 3rd quantile 75th–94th percentile 75th–94th percentile 95th–100th percentile 95th–100th percentile

93.12 94.68 76.27 78.33 48.07 47.24 32.98 30.65 25.66 21.01

1997 2003 1997 2003 1997 2003 1997 2003 1997 2003

1st quantile 1st quantile 2nd quantile 2nd quantile 3nd quantile 3nd quantile 75th–94th percentile 75th–94th percentile 95th–100th percentile 95th–100th percentile

97.02 98.01 95.02 97.41 70.69 80.30 32.78 38.79 15.16 27.54

Frequency share All retail 93.77 95.73 83.75 85.55 60.69 60.31 48.62 45.80 38.64 32.98

Mean shop Median shop Sd of shop employment employment employment

2.71 2.65 4.39 4.30 6.10 6.11 9.59 9.92 30.64 30.71

2.69 2.63 4.32 4.24 5.93 5.94 8.70 8.98 25.62 26.25

1.01 1.01 1.76 1.73 2.68 2.76 5.64 5.69 22.94 23.93

521—Nonspecialized 96.52 3.90 96.83 3.82 94.69 7.15 96.77 6.64 76.40 13.52 85.36 12.46 47.81 38.94 52.58 39.87 29.21 195.26 39.32 121.84

3.90 3.80 7.12 6.62 13.42 12.25 33.82 36.49 176.23 102.82

1.01 1.04 2.04 2.24 4.43 3.79 27.64 27.98 195.38 99.26

Notes: The table is constructed by ranking all firms (reporting units) into size quantiles and then calculating for each quantile the fraction of shops (local units) in that quantile who are small, where small is defined as the shops below the 1997 median of three-digit industry employment of the shop. Source: Authors’ calculations from ARD.

stores is much less (i.e., when control for the store mix of products productivity is similar), suggesting that a lot of this is diﬀerent mixes of stores. Department stores are about the same productivity (the United States lead about 5 percent in multicategory stores) but lead is 25 percent in single category stores (e.g., Home Depot). The value added per hour worked gap was from Griﬃth and Harmgart (2005), citing the EUKLEMS study, 46 percent in 2001. Second, there are two potential impacts of size on productivity. The first is simply the impact of total size of shops, which changes productivity via economies of scale. The second is the impact of the sizes of shops within a chain, which has an economy of scale eﬀect, but also an economy of scope eﬀect if the organizational capital required to run hitherto large stores cannot be perfectly substituted to running small stores. On the first eﬀect, the average size of stores is 14.42 from Jarmin, Klimek, and Miranda (2005), comparable with 9.73 from us for 2003, a log diﬀerence with the United States of 39 percent. How much of the productivity gap can this ac-

300

Jonathan Haskel and Raﬀaella Sadun

count for? It depends on the returns to scale, if any, in shop size. Table 7.10 shows that larger stores have a higher value added per full-time employee. To explore this further we regressed, for all reporting units, 1998 to 2003, log value added per FTE on log FTEs, plus dummies for regions, and year and four-digit industry interacted. We obtained a coeﬃcient on log FTE employment of 0.061 (t 20.80, 38, 910 observations, R2 0.10). When we ran the same regression for only single-unit reporting units we obtained a coeﬃcient on log FTE employment of 0.039 (t 6.74, 29,390 observations, R2 0.10). This, then, is consistent with increasing returns to scale (although we stress that we have not controlled, due to data availability, for other inputs such as capital) with a 1 percent increase in employment, raising productivity by 4 to 6 percent (so that a 100 percent increase in employment raises output by 107 percent). Thus, a 39 percent diﬀerence in employment would give a 0.39 0.05 1.95 percent increase in productivity (taking a 5 percent returns to scale figure). This is 4 percent of the 46 percent productivity gap, which seems rather small, although this gap estimate is the largest. The second (within-chain) eﬀect is analyzed in Haskel and Sadun (2007), where we document that average store size within a chain has, indeed, an important role on chain-level TFP. According to our firm-level estimates, the fall in within-chain shop sizes lowered annual TFP growth in U.K. retailing by 0.2 percent. This is about 20 percent of the post-1995 slowdown in U.K. retail TFP growth of about 1 percent, documented by Basu et al. (2003). 7.7 Conclusions We have used a new micro-level data set to study productivity in U.K. retailing, 1997 to 2003. We have used store-level data to look at concentration and entry and exit, but, due to data limitations, chain of store-level data to look at productivity and productivity growth. Among our findings are: 1. In 2003, there were 285,291 stores in U.K. retailing and 196,286 firms/ chains. But just 171 chains accounted for 60 percent of total retail employment. 2. Entry/exit/stayers are fairly stable fractions of all stores, being about 12 percent, 12 percent and 70 percent (the rest are stores who survive one year). 3. Productivity levels are strongly aﬀected by whether productivity is measured by heads or full-time equivalents. 4. Labor productivity is higher in larger stores, especially so in supermarkets. 5. The variation in labor productivity across retailers is rather larger than in the United States.

Entry, Exit, and Labor Productivity in U.K. Retailing

301

6. Data diﬀerences with the United Kingdom make comparisons hard, but the contribution of entry and exit to productivity growth is somewhat smaller than in the United States. 7. There was a change in planning regulations in 1996 that greatly stopped retailers developing out-of-town shops. This has little discernible eﬀect on the retailing stores as a whole but a noticeable eﬀect on supermarkets, where the average size of stores in the largest chains has fallen as large chains operate an increasingly large fraction of small stores. 8. U.S. stores are on average 39 percent larger than U.K. stores (in terms of employment) and so the increases in preponderance of small stores might be expected to lower the productivity of U.K. retailing if there are increasing returns to scale in retailing. Recent research suggests that this seems indeed to be the case.

References Baily, M. 1993. Compassion, regulation and eﬃciency in service industries. Brookings Papers on Economic Activity, Issue no. 2:71–130. Washington, D.C.: Brookings Institution. Basu, S., J. Fernald, N. Oulton, S. Srinivasan. 2003. The case of the missing productivity growth: Or, does information technology explain why productivity accelerated in the United States but not the United Kingdom? In NBER macroeconomics annual 2003, Volume 18, ed. Mark Gertler and Kenneth Rogoﬀ, 9–63. Cambridge, MA: MIT Press. Barnes, M., and Martin, R. 2002. Business data linking: An introduction. Economic Trends 581 (April): 34–41. Betancourt, R., and D. Gautschi. 1993. The outputs of retail activities: Concepts, measurements, and evidence from U. S. census data. The Review of Economics and Statistics 75 (May): 294–301. Betancourt, R., and M. Malinoski. 1999. An estimable model of supermarket behavior: Prices, distribution services, and some eﬀects of competition. Empirica 26:55–73. Bosworth, B. P., and J. E. Triplett. 2003. Productivity in services industries: Trends and measurement issues. Paper presented at Brookings Conference, A Brookings Economic Studies Event. Productivity in services industries: Trends and measurement issues, summary of what we have learned from the Brookings Economic Measurement Workshops. November 21, 2003. Available at http:// www.brookings.edu/es/research/projects/productivity/workshops/ 20031121_chapter4.pdf Bruno, M. 1978. Duality, intermediate inputs, and value-added. Chapter 1 in Production economics: A dual approach to theory and application. Elsevier NorthHolland. Competition Commission. 2000. Supermarkets: A report on the supply of groceries from multiple stores in the United Kingdom—Report Summary. Criscuolo, C., J. Haskel, and R. Martin. 2003. Building the evidence base for productivity policy using business data linking. Economic Trends 600:39–51.

302

Jonathan Haskel and Raﬀaella Sadun

DuMouchel, W. H., and G. J. Duncan. 1983. Using sample weights in multiple regression analysis of stratified samples. Journal of the American Statistical Association 78 (383): 535–43. EU KLEMS Database. 2008. In The EU KLEMS Growth and Productivity Accounts: An Overview, ed. M. Timmer, M. O’Mahony, and B. van Ark. University of Groningen and University of Birmingham. Available at www.euklems.net Foster, Haltiwanger, and Krizan. 2006. Market selection, reallocation, and restructuring in the U. S. retail trade sector in the 1990s. The Review of Economics and Statistics 88 (4): 748–58. Griﬃth, R., and H. Harmgart. 2005. Retail productivity. The International Review of Retail, Distribution and Consumer Research 15 (3): 281–90. Haskel, J., R. Jarmin, K. Motohashi, and R. Sadun. 2007. Retail market structure and dynamics: A three country comparison of Japan, the U.K., and the U.S. Working Paper prepared for National Bureau of Economic Research and the Conference on Research in Income and Wealth. Haskel, J., and N. Khawaja. 2003. Productivity in UK retailing: Evidence from micro data. CERIBA Working Paper. Haskel, J., and R. Sadun. 2007. Entry regulation and productivity: Evidence from the UK retail sector. CERIBA and Center for Economic Performance. Unpublished Manuscript. Jarmin, R., S. Klimek, and J. Miranda. 2005. The role of retail chains: National, regional, and industry results. Working Papers 05-30, Center for Economic Studies, U. S. Census Bureau. Jones, G. 2000. The development of the annual business inquiry. Economic Trends 564 (November). Martin, R. 2004. Globalisation, ICT, and the nitty gritty of plant level datasets. CEP Discussion Papers dp0653, Center for Economic Performance, London School of Economics. Oi, W. Y. 1993. Productivity and the distributive trades. NBER conference on income and wealth, ed. Z. Griliches. Cambridge, MA: National Bureau of Economic Research. Oﬃce for National Statistics. 2001. Review of the inter-departmental business register: National Statistics Quality Review Series Report No. 2. Partington, J. 2001. The launch of the annual business inquiry. Labour Market Trends 109 (5): 25–68.

8 The Dynamics of Market Structure and Market Size in Two Health Services Industries Timothy Dunne, Shawn D. Klimek, Mark J. Roberts, and Daniel Yi Xu

8.1 Introduction The relationship between the size of a market and the structure of production—the number of firms, their relative size, and the magnitude of entry and exit flows—is determined by a large set of underlying structural factors, including the competitiveness of the market, the magnitude of sunk entry costs, the importance of economies of scale in production, the relationship between production cost and product quality, and the magnitude of cost heterogeneity among producers. A number of empirical studies in industrial organization have used products that are sold in small geographic markets to study the crosssectional relationship between market size and market structure. Studies of the relationships between market size, generally measured as population, and the number of firms in the market (Bresnahan and Reiss 1989, 1991; Asplund and Sandin 1999; Berry and Waldfogel 2003), the average sales of the firms (Campbell and Hopenhayn 2005), and the magnitude of cost heterogeneity (Syverson 2004) have been used to indirectly draw inferences Timothy Dunne is a senior economic advisor in the Research Department at the Federal Reserve Bank of Cleveland. Shawn D. Klimek is a senior economist at the Center for Economic Studies at the U. S. Census Bureau. Mark J. Roberts is a professor of economics at the Pennsylvania State University, and a research associate of the National Bureau of Economic Research. Daniel Yi Xu is an assistant professor of economics at New York University. This chapter reports the results of research and analysis undertaken by the U. S. Census Bureau staﬀ. It has undergone a Census Bureau review more limited in scope than that given to oﬃcial Census Bureau publications. This report is released to inform interested parties of ongoing research and to encourage discussion of work in progress. We are grateful to Steve Berry for helpful discussion and comments on this research. The views stated herein are those of the authors and are not necessarily those of the Federal Reserve Bank of Cleveland, the Board of Governors of the Federal Reserve System, or the U. S. Census Bureau.

303

304

Timothy Dunne, Shawn D. Klimek, Mark J. Roberts, and Daniel Yi Xu

about these underlying structural factors, particularly the degree of competition in the market.1 A common factor in virtually all of the empirical literature is that they are based on a two-period long-run equilibrium model that explains the number and size of firms as a function of market size. Each paper tends to focus on the cross-sectional correlation between one aspect of market structure and market size. The data and theoretical framework used in these studies are not well suited to examining entry and exit flows or to identifying the role that sunk entry costs might play in the evolution of market structure. More recently, explicit dynamic models have been developed that generate a relationship between market size and firm turnover (Asplund and Nocke 2006; Pakes, Ostrovsky, and Berry 2007). These dynamic models distinguish incumbent producers from potential entrants and recognize that, when sunk entry costs are present, the value functions of the two groups are diﬀerent. This makes an incumbent’s decision to remain in operation or exit diﬀerent from the decision of a potential entrant and leads to a framework in which market history, specifically past market structure, is a determinant of current market structure. With the exception of Bresnahan and Reiss (1994), the role of past market structure has not been examined in the empirical market structure literature. In this chapter we utilize data from the U. S. Census of Service Industries to study the evolution of market structure in two health services industries, dentists and chiropractors. We use data for the period 1977 to 2002 to document a set of empirical facts linking the number of firms—and the flows of entering and exiting firms—to both market size and past market structure. It is particularly interesting to examine the market structures of health service industries because the market demand is closely tied to population, so that market size should be important, and there are substantial sunk costs involved in establishing a practice, so that the history of market structure should also be a significant determinant. The empirical results indicate that past market structure, specifically the number of firms in the market in previous periods and the number of potential entrants to the market, play an important role in determining the flow of entering and exiting firms. Together these imply that market history is a significant determinant of current market structure, as the dynamic models of entry and exit imply. The inclusion of lagged market structure also leads to a large reduction in the role of current market size and thus would have a significant impact on conclusions about market competition that are based on the two-period long-run equilibrium models. In the next section of the chapter we review the theoretical arguments 1. Berry and Reiss (2006) summarize this literature and discuss the modeling assumptions needed to separately identify the degree of competition in the market from other structural factors, particularly the magnitude of fixed costs.

Market Structure and Size in Two Health Services Industries

305

and empirical findings from the two-period long-run models and then contrast them with an explicitly dynamic model of entry and exit. The third section summarizes the market-level data we utilize, focusing on the number of dentists and chiropractors in geographic markets in the U. S. The fourth section summarizes the empirical model of entry and exit we estimate, and the fifth section presents the empirical results. 8.2 Models of Entry, Exit, and Market Structure 8.2.1 The Number of Firms The primary model that has guided empirical work on entry and exit is first outlined in a series of papers by Bresnahan and Reiss (1989, 1991) and Sutton (1991). There are two time periods. In the first period, a large group of ex ante identical potential entrants make a decision to enter the market after paying a fixed cost to enter. In the second period, production occurs and profits are realized. The second period profits are determined by the nature of competition in the market (e.g., Cournot versus Bertrand versus collusion) and the number of firms that entered in the first stage. In equilibrium, the number of firms that enter will be determined by a zero net profit condition; entry will occur until the second-stage profits fall below the fixed entry cost. What Bresnahan and Reiss and Sutton show is that the zero profit condition can be used to guide empirical work explaining the number of firms in the market. In the simplest version of this framework, all firms in a market are identical. Let Z represent a set of exogenous market-level variables that determine demand and cost conditions in a market such as the number of consumers and factor prices. Let V(N,Z) be the profits earned by each firm in a market when there are N producers. If is the common fixed cost of entry then the equilibrium number of firms N∗ can be described by two entry conditions: (1)

V(N ∗,Z) and V(N ∗ 1,Z) .

In a market with N ∗ firms, profits for each will cover the fixed cost, while in a market with N ∗ 1 firms they will not. Almost all empirical applications of this framework have used data on N and Z from a cross-section of geographic markets to estimate parameters of the profit function, particularly the eﬀect of a change in the number of firms on profits, and the fixed cost. If we assume that the fixed cost in each market is an independent draw from a common normal distribution for , then the equilibrium entry conditions imply an ordered probit model for the number of firms with Z and any variables that shift the distribution of as the explanatory variables in the model. Berry and Reiss (2006) and Berry and Tamer (2006) discuss the assumptions on V and that are necessary to estimate the parameters of

306

Timothy Dunne, Shawn D. Klimek, Mark J. Roberts, and Daniel Yi Xu

the profit function and fixed cost distribution in this homogenous firm framework. In general, they show that if only N and Z are observed and profit data is not observed, then distributional assumptions on and functional form restrictions on V are needed to estimate the profit and fixed cost parameters. A particularly interesting special case, which has played a large role in the empirical studies, is where the profit function can be written as the product of a per-customer profit function V C(N,Z ) and the number of customers S: V(N,Z ) V C(N,Z )S. In this case, S is interpreted as a measure of market size and so cross-sectional variation in market size generates cross-sectional variation in firm profits. In this case, the empirical relationship between N and S can be used to draw inferences about the competitive eﬀect of entry, that is, the eﬀect of N on the profit function V, without using profit data. If this competitive eﬀect is present in a market, then entry of additional firms into a market compresses the average markup of all firms in operation, lowering V. At the entry stage the market size needed to support an additional firm will be larger than if this competitive eﬀect is absent. Alternatively, larger markets will support more firms but will also have a larger average market size per firm (S/N ). This competitive eﬀect can show up in the market size correlations in other ways as well. Campbell and Hopenhayn (2005) develop the implications of increased market size on the average size of firms in the market. If larger markets are more competitive and hence have lower markups, then average firm size will be larger because the firms must sell more output to cover their fixed costs. They find evidence of this correlation in thirteen U. S. retail industries. Syverson (2004) incorporates firm heterogeneity into the two-period framework.2 Firms are allowed to diﬀer in marginal costs and he shows that competitive eﬀects can be reflected in the distribution of costs or productivities in a market. In this case a homogeneous product is produced by plants with diﬀerent marginal cost. Product diﬀerentiation is introduced through the spatial dispersion of customers and the presence of high transport costs. Together these make each producer’s output an imperfect substitute for the output of others. An increase in demand density (the number of customers per unit of area) leads to an increase in producer density, which, in turn, lowers prices and profit margins for all plants in the 2. A number of other papers have incorporated firm heterogeneity in the two-period framework. In his study of airline markets, Berry (1992) allows for diﬀerences in fixed costs across firms and models the number of firms as a function of market and firm characteristics. He finds that average firm profits are negatively aﬀected by an increase in the number of producers. Mazzeo (2002) and Seim (2006) allow for diﬀerent degrees of product diﬀerentiation across firms within the same market. Mazzeo models the number of high-quality and lowquality firms in a market and finds significant own and cross-eﬀects of the number of firms of each type on the average profits of each type. Seim allows firms to diﬀer in their geographic location within the market and studies the location decision of new firms. She finds that increasing distance between firms insulates them from the competitive eﬀects.

Market Structure and Size in Two Health Services Industries

307

market and raises the failure rate of high-cost producers. As a result, more densely populated markets will be more eﬃcient, having a higher proportion of low-cost producers. He finds empirical evidence of higher eﬃciency in larger markets for the U. S. ready-mix concrete industry. Asplund and Nocke (2006) move beyond the two-period framework and develop a dynamic equilibrium model in which market size has implications for the rate of firm turnover. The underlying competitive mechanism is similar to these other papers: an increase in competition as market size increases results in large markets having more firms with higher per-firm sales but lower price-cost margins. This results in the marginal surviving firm being more productive in larger markets, which is reflected in higher turnover and a younger age distribution of firms in larger markets. They find empirical evidence supporting this for Swedish hair salons. 8.2.2 Entry and Exit Flows With the exception of Asplund and Nocke (2006), the empirical literature summarized in section 8.2.1 focuses on long-run diﬀerences in the number of firms, not on entry and exit flows directly. While the underlying two-period framework can allow for producer heterogeneity in fixed costs or profits, which leads one firm to choose to be in the market while another chooses to be out, it does not provide any insights into what determines the magnitude of entry and exit flows. One aspect that is generally missed in the two-period model is the distinction between the role of fixed costs that all producers pay and the role of sunk entry costs that are only paid by firms at the time of entry.3 This leads to a diﬀerence in the objective function and participation decision of incumbent and potential entrant firms. Incumbents compare the expected sum of discounted future profits with the scrap value they would earn by liquidating the firm. In contrast, potential entrants compare the discounted future payoﬀ from entering with the sunk entry cost they must incur. This distinction has important implications for the way that the number of firms responds to exogenous factors that change profits. Sunk entry costs combined with uncertainty about future market conditions gives rise to hysteresis in market structure (Dixit and Pindyck 1994). For example, suppose there is an exogenous increase in market size that raises firm profits suﬃciently to induce potential entrants to pay the sunk cost and enter the market. If the market size and profits then return to their initial levels, those new firms may find it profitable to remain in operation rather than exit. The number of firms thus responds asymmetrically to changes in market size. Equivalently, the history of market structure, and 3. Berry (1992) allows the fixed cost of an airline on a route to depend on whether or not the airline had a presence at the endpoint cities, which distinguishes incumbents from potential entrants in the endpoint markets.

308

Timothy Dunne, Shawn D. Klimek, Mark J. Roberts, and Daniel Yi Xu

not just current and future profit determinants, matters in explaining the current number of firms. The level of the sunk entry cost also aﬀects the magnitude of entry and exit flows in the market. Using a competitive, industry-equilibrium model, Hopenhayn (1992) shows that an increase in the sunk entry cost will reduce the flow of entering firms in the market but also reduce the flow of exiting firms. The higher entry cost acts to insulate incumbent firms from the pressure of entry and allows more ineﬃcient incumbents to survive. Thus, the entry cost is an important structural element aﬀecting entry and exit flows and the degree of market eﬃciency. Recently, fully dynamic models that recognize the distinction between incumbent and potential entrant firms have been developed that can explain diﬀerences in entry and exit flows across markets and/or industries. Pesendorfer and Schmidt-Dengler (2003), Aguirregabiria and Mira (2007), Das, Roberts, and Tybout (2007), Collard-Wexler (2006), and Ryan (2006) all use micro data on firm participation patterns in a market to estimate structural models of entry and exit, including the sunk costs of entry. Pakes, Ostrovsky, and Berry (2007) develop a dynamic model that is very useful as a framework for studying the flows of entering and exiting firms in market-level data. This leads to a formulation for regressions of entry, exit, and the number of firms that can be distinguished from the two-period models but estimated with the same type of cross-sectional or panel marketlevel data. We will briefly summarize this model and then use it to specify regression models of entry and exit. The model assumes that in a market each firm earns identical profits given by (N, Z ) where N is the number of firms that operate in the period and Z is a set of exogenous cost and demand shifters. The state variables Z evolve exogenously over time as market conditions change, while N evolves endogenously as firms make optimal entry and exit decisions. In each period an incumbent faces a choice of remaining in the market in the next period or exiting. If firm i chooses to exit they earn a scrap value i, which is modeled as an independent draw from an underlying distribution F (• | ) where is a parameter that characterizes the distribution. The distribution is common for all firms and time periods. If they remain in they earn expected profits VC(N, Z, ), which is the expectation of the firm value in the next period and is identical for all incumbents in the market.4 Incumbent firm i chooses to remain in the market if VC(N, Z, ) i . Similarly, each potential entrant faces a decision to enter at the start of the next period. The payoﬀ from entering is represented as VE(N, Z, ) and is the same for all potential entrants in the market. Each potential entrant diﬀers in their entry cost i , which is modeled as an independent draw from a common distribution F (• | ), where the parameter characterizes the entry cost distribution. The firm enters the market if VE(N,Z,) i. 4. The expectation is over the future values of the state variables N, Z, and the scrap value .

Market Structure and Size in Two Health Services Industries

309

This framework diﬀers from the two-period model outlined above in several important ways. First, incumbents and potential entrants diﬀer. The latter must pay an entry cost, but also the expected firm value from continuing in production VC is diﬀerent than the expected firm value from entering VE.5 Second, firm-level heterogeneity, which is absent from the simplest two-period models, is introduced through the random scrap value and entry cost. This model is capable of generating simultaneous flows of entering and exiting firms into the same market, something that simpler models with homogenous firms cannot do. This model results in simple expressions for the probability of entry and exit. The probability an incumbent firm exits from a market with current state N,Z is given by: (2)

PX (N,Z,) Prob [VC(N,Z,) ] 1 F [VC(N,Z,)|].

Since this probability is the same for all incumbents in the market, the numbering of exiting firms NX is a binomial random variable with the parameters PX and N, where the number of incumbent firms is the number of trials in the process. Similarly, the number of entering firms is also a binomial random variable. The probability of one firm entering the market with current state (N, Z) is: (3)

PE (N,Z,, ) Prob [VE(N,Z,) ] F [VE(N,Z,)| ].

This probability is the same for all potential entrants to the market, so the number of entering firms NE is a binomial random variable with parameters PE and NPE, where the latter is the number of potential entrants to the market. To summarize, the theoretical model provides a basis for a statistical model of the number of entering and exiting firms in a market. The entry and exit flows (NX and NE ) over a time period are a function of exogenous state variables Z that aﬀect profits in the beginning time period (and determine values in future time periods) and the number of firms N and number of potential entrants NPE at the beginning of the period. In section 8.4 we estimate equations for the entry and exit flows based on this specification. 8.3 Measuring Entry and Exit for Dentists and Chiropractors Using Census Data The data used in the analysis are from the U.S. Census Bureau’s Longitudinal Business Database (LBD). The LBD contains establishment-level data on all employers in the United States from 1976 through 2005. The 5. VE(N,Z,) and VC(N,Z,) are not identical. The former is the expected firm value from the perspective of a firm that chooses to enter, and thus knows there is at least one entering firm in the market. The latter is the expected firm value from the perspective of an incumbent that chooses to remain in the market and thus knows there is at least one survivor. Each group thus has a diﬀerent expectation for the number of firms N in the market in future periods.

310

Timothy Dunne, Shawn D. Klimek, Mark J. Roberts, and Daniel Yi Xu

database allows for the measurement of establishment and firm dynamics across almost all sectors of the U. S. economy.6 While prior research focused primarily on the analysis of firm dynamics in the manufacturing sector, the recent development of nonmanufacturing data allows for the analysis of producer turnover across a much wider range of industries. In this chapter, we examine entry and exit in two health services industries—dentists (NAICS 621210) and chiropractors (NAICS 621310)— where little is known about the patterns of firm dynamics. Dental and chiropractic services are dominated by small, single location firms typically owned by the practicing doctor(s). While Census data collection is establishment-based, for these industries virtually all firms are single establishment practices, particularly in the small markets we will utilize, and we use the terms interchangeably in this chapter. These firms provide their services in relatively small markets with the demand for services tied closely to local market conditions, particularly population. The technologies are also similar across dental and chiropractic establishments in that they combine oﬃce staﬀ, specialized capital equipment, and doctors’ time to provide health services. Our analysis augments the LBD with revenue, payroll, employment, and geographic coding data from the Census of Services, limiting the data set to the Census years of 1977, 1982, 1987, 1992, 1997, and 2002. The remainder of this section discusses market definitions, the measurement of entry and exit, and the construction of market-level variables. 8.3.1 Defining Markets and Market Participation Throughout this chapter, we focus our attention on small and isolated geographic areas so that we can better define the market served, similar in spirit to the approach taken by Bresnahan and Reiss (1989, 1991). We first identified a set of cities and towns that are geographically distinct from large population centers. From this list of potential markets, we kept only those locales with populations of less than 50,000 and consistent place coding in the Census of Services over time. Our markets include 754 incorporated places that vary in population from 2,500 to 50,000 people, and are, on average, larger than the locales used by Bresnahan and Reiss. All 754 geographic areas had a dental practice present in at least one year; but because they require a larger population to sustain a practice, only 689 of the geographic areas had a chiropractic practice present. The measure of entry used in this chapter is the entry of an establishment into one of these geographic markets. An entrant in a market is defined as an establishment that is not producing in market m in period t but is producing in market m in period t 5 (the next Census year).7 An exit is simi6. Jarmin and Miranda (2002) discuss the measurement issues involved in constructing the LBD. 7. Almost all entering and exiting establishments in these data represent establishment birth and deaths. However, some establishments in the data switch geographic codes over time

Market Structure and Size in Two Health Services Industries

311

Fig. 8.1 Cumulative distribution of the number of dentists and chiropractors in local markets Source: Authors’ own calculations, Census of Services 1977–2002.

larly defined as an establishment that is in market m in period t and is not in that market in period t 5. For each market m and in each time interval, we construct the number of entering establishments (NEmt), the number of exiting establishments (NXmt), and the number of establishments (Nmt). The data allow us to measure entry and exit for five time intervals (1977–1982, 1982–1987, 1987–1992, 1992–1997, and 1997–2002) and for 754 geographic markets, yielding a data set of 3,770 market-time observations. To give a sense of how entry and exit varies across the markets in our data, figures 8.1–8.3 show the cumulative density of the number of establishments, number of entrants, and number of exiting establishments. The distribution for dentists is shifted to the right in all three panels, indicating that the number of oﬃces per market, as well as the number of entering and exiting producers, is larger for dentists compared to chiropractors. The graph also shows that many of our markets have a relatively small number of producers. In fact, the majority of markets support less than three and such geographic market switching can generate entry and exit under our definitions. We restrict certain types of these geographic market switching. In particular, establishments within a county will sometimes switch between a rest of county place code and a place code identifying a city. We do not allow these within-county changes in geographic coding to generate entry and exit. In these cases, we fix the place code to the code that identifies the city and then treat the establishment as continuing in that location.

Fig. 8.2 Cumulative distribution of the number of entering dentists and chiropractors in local markets Source: Authors’ own calculations, Census of Services 1977–2002.

Fig. 8.3 Cumulative distribution of the number of exiting dentists and chiropractors in local markets Source: Authors’ own calculations, Census of Services 1977–2002.

Market Structure and Size in Two Health Services Industries

313

chiropractors and less than eight dentists. One diﬀerence between these two industries is that the number of chiropractic oﬃces grew rapidly over the period of analysis while dentists experienced slower growth. The average number of dentists oﬃces in our markets grew by 16.5 percent whereas the growth in chiropractors oﬃces increased by 142.5 percent. Still, the number of chiropractic oﬃces was only 31.5 percent of the number of dentists oﬃces by the end of the sample period. 8.3.2 Market-Level Variables Throughout the analysis, three variables are used to characterize market structure: the number of establishments (discussed previously); the average size, measured as real revenue, of producers; and the average labor productivity of producers. We use the data from the Census of Services to measure the average revenue of practices in a market and deflate this by the Consumer Price Index (CPI). Average labor productivity for a market is similarly constructed by taking real revenue of a practice, dividing it by the establishment’s total employment, and then averaging across all producers in the market.8 Our empirical models use three variables to capture diﬀerences in demand and cost conditions across markets. To control for demand diﬀerences, we include the population of the geographic market and per capita income. Population in a market (Smt) has been the main proxy used to measure market size in most previous studies of entry. Population figures on incorporated places are obtained from the Census Bureau’s series on population estimates, but we interpolate the data for our places in some earlier years from the Decennial Census when population estimates are unavailable. The real per capita income variable (PCImt) from the Bureau of Economic Analysis is measured at the county level and deflated by the CPI. To control for cost diﬀerences, we construct the average real wage (Wmt) paid to workers employed in the health practitioners’ oﬃces (NAICS 62111621399) in the county, again deflating by the national CPI. Because we do not have local price deflators, variation in the wage and income variables will also reflect price-level diﬀerences across geographic markets, which are likely to be important in these data. The dynamic entry model from section 8.2.2 implies that the history of market structure matters in determining current market structure. Two variables are used to control for history in our empirical models—the lagged or beginning period number of firms in a market and the lagged or beginning period number of potential entrants (NPEmt) that were present 8. In these industries, dentists and chiropractors are the main generators of revenue, but if the legal form of organization is a sole proprietor or a partnership these owner-practitioners will not be counted in employment. To account for this omission, we modify total employment at an establishment for sole proprietors by adding one employee and for partnerships by adding two employees.

314

Fig. 8.4

Timothy Dunne, Shawn D. Klimek, Mark J. Roberts, and Daniel Yi Xu

Number of producers by market size

Source: Authors’ own calculations, Census of Services 1977–2002.

in the market. The number of potential entrants into a geographic market in a time period is equal to the maximum number of diﬀerent establishments that appear in the market over time minus the current number of establishments in operation. The rationale behind this definition is that in each geographic market we observe all potential entrants being active at some point in time. In each time period the pool of potential entrants is the set of establishments that are not currently active. A main focus of prior work has been an examination of how the number of firms in a market varies with market size. Figure 8.4 graphs this relationship for dentists and chiropractors using a locally weighted regression.9 Larger markets support a larger number of practices in both industries and, as noted previously, the number of dental practices per capita is significantly greater than chiropractic practices per capita. Our largest markets support in excess of twenty dentists per market while for chiropractors the largest markets only support about five producers. Since chiropractic oﬃces also have lower revenue, on average, than dentist oﬃces, the data reflect the fact that per capita demand for chiropractic services is much less than dental services. Figures 8.5 and 8.6 graph the relationship 9. Figures 8.1 through 8.9 are produced using STATA’s lowess command with a bandwidth of .3 and the default weighting procedure. The lowess command estimates a weighted regression at each observation in the data using nearby data points to construct a smoothed value of the dependent variable conditional on the x variable.

Fig. 8.5

Number of entrants by market size

Source: Authors’ own calculations, Census of Services 1977–2002.

Fig. 8.6

Number of exits by market size

Source: Authors’ own calculations, Census of Services 1977–2002.

316

Timothy Dunne, Shawn D. Klimek, Mark J. Roberts, and Daniel Yi Xu

between market population and the number of entering and exiting establishments. The same general patterns hold—there are a greater number of entering and exiting dentists per capita than chiropractors. 8.4 Empirical Model of Entry, Exit, and Number of Firms The dynamic model of section 8.2.2 implies that the number of entering and exiting firms is a function of the market characteristics that aﬀect current and future profits, the distribution of scrap values and entry costs, and the number of firms and potential entrants that are present. In this section we specify the estimating equations that we will apply to the market-level data for dentists and chiropractors. The theoretical model from section 8.2.2 specifies the number of entering and exiting firms as binomial random variables. We could specify the probabilities of entry and exit as functions of the observable state variables and unknown parameters and estimate them using maximum likelihood. The estimation of entry probabilities depends critically on the measurement of the number of potential entrants (NPEmt) and this variable is diﬃcult to measure accurately. Instead, we choose to model the mean number of entering and exiting firms directly and use NPE as one explanatory variable. If it is measured poorly it may still be possible to determine if the other state variables are significant determinants of entry flows. Since the entry and exit flows are count variables, we use an extension of the Poisson model. Given the panel of 754 geographic markets for five five-year periods, we specify the flows of entering and exiting practices using a negative binomial regression model. The mean number of exiting plants is a function of the market-level state variables and the number of potential entrants: (4) E(NXmt |Nmt 1, Zmt 1, NPEmt 1) exp [ 0 1 log Smt 1 2 log (PCImt 1 ) 3 log Wmt 1 4Nmt 1 5NPEmt 1 j Dmj ] where the Dmj is a set of four dummy variables to distinguish the five time periods in the data. The variables are all specified at the start of the time period (t – 1) and the number of exits is measured over the time interval t – 1 to t. The negative binomial model generalizes the Poisson model to allow the variance of the distribution to be greater than the mean using the specification: (5)

Var (NXmt ) E(NXmt )[1 x E(NXmt )] .

This introduces one new parameter X, which is referred to as the overdispersion parameter. The Poisson model is the special case where X 0. A similar equation is specified for the mean entry flow:

Market Structure and Size in Two Health Services Industries

(6)

317

E(NEmt | Nmt 1, Zmt 1, NPEmt 1 ) exp [0 1 log Smt 1 2 log (PCImt 1 ) 3 log Wmt 1 4 Nmt 1 5 NPEmt 1 jDmj ]

where NE is measured over the time interval t –1 to t. We also allow for overdispersion in the entry model using the specification in equation (5). It is important to recognize that the coeﬃcients in these entry and exit flow regressions are combinations of profit function, sunk cost, and entry cost parameters, and these underlying structural parameters are not separately identified. Combining the entry and exit models also provides a way to describe the number of firms in year t that is consistent with the dynamic model of entry and exit. The number of firms in year t can be written as Nmt Nmt–1 NEmt – NXmt. Using equations (4) and (6) for NE and NX we can write Nmt as a function of the state variables, lagged number of firms, and number of potential entrants: (7)

E(Nmt | Nmt 1, Zmt 1, NPEmt 1) exp [0 1 log Smt 1 2 log (PCImt 1) 3 log Wmt 1 4Nmt 1 5 NPEmt 1 j Dmj].

Notice that this diﬀers from the specification of N from the two-period models because it depends on lagged N and the number of potential entrants NPE. One way to distinguish the two-period and fully dynamic models is by whether these two variables are significant in the model for the number of firms. We have estimated both negative binomial and Poisson models and in most cases the amount of overdispersion is relatively small and the coeﬃcient estimates are similar. To simplify the discussion, we report only the negative binomial estimates of equations (4) through (7) in the next section. 8.5 Empirical Evidence on Entry, Exit, and Market Structure 8.5.1 Market Structure and Market Size Before turning to the regression analysis, we present graphs showing the relationship between market size and market structure for our health service industries. Figures 8.7 through 8.9 depict a set of locally-weighted regressions for three variables that describe features of our local markets— population per firm (S/N ), average revenue per firm, and average firm productivity. These variables, measured in logs, are plotted against log S in the market. For both industries, average revenue per producer and average labor productivity rise as market size increases, though there are some

Fig. 8.7

Log of population-per-producer by market size

Source: Authors’ own calculations, Census of Services 1977–2002.

Fig. 8.8

Log of average revenue per producer by market size

Source: Authors’ own calculations, Census of Services 1977–2002.

Market Structure and Size in Two Health Services Industries

Fig. 8.9

319

Log of average labor productivity

Source: Authors’ own calculations, Census of Services 1977–2002.

diﬀerences in the shape of this relationship between the industries. The rise in average revenue and average labor productivity is greatest for small markets and then flattens out in the larger markets for dentists. However, for chiropractors, the rise in average revenues and labor productivity is most pronounced in mid-size markets. Recall that small markets support very few chiropractors and it is only as markets become mid-sized that we are likely to see multiple chiropractors operating. These increases in average revenue and average labor productivity with market size are consistent with previous empirical research. Campbell and Hopenhayn (2005) find that average sales per producer rises with market size and Syverson (2004) reports that productivity is higher in larger markets. Both interpret these patterns as consistent with more intense competition in larger markets. The relationship between log(S/N) and log S shown in figure 8.7 does not provide a uniform picture across the industries. For chiropractors, population per producer rises as market size increases. Again, under the conditions described in Berry and Reiss (2006), this pattern is consistent with more competition in larger markets. Alternatively, dentists show a much more muted rise in average population per practice as market size increases, suggesting less of a competitive eﬀect. To control for other variables, table 8.1 reports the coeﬃcients from regressions of the number of producers, the average revenue of practices, and average labor productivity in a market on our demand and cost variables.

320

Timothy Dunne, Shawn D. Klimek, Mark J. Roberts, and Daniel Yi Xu

Table 8.1

Results for market structure regressions Dentists

Constant Log market population Log per capita income Log market wage yr = 1982 yr = 1987 yr = 1992 yr = 1997 Alpha N R-square a

Chiropractors

Number Log average Log labor of firmsa revenue productivity

Number of firmsa

Log average revenue

Log labor productivity

–5.593 (.472) .830 (.013) .416 (.051) .733 (.048) .141 (.022) .026 (.023) –.015 (.024) –.132 (.026) .054 (.004)

1.116 (.423) .108 (.010) .287 (.044) .413 (.042) –.079 (.018) .005 (.019) .153 (.019) .253 (.024) —

1.825 (.274) .053 (.007) .132 (.028) .212 (.028) –.241 (.012) –.220 (.012) –.163 (.012) –.140 (.016) —

–11.265 (.773) .671 (.020) 1.032 (.084) .122 (.074) .266 (.044) .420 (.042) .572 (.041) .597 (.044) .017 (.012)

–.821 (.695) .137 (.019) .461 (.078) .278 (.067) –.071 (.037) .125 (.034) .208 (.032) .082 (.035) —

.733 (.534) .074 (.014) .230 (.059) .187 (.053) –.199 (.030) –.042 (.026) –.020 (.026) –.106 (.028) —

3,762 .189

3,740 .299

3,739 .212

3,762 .145

3,052 .113

3,052 .066

Use negative binomial model for the number of firms. Remaining regressions are OLS.

The regression of the number of producers is estimated using the negative binomial count data model for comparability with the dynamic model results that are reported below. The log average revenue and log labor productivity models are estimated using ordinary least squares, and all models contain year dummies. The coeﬃcient on market population in the regression for number of establishments (N), is an estimate of the elasticity of N with respect to S and is less than one for both industries, .830 for dentists and .671 for chiropractors. If a competitive eﬀect is present, then the coeﬃcient on market size should be less than one. As observed in figure 8.7, chiropractic services appear to have a somewhat stronger competitive eﬀect. One reason for this diﬀerence between these industries is that, compared to dentists, a much higher percentage of the markets served by chiropractors have two or fewer producers (see fig. 8.1). Bresnahan and Reiss found that for their industries the competitive eﬀect was dissipated by the time there were three firms in the market. Given the relatively small number of chiropractors operating in many of the markets, it is more likely to find deviations from competitive outcomes in this industry.

Market Structure and Size in Two Health Services Industries

321

The positive relationship between market size and both revenue and labor productivity in these industries can also be interpreted as evidence that a competitive eﬀect of entry is present. The estimated elasticity of average revenue is .108 in dentists and .137 in chiropractic services. Although these elasticities are somewhat greater than those reported by Campbell and Hopenhayn (2005) using similar models, it is consistent with their argument that a competitive eﬀect will result in larger firms in larger markets. The larger magnitudes found here may reflect the fact that the markets used in this study are generally smaller and thus more likely to be aﬀected by competitive pressure from entry. Syverson’s (2004) model predicts that productivity will be higher in larger markets due to a more intense selection eﬀect, and the results are consistent with his findings. The elasticity of average productivity with respect to market size is .053 and .074 in the dentist and chiropractor industries, respectively. The other demand variable, the log of per capita income, has a positive elasticity in all three market structure regressions for both industries. These results probably reflect the fact that as incomes rise a greater share of the population uses the services, and/or consumers use the service more intensively, resulting in higher market demand, more firms, and higher average revenue. The latter eﬀect could also arise from increased product differentiation and higher prices in wealthier markets. The wage variable has the expected positive sign in the average revenue and productivity regressions, since revenue must rise to cover higher costs, but the positive coeﬃcient in the regressions for the number of producers is unexpected. This suggests that the market wage variable in these regressions not only picks diﬀerences in factor costs across locations but also diﬀerences in cost-ofliving. Markets with higher cost of living will have higher output prices and the net eﬀect of these input and output price changes on profitability is ambiguous. Finally, the alpha parameter estimated from the negative binomial models of the number of producers indicates that overdispersion is present in the model for dentists but not for chiropractors. 8.5.2 Entry and Exit The regression specifications in table 8.1 are motivated by the twoperiod models of entry summarized in section 8.2.1. The empirical results for the eﬀect of market size are similar to other empirical studies using this framework and these have generally been interpreted as reflecting a competitive eﬀect of entry. The dynamic model of section 8.2.2 provides a diﬀerent starting point for the modeling of market structure and indicates that the lagged number of firms and number of potential entrants are determinants of entry and exit flows and thus current market structure. Tables 8.2 and 8.3 provide results for models that include these additional explanatory variables. Table 8.2 reports the results of negative binomial models using the num-

322

Timothy Dunne, Shawn D. Klimek, Mark J. Roberts, and Daniel Yi Xu

Table 8.2

Models of the number of entrants and exits (negative binomial) Dentists

Constant Log market population Log per capita income Log market wage yr = 1982 yr = 1987 yr = 1992 yr = 1997 N(t) Potential entrants NPE(t) A1pha N R-square

Chiropractors

Entry

Entry

Exit

Exit

Entry

Entry

Exit

Exit

–6.331 (1.120) .839 (.027) .336 (.120) .925 (.090) –.344 (.040) –.723 (.043) –.689 (.048) –.751 (.053) —

–7.678 (.905) .847 (.025) .487 (.099) .698 (.087) .024 (.042) –.214 (.043) –.260 (.047) –.234 (.049) —

.164 (.035)

–4.144 (1.038) .295 (.043) .233 (.118) .120 (.110) .274 (.061) .324 (.061) .475 (.069) .657 (.069) –.005 (.011) .115 (.010) .088 (.022)

–5.403 (1.674) .510 (.044) .309 (.186) .310 (.163) –.073 (.077) –.130 (.079) .185 (.075) .415 (.081) —

.090 (.014)

–3.112 (.769) .300 (.031) .208 (.083) .209 (.081) –.057 (.039) –.218 (.040) –.250 (.042) –.149 (.044) .050 (.003) .004 (.002) .005 (.009)

–9.388 (1.179) .684 (.032) .700 (.131) .416 (.116) .205 (.058) .090 (.059) .124 (.065) .183 (.063) —

.187 (.019)

–2.019 (.869) .338 (.032) .069 (.094) .408 (.077) –.307 (.035) –.578 (.038) –.535 (.040) –.490 (.045) .007 (.003) .043 (.003) .062 (.010)

.119 (.057)

2.711 (1.301) .106 (.044) –.428 (.143) .011 (.134) –.178 (.071) –.240 (.073) –.051 (.072) .141 (.079) .211 (.012) .004 (.007) .004 (.015)

3,762 .128

3,762 .189

3,740 .130

3,740 .176

3,762 .080

3,762 .148

3,052 .053

3,052 .134

—

—

—

—

ber of entering and exiting firms as the dependent variables. The most interesting finding in the table is the role played by the controls for the number of firms and the number of potential entrants at the beginning of the period. These regressions are reported in the even-numbered columns in the table. The model in section 8.2.2 implies that an increase in N should reduce the profit stream from being in the market and so reduce the number of entrants (NE) and increase the number of exits (NX). The model implies an additional eﬀect on exit, holding the profit stream fixed, because an increase in the number of firms making the continuation decision means that the expected number of firms observing a scrap value larger than the value of remaining in the industry (and thus choosing to exit) will rise. Overall, there should be a positive correlation between N and NX and a negative correlation between N and NE. The predicted positive correlation between N and NX observed for both industries and the coeﬃcients, .050 in dentists and .211 in chiropractors, are highly significant. The correlation can reflect either, or both, of the exit linkages in the model and it is not possible using these regressions to identify the separate contribution of each

Market Structure and Size in Two Health Services Industries

323

mechanism. The predicted negative relationship between N and NE is not observed in the dentist industry and is observed, but is not statistically significant, for chiropractors. Both coeﬃcients are small when compared with the impact of an increase in the number of potential entrants, and this may reflect the second-order impact of an increase in N aﬀecting the number of entrants through its eﬀect on the value of entering. The dynamic model predicts that an increase in the number of potential entrants NPE lowers the discounted value of expected future profits by increasing the expected number of firms operating in the future. This will lead to less entry and more exit. It has a second impact on the number of entrants. An increase in the number of potential entrants, holding the profit stream fixed, will increase the expected number of firms that draw an entry cost less than the value of entering and thus choose to enter. So the correlation between NPE and NX should be positive, while the correlation between NPE and NE is ambiguous. The positive relationship between NPE and NX is observed, the coeﬃcient is .004 for both industries but is only statistically significant for dentists. The estimated relationship between NPE and NE is positive and statistically significant, .043 in dentists and .115 in chiropractors. This suggests that the direct eﬀect of an increase in the size of the pool of potential entrants is more important for the entry flow than the indirect or secondary eﬀect this has on the future profit stream. Overall, the importance of N and NPE as control variables in these regressions provides some support for the dynamic framework. The second set of results that is of interest in table 8.2 concerns the coeﬃcients on the demand and cost variables and how these are aﬀected when lagged market structure variables are included. In models that do not control for N and NPE, reported in the odd-numbered columns of table 8.2, the coeﬃcients on the demand and cost variables are all positive, indicating that markets with larger population, per capita income, and wage rates (or price levels) have more producer turnover, that is, both higher entry and exit. Unlike the regressions in table 8.1, there is no competition interpretation linked to these results. When N and NPE are included in the regressions the magnitude of each of these coeﬃcients is substantially reduced. For example, comparing the first two columns, the coeﬃcient on the market size variable falls from .839 to .338 when the two variables are included. This is true in all the entry and exit models and suggests that any conclusions we draw based on the relationship between the demand and cost variables and market structure may be sensitive to whether we base the empirical model on the two-period, long-run equilibrium model or an explicitly dynamic one. To explore this last point further we reestimate the market structure models reported in table 8.1, but now include the lagged number of firms and potential entrants as additional control variables. These findings are reported in table 8.3. First, the lagged number of firms and potential en-

324

Timothy Dunne, Shawn D. Klimek, Mark J. Roberts, and Daniel Yi Xu

Table 8.3

Results for market structure regressions with history variables Dentists Number of firmsa

Constant

Chiropractors

Log average Log labor revenue productivity

Number of firmsa

Log average Log labor revenue productivity

Potential entrants NPE (t – 1) Alpha

–.928 (.365) .316 (.017) .119 (.039) .305 (.039) –.089 (.017) –.116 (.018) –.153 (.021) .042 (.002) .008 (.001) .000

1.313 (.389) .074 (.018) .286 (.042) .298 (.038) .103 (.018) .251 (.019) .366 (.022) –.014 (.020) .057 (.013) —

1.871 (.249) .033 (.012) .115 (.027) .157 (.025) .035 (.011) .092 (.012) .124 (.014) –.009 (.012) .034 (.008) —

–4.049 (.719) .296 (.030) .378 (.079) –.046 (.067) .173 (.033) .270 (.035) .266 (.038) .155 (.013) .032 (.005) .012 (.009)

.807 (.786) .092 (.026) .278 (.086) .317 (.070) .188 (.040) .239 (.041) .119 (.044) .063 (.023) .021 (.019) —

1.482 (.578) .065 (.020) .119 (.063) .212 (.054) .152 (.030) .173 (.031) .109 (.034) –.003 (.018) .033 (.015) —

N R-square

3,010 .282

2,945 .326

2,945 .183

3,010 .222

2,147 .104

2,144 .072

Log market population Log per capita income Log market wage yr = 1987 yr = 1992 yr = 1997 N (t – 1)

a

Use negative binomial model for the number of firms. Remaining regressions are OLS.

trants are both significant variables in the regressions for the number of firms (columns 1 and 4). The importance of these variables is one way to discriminate between the dynamic models that allow for the possibility of hysteresis in market structure, and the two-period models that do not. In addition, for both industries, the other estimated coeﬃcients in these two columns are sensitive to the inclusion of these market history variables. The magnitude of the coeﬃcients on population, per capital income, and the wage rate all drop markedly in comparison to those reported in table 8.1. Of particular interest is the eﬀect on the coeﬃcients of the market size variable, since these have been the focus of most attention in the empirical literature. In the dentist industry, the coeﬃcient on market population drops from .830 in table 8.1 to .316 in table 8.3. The corresponding coeﬃcients for chiropractors are .671 and .296. While all the coeﬃcients are significantly diﬀerent than one and would, if the two-period model was taken literally, imply a competitive eﬀect of entry, the magnitude of the eﬀect is clearly very diﬀerent between model specifications. This raises questions

Market Structure and Size in Two Health Services Industries

325

about how to interpret the estimated relationship between the size of a market and the number of firms. In contrast to the model of the number of firms, there is little systematic relationship between the lagged market structure variables and either the average revenue or labor productivity variables. The most significant correlations appear between the number of potential entrants and the average revenue and productivity variables for dentists. Still, the results in table 8.3 suggest that the lagged market structure variables are determinants of current market structure, particularly the number of firms in the market, as implied by the dynamic model of entry and exit. This only serves to further complicate the interpretation of the regression coeﬃcients in this type of model, since they now reflect much more than a possible competitive eﬀect of entry. 8.6 Conclusion This chapter utilizes U. S. Census micro data to study patterns of producer dynamics in two health service industries, dentists and chiropractors. The analysis is guided by studies in the industrial organization literature that quantify the relationship between market size and market structure, where the latter is measured in several ways, including the number of firms, the average size of firms, and the level of productivity. The framework is extended to incorporate the analysis of entry and exit flows. Recent models of producer dynamics stress the diﬀerent decisions faced by incumbent firms and potential entrants. In particular, the existence of a sunk entry cost implies that the decision of an incumbent producer to remain in a market diﬀers in fundamental ways from the decision of a new firm to enter the market. One implication of these recent models is that entry flows, exit flows, and current market structure depend not only on current demand and cost conditions but also on the history of participation decisions. In order to empirically examine the determinants of entry, exit, and market structure, we use census data for 754 small, geographically isolated markets for dentists and chiropractor services and follow these markets over a twenty-five-year time span. We find a significant role for both the past number of firms and the number of potential entrants as determinants of current market structure, and this is consistent with the dynamic model of entry and exit we rely on. Our empirical findings also show that as market size increases, the number of firms rises less than proportionally, firm average size increases, and average productivity in the market increases. All of these patterns replicate findings of other studies that have been used to infer that markets become more competitive as they increase in size. However, we find the magnitudes of these correlations, particularly for the number of firms, are sensitive to the inclusion of the market history vari-

326

Timothy Dunne, Shawn D. Klimek, Mark J. Roberts, and Daniel Yi Xu

ables, and this suggests caution in interpreting the cross-market regressions as revealing much about the competitive structure of the market. The relationship between the size of a market, the magnitude of barriers to entry including the size of entry costs, and the competitiveness of a market is an issue of long-standing interest in industrial organization. Changes in market size aﬀect firm profitability but also generate flows of entry and exit that also impact profitability. These entry and exit flows are determined, in part, by the magnitude of sunk entry costs, which are very hard to measure and control for in empirical work. The correlations between market size and the number of firms, entry flows, and exit flows will reflect the interaction of these entry barriers, magnitude of competition in the market, and expected future changes in market conditions and profitability. In Dunne et al. (2008) we utilize U. S. Census micro data to estimate an empirical model of entry, exit, and profitability to identify these separate components of market structure and performance. One key to future empirical work in the area of producer dynamics is micro data that can track the evolution of firms and markets over time. Producer data sets that cover a broader range of sectors, countries, and time periods are a crucial component of future research on the sources and impacts of producer dynamics.

References Aguirregabiria, V., and P. Mira. 2007. Sequential estimation of dynamic discrete games. Econometrica 75 (1): 1–53. Asplund, M., and V. Nocke. 2006. Firm turnover in imperfectly competitive markets. The Review of Economic Studies 73 (2): 295–327. Asplund, M., and R. Sandin. 1999. The number of firms and production capacity in relation to market size. Journal of Industrial Economics 47 (1): 69–85. Berry, S. 1992. Estimation of a model of entry in the airline industry. Econometrica 60 (4): 889–917. Berry, S., and P. Reiss. 2006. Empirical models of entry and market structure. In Handbook of industrial organization vol. III, ed. M. Armstrong and R. Porter, chapter 29. Amsterdam: North-Holland Press. Berry, S., and E. Tamer. 2006. Identification in models of oligopoly entry. In advances in economics and econometrics vol. II, ed. R. Blundell, W. K. Newey, and T. Persson, 46–85. Cambridge: Cambridge University Press. Berry, S., and J. Waldfogel. 2003. Product quality and market size. NBER Working Paper no. 9675. Cambridge, MA: National Bureau of Economic Research, April. Bresnahan, T., and P. Reiss. 1989. Do entry conditions vary across markets? Brookings Papers on Economic Activity, Microeconomics Annual 1:833–82. ———. 1991. Entry and competition in concentrated markets. Journal of Political Economy 99 (5): 977–1009. ———. 1994. Measuring the importance of sunk costs. Annales D’Economie et de Statistique 34:181–217.

Market Structure and Size in Two Health Services Industries

327

Campbell, J., and H. Hopenhayn. 2005. Market size matters. Journal of Industrial Economics 53 (1): 1–25. Collard-Wexler, A. 2006. Plant turnover and demand fluctuation in the ready-mix concrete industry. Center for Economic Studies Working Paper 06-08. U. S. Census Bureau. Das, S., M. J. Roberts, and J. Tybout. 2007. Market entry costs, producer heterogeneity, and export dynamics. Econometrica 75 (3): 837–73. Dixit, A., and R. Pindyck. 1994. Investment under uncertainty. Princeton, NJ: Princeton University Press. Dunne, T., S. D. Klimek, M. J. Roberts, and Y. Xu. 2008. Entry and exit in geographic markets. Pennsylvania State University. Working Paper. Hopenhayn, H. 1992. Entry, exit, and firm dynamics in long run equilibrium. Econometrica 60 (5): 1127–50. Jarmin, R., and J. Miranda. 2002. The longitudinal business database. Center for Economic Studies Working Paper 02-14. U. S. Census Bureau. Mazzeo, M. J. 2002. Product choice and oligopoly market structure. The RAND Journal of Economics 33 (2): 221–42. Pakes, A., M. Ostrovsky, and S. Berry. 2007. Simple estimators for parameters of discrete dynamic games with entry/exit examples. The RAND Journal of Economics, 38 (2): 373–99. Pesendorfer, M., and P. Schmidt-Dengler. 2003. Identification and estimation of dynamic games. NBER Working Paper no. 9726. Cambridge, MA: National Bureau of Economic Research, April. Ryan, S. 2006. The costs of environmental regulation in a regulated industry. MIT Working Paper. Seim, K. 2006. An empirical model of firm entry with endogenous product-type choices. The RAND Journal of Economics 37 (3): 619–42. Sutton, J. 1991. Sunk costs and market structure. Cambridge, MA: MIT Press. Syverson, C. 2004. Market structure and productivity: A concrete example. The Journal of Political Economy 112 (6): 1181–1222.

9 Measuring the Dynamics of Young and Small Businesses Integrating the Employer and Nonemployer Universes Steven J. Davis, John Haltiwanger, Ronald S. Jarmin, C. J. Krizan, Javier Miranda, Alfred Nucci, and Kristin Sandusky

9.1 Introduction The measurement of economic activity by federal statistical agencies focuses greater attention on larger, more mature business units. This datagathering strategy has two clear advantages. First, it yields greater accuracy in estimating the level of economic activity, whether greater attention takes the form of higher sampling probabilities or more careful auditing and editing. Second, it is easier to identify and promptly capture the activity of large, long-established business units. On both counts, the desire for a cost-eﬀective approach to measuring the level of economic activity leads naturally to a focus on larger, more mature units. Steven J. Davis is the William H. Abbott Professor of International Business and Economics at the Graduate School of Business, University of Chicago, and a research associate of the National Bureau of Economic Research. John Haltiwanger is a professor of economics at the University of Maryland, and a research associate of the National Bureau of Economic Research. Ronald S. Jarmin is director of research at the Center for Economic Studies, U.S. Census Bureau. C. J. Krizan is an economist at the Center for Economic Studies, U.S. Census Bureau. Javier Miranda is an economist at the Center for Economic Studies, U.S. Census Bureau. Alfred Nucci is an economist at the Center for Economic Studies, U.S. Census Bureau. Kristin Sandusky is an economist in the Longitudinal Employer-Household Dynamics (LEHD) program at the U.S. Census Bureau. We gratefully acknowledge support from the Kauﬀman Foundation, the U.S. Census Bureau, and the Initiative on Global Financial Markets at the University of Chicago. For helpful comments on earlier drafts, we thank Mark Roberts, Tim Dunne, Thomas Holmes, Robert Strom, Robert Litan, E. J. Reedy, Ying Lowery, Richard Boden, Lucia Foster, and participants at the NBER/CRIW conference on Producer Dynamics in April 2005 and at an AEA/SBA session in January 2005. We also would like to thank Paul Hanczaryk for helping us understand the Census Bureau’s nonemployer data. This work has undergone a more limited review than oﬃcial Census Bureau publications. The views, findings, and opinions expressed in this work are those of the authors and not the U.S. Census Bureau. All results have been reviewed to ensure confidentiality.

329

330

S. Davis et al.

There are, however, drawbacks to this data-gathering strategy. When responses to shocks and new developments in the economy vary systematically with business size or age, a focus on larger and more mature units can yield less accurate, potentially misleading measures of changes in economic activity. As a simple example, consider the situation when younger and smaller business units are relatively sensitive to aggregate shocks. In this case, a cost-eﬀective approach to estimating short-term growth rates can require the oversampling of younger and smaller business units, and there is tension between a sample design optimized for the level of activity and one optimized for the growth rate. More important, the traditional focus on larger and more mature units limits our ability to measure and study the early life cycle dynamics of businesses and to evaluate theories of business formation, selection, and growth. This chapter reports our initial eﬀorts to remedy these drawbacks. We develop a preliminary version of an Integrated Longitudinal Business Database (ILBD) that combines administrative records and survey-based data for all nonfarm employer and nonemployer business units in the United States. In the process, we confront conceptual and practical issues that arise in measuring the importance and dynamic behavior of younger and smaller businesses. We also document some basic facts about younger and smaller businesses. In doing so, we exploit the ability of the ILBD to follow business transitions between employer and nonemployer status. This aspect of the ILBD opens a new frontier for the study of business formation and the precursors to job creation in the U.S. economy. As of 2000, there are 5.4 million nonfarm business firms with employees in the United States and another 15.5 million with no employees. Most nonemployer business units are quite small, never become employers, and do not link to the employer universe by way of any ownership relation. Nonemployer businesses account for a modest 4 percent of aggregate U.S. business revenue within the year, but we show that a substantial share of employers originate as nonemployer businesses. Our analysis focuses on forty industries for which smaller and younger businesses play especially important roles. These industries account for nearly half of nonemployer business units in the U.S. economy and 36 percent of nonemployer revenues. Within these industries, nonemployers account for 14 percent of business revenues. In addition, more than 11 percent of employers in these industries are connected by some type of ownership link to the nonemployer business universe within the previous eight years. Many of these linkages reflect nonemployer businesses that expand over time, hire workers, and become employers, but other linkage patterns also arise. For example, some employer and nonemployer business units operate simultaneously under common ownership for many years. In the analysis below, we identify and quantity several types of ownership linkages between the employer and nonemployer business universes.

Measuring the Dynamics of Young and Small Businesses

331

Over a three-year horizon, 3 percent of the roughly seven million nonemployers in our selected industries become Migrants to the employer universe. By Migrants, we mean that the firm starts as a nonemployer business and later hires one or more paid employees. While 3 percent is a small share of nonemployers, it amounts to 220,000 transitions. Indeed, Migrants play a nontrivial role in the formation of employer businesses. They account for 28 percent of firms and 20 percent of revenues among young employers (three years or less since first hire) in our selected industries. Migrants also grow very rapidly around the transition event. Mean annual revenue growth for Migrants is 31 percent in the year prior to transition and 101 percent in the year of transition, much higher than contemporaneous growth for other nonemployers. As another step toward an integrated perspective on the dynamics of young and small businesses, we compare the revenue growth patterns of employers and nonemployers. Conditional on survival, net revenue growth rates decline strongly with business age and size for employers and nonemployers alike. When we include business exits, however, revenue growth shows much weaker and less clear-cut relationships to size and age. The dispersion in growth rates is also much higher for younger and smaller businesses, mainly because of much higher business turnover rates. The chapter proceeds as follows. Section 9.2 describes the construction of the ILBD. Section 9.3 presents some facts about business numbers, activity levels, and the business age and size distributions. Section 9.4 investigates ownership linkages over time between employer and nonemployer businesses. We quantify transitions between employer and nonemployer status and other linkages between employer and nonemployer businesses. Section 9.5 reports revenue growth and dispersion patterns by business size and age. Section 9.6 discusses next steps in our research program, and section 9.7 oﬀers concluding remarks. 9.2 Constructing an Integrated Longitudinal Business Database 9.2.1 Overview of Main Tasks and Previous Work In terms of data development, our objective is to build a fully Integrated Longitudinal Business Database (ILBD) that covers all employer and nonemployer business units in the nonfarm private sector of the U.S. economy. We construct the initial version of the ILBD for the years 1992 and 1994 to 2000, and we plan updates for later years in future ILBD enhancements. Key data on nonemployers are unavailable for 1993. From an analytical perspective, the presence or absence of employees is simply another business characteristic to be measured. From a database development perspective, however, integrating the Census Bureau’s employer and nonemployer business universes is a major undertaking. The main

332

S. Davis et al.

tasks fall into three categories. One set of tasks involve the construction of longitudinal links for business units within each universe. A second task is to integrate the employer and nonemployer universes on a year-by-year basis, ensuring that each unique business entity is counted once, and only once. A third task is to construct contemporaneous and dynamic ownership links across universes between employer and nonemployer business units. To carry out these tasks, we build on previous work by Jarmin and Miranda (2003) to create the Longitudinal Business Database (LBD), which contains annual data from 1975 to 2001 for all nonfarm private employers. We also build on previous eﬀorts to construct longitudinal links within the nonemployer universe by Nucci and Boden (2003) and Boden and Nucci (2004). 9.2.2 Source Data for the ILBD Census Bureau business registers draw on payroll tax records, corporate and individual income tax returns, applications for an Employer Identification Number (EIN), and various Census Bureau business surveys. The data available to the Census Bureau vary with the legal and tax status of a business and, in certain respects, its size and number of locations. For large corporations, routine data inputs include payroll records and particular items from corporate income tax returns, augmented by direct Census Bureau collections for multi-location companies. For sole proprietors, partnerships, and single-location corporations with employees, routine data inputs include payroll records, certain items from income tax returns, and periodic Census Bureau surveys such as the quinquennial Economic Census. For nonemployer businesses, routine data inputs derive mainly from income tax returns. Tables 9.1, 9.2, and 9.3 list the most important administrative and survey sources for key variables in the employer and nonemployer universes. To construct the ILBD, we must first ensure that administrative data from each universe are cleaned and ready for integration. On the employer side, this task has been largely accomplished in the work to create the Longitudinal Business Database (LBD).1 The Census Bureau’s Employer Business Register, which underlies the LBD, is a list of establishments (physical locations) maintained to serve as a mailing list for the Economic Census and as a sample frame for surveys. The Employer Business Register relies heavily on administrative data and is supplemented by direct Census Bureau collections.2 Longitudinal linking is facilitated by establishment 1. See Jarmin and Miranda (2003). The main outstanding issue with respect to the LBD concerns the delayed identification of new establishments owned by certain multi-unit companies. We are developing algorithms to retime these births. The retiming issue pertains only to the recognition date of establishment birth, not the company-wide level of revenues or other measures of economic activity. 2. In order to track the establishment structure of multi-unit enterprises, the Census Bureau conducts an annual Company Organization Survey. This survey covers all large multi-unit companies and a sample of smaller ones. During an economic census, all establishments of multi-unit companies receive survey forms.

Employment

Industry

Name and address

Variable

Table 9.1

Surveya Administrative

Administrative

Survey

Administrative

Survey

Source type Source details

Company Organization Survey, Annual Surveys, or Economic Censuses in Census years. IRS Form 941 IRS FICA wages IRS total compensation Imputed

Reported industry code from Company Organization Survey, Annual Surveys, or Economic Censuses in Census years. Derived from the 1992 Economic Census—respondent reported classification Derived from a current survey (County Business Patterns (CBP), Company Organization Survey (COS)/Annual Survey of Manufactures (ASM), Current Industrial Report (CIR), Business Sample Revision (BSR) CBP Safeguard Review or intercensal refiles Derived from the Bureau of Labor Statistics Derived from the Social Security Administration Derived from the Internal Revenue Service Principal Business Activity code, obtained from the Business Master File.

Physical address from Company Organization Survey, Annual Surveys, or Economic Censuses in Census years. Beginning in 1998, physical address from form ss-4 for births; before 1998, mailing address from form ss-4. Beginning in 1998, physical address from form 941 for all businesses; before 1998, mailing address from form 941 for all businesses. Beginning in 1998, physical address from IRS income tax form for all businesses; before 1998, mailing address from IRS income tax form for all businesses.

Employer business register

(continued )

Line 1

Top of form

Top of form

4a,4b; 5a,5b

Line number

Company organization survey, Annual surveys, or Economic Censuses in Census years 1120 – Gross receipts or sales less returns and allowances 1120-A – Gross receipts or sales less returns and allowances 1220 F – Gross receipts or sales less returns and allowances 1120L – Gross income 1120-PC – Gross income 1120-RIC – Total income 1120S – Gross receipts or sales less returns and allowances 1065 – Gross receipts or sales less returns and allowances 990 – Total revenue 990-C – Gross receipts or sales less returns and allowances 990EZ – Total revenue 990-PF – Total revenue 1040C – Gross receipts or sales less returns and allowances

Surveya Imputed from EIN-level data

Revenue

Line 1c Line 1c Section II, line 1a Line 9 Sch A: line 14 Line 8 Line 1c Line 1c Line 12 Line 1c Line 9 Line 12 Line 3

Line 2

Line number

Data for multi-unit firms disaggregated to the establishment level using survey information, except where the firm reports individual establishments under separate EINs.

a

Company Organization Survey, Annual Surveys, or Economic Censuses in Census years. IRS Form 941 IRS FICA wages IRS total compensation Imputed

Source details

Survey Administrative

a

Source type

(continued)

Payroll

Variable

Table 9.1

Measuring the Dynamics of Young and Small Businesses Table 9.2

335

Nonemployer business register, proprietors—SSN Records

Variable Name Mailing address Legal form of organization Industry code Revenue

Source from IRS Form 1040 including Schedule C Form 1040 Form 1040 Implied by filing of IRS form 1040 Schedule C Line B Line 3: Gross receipts or sales less returns and allowances

Notes: All data for sole proprietors (including proprietorships jointly operated by husband and wife) are obtained form IRS form 1040 including Schedule C. Line B of Schedule C reports the “Principal business or profession, including product or service.” Based on this information, the IRS codes the Principal Business Activity (PBA) of the proprietorship. Based on the PBA code, the Census assigns a Tabulated Kind of Business (TKB) code. If the PBA is not reported on the tax form, then the Census uses the historic TKB code, if available.

IDs (LBD Numbers and Permanent Plant Numbers [PPNs]), EINs, enterprise IDs (Alphas), and business name and address information. Longitudinal establishment links are relatively straightforward to construct because they are one to one, and because establishments typically have well-defined physical locations. Firms are more diﬃcult to track over time, partly because firm-level links can be many to many. It should be noted that the Census Bureau continues to refine its longitudinal firm linkages. Longitudinal links are diﬃcult to construct in the nonemployer universe for some of the same reasons that they are diﬃcult to construct for firms in the employer universe. Some nonemployer businesses have an EIN, but most do not. Instead, they are tracked by the person ID of the business owner, that is, his or her Social Security Number (SSN).3 When there is a change in the legal or tax status of a nonemployer business, its EIN or person ID can also change. (Person IDs do not change for individuals, but ownership changes can yield a change in the person ID associated with a business.) In these instances, it is not straightforward to maintain longitudinal links for nonemployer businesses using data items that are routinely included in the Census Bureau’s administrative records systems. Direct Census Bureau collections provide this additional information on the employer side, but there is no ready equivalent on the nonemployer side. Our longitudinal links for nonemployer businesses currently exploit EINs, person IDs in the form of SSNs, business name information, and geographic information. Our main source of revenue data for nonemployer businesses are Schedule C forms attached to personal income tax filings. Two complications arise in this regard. First, multiple Schedule C forms can be attached to a single 1040 tax form. In these cases, we aggregate to the level of a single 3. A nonemployer business has an Employer Identification Number if it previously had paid employees or applied for an EIN in anticipation of hiring paid employees.

U.S. Corporation Income Tax Return

U.S. Corporation Short Form Tax Return

U.S. Income Tax Return for an S Corporation U.S. Income Tax Return of a Foreign Corporation U.S. Casualty and Property Insurance Company income Tax Return U.S. Life Insurance Company Income Tax Return U.S. Income Tax Return for Regulated Investment Companies U.S. Income Tax Return for Real Estate Investment Trusts

Form 1120

Form 1120A

Form 1120 S

Inferred by Form Type

Schedule M, 2a-c (Kind of Company, Principal Business) Inferred by Form Type

Schedule I, Line 2

Line F1 (Business Activity Code)

Line B (Business Code)

Schedule K, Line 2a (Business Activity Code) Part 2, line 1a (Business Activity Code)

Line A (principal business activity)

Industry code source

8: Total Income

8: Total Income

9 Gross Income

Line 1c: Gross receipts or sales less returns and allowances Line 1c: Gross receipts or sales less returns and allowances Line 1c: Gross receipts or sales less returns and allowances Line 1c: Gross receipts or sales less returns and allowances Section II, 1c: Gross receipts or sales less returns and allowances Schedule A: line 14 (gross income)

Revenue source

Notes: All data for these businesses are derived from IRS income tax returns filed by the businesses. Name and Mailing Address taken from top of forms. Industry codes on Census Nonemployer database are IRS PBA codes obtained from the sources noted above and then converted to Census TKB codes. If the PBA is not available from the tax form, then the Census uses the historic TKB code, if available. Legal form of organization implied by type of form submitted: 1065 filers are partnerships, all others are corporations.

Form 1120 REIT

Form 1120 RIC

Form 1120 L

Form 1120 PC

Form 1120 F

U.S. Partnership return of income

Form description

Nonemployer business register, partnerships and corporations—EIN records

Form 1065

Form number

Table 9.3

Measuring the Dynamics of Young and Small Businesses

337

1040 tax filing and associate the revenue to the SSN of the primary tax return filer. Second, when married couples with Schedule C income file jointly, either the husband or the wife can be the primary filer. In these cases, we associate the Schedule C income to the SSN of the primary filer, but we retain the spouse’s SSN as well. We then create longitudinal links based on the SSN of the primary filer. In addition, for married couples who file jointly in year t and t k, we create additional links that capture a primary filer in year t who matches to his or her spouse’s SSN in year t k (for k positive or negative). These additional cases account for less than one percent of all longitudinal linkages among nonemployers. 9.2.3 Cross-Universe Matching In addition to the longitudinal links within each universe, we create a set of firm-level matches between employers and nonemployers for our selected industries. These matches rely on numeric identifiers and exact literal matches on business names. In matching on numeric identifiers, we exploit the fact that many business records contain both an EIN and an SSN. For example, when a business owner or oﬃcer applies for an EIN, he or she must fill out an SS-4 form for the IRS. This form includes the business name, the EIN, and the SSN of the business owner or chief oﬃcer, all of which are included in Census Bureau business registers. These data allow us to build a crosswalk between EINs and SSNs, which we then use to match business records across universes. We take a conservative approach in matching records between employers and nonemployers. In particular, we rely only on the EIN-SSN crosswalk and exact literal matches on business name. As an example of how our matching algorithm works, consider all establishments with employees in our selected industries as of 2000. Using the longitudinal links in the LBD, we first create a set of identifiers (EINs, SSNs, and business names) associated with each establishment with employees in 2000 for each year back to 1992. Given the list of identifiers for a particular establishment, we then identify its matches to nonemployers in the years from 1992 to 2000. Since we are most confident about cross-universe matches based on numeric identifiers, we match on EIN, SSN, and business name—in that order. For matches on business name, we also require that the employer and nonemployer be in the same state. The industry restriction applies to the business under consideration in the origin universe (i.e., employer universe) and not to its potential matches in the other business universe. We create matches in the other direction using a similar method. That is, for each nonemployer, we match its numeric and name identifiers to the identifiers of establishments in the employer universe. About 17 percent of our employer-nonemployer matches rely on exact literal matches on business name strings. We experimented with alternative name-matching procedures, including the removal of vowels, various sym-

338

S. Davis et al.

bols, and commonly used abbreviations (inc., ltd., etc.). However, after examining a sample of matches, we concluded that literal name strings produced highly reliable matches and appeared to miss very few good matches. Nevertheless, previous eﬀorts to develop longitudinal identifiers for the LBD have shown that linkages can be improved by the use of more sophisticated probabilistic matching algorithms that exploit all relevant available information, and that take into account the reliability of the information. We plan to incorporate some of these techniques in future enhancements of the ILBD. Given our current algorithm, we probably understate the incidence of linkages between the two universes. For the same reason, we probably understate the contribution of nonemployer businesses to the formation of employer businesses. Given a set of matches between the two universes, we aggregate the establishment data within an industry to the firm level. The result is a firmlevel data set with ownership linkages to nonemployer businesses and additional variables that describe the nature of the nonemployer records to which the employer firm links. At this point, the unit of observation is a business firm with at least one establishment operating in one of our selected industries. If a firm operates in more than one of our selected industries, we maintain separate records for each industry in which it operates. 9.3 Basic Facts about Employers and Nonemployers 9.3.1 Business Numbers and Activity Levels Table 9.4 provides summary statistics for the employer and nonemployer business universes in 2000. There are about 15.5 million nonemployer businesses. Of these, 13.4 million are person ID units (sole proprietorships with no employees) and 2.1 million are EIN units (corporations, partnerships, and other nonemployer business entities with EINs).4 There are also about 5.4 million employer businesses. Of these, 182 thousand are multi-unit enterprises with more than one establishment, and the rest are single-unit businesses. While comparatively small in number, multi-unit enterprises account for 61 percent of aggregate U.S. business revenue. Nonemployer business units account for 4 percent of aggregate revenue, and single-unit employers account for 35 percent. Given the sheer size of the Census Bureau business registers and some complex issues of measurement, we focus on a selected set of forty industries for this chapter. We choose industries with large numbers and relatively 4. The distinction between person ID and EIN units can be complex. A sole proprietor with no payroll but positive receipts who has applied for an EIN can appear in both the person ID and EIN sections of the Nonemployer Business Register. That same proprietor can appear in the Employer Business Register as well. We assign all zero-payroll units to the nonemployer universe, even if they reside in the Employer Business Register.

13.38 2.15 15.54 459.53 251.74 711.26

1.43

EIN units

6.84 0.54 7.38 199.87 55.58 255.45

Number or revenue

2.61

SSN units 34.77

Single-unit

Single-unit Multi-unit All Single-unit Multi-unit All

61.19

Multi-unit

3.09

EIN units

1.9 0.06 1.96 877.92 664.52 1,542.44

Number or revenue

11.12

SSN units

36 31 36 14 6 9

As percent of all

48.83

Single unit

36.96

Multi-unit

Employer businesses

40 selected industries

5.26 0.18 5.44 6,113.43 10,758.04 16,871.47

All industries

40 selected industries

Employer business universe

Nonemployer businesses

Integrated business universe

51 25 48 43 22 36

As percent of all

Employer businesses

All industries

Nonemployer businesses

SSN units EIN units All SSN Units EIN units All

All industries

40 selected industries

Nonemployer business universe

Summary statistics for the employer and nonemployer business universes, 2000

Percent of aggregate revenue

Revenue (billions $)

# of units (millions)

Table 9.4

340

S. Davis et al.

high revenue shares for young and small businesses. Dynamic links between employers and nonemployers are likely to be more important for these industries. We avoid industries with complex measurement issues related to financial holding companies, tax shelters, and special purpose financial entities. These aspects of corporate organizational structure are interesting, but they are not the focus of our eﬀorts. Our analysis period overlaps with the transition from SIC to NAICS industry classifications, and the SIC-NAICS crosswalk is a many-to-many mapping. Our nonemployer data files contain three-digit SIC codes prior to 1997 and four-digit NAICS codes thereafter. The employer data files contain codes for both classifications from 1997 to 2000. Accordingly, we proceed as follows. For many exercises, we look backwards for businesses in selected four digit NAICS industries. For other exercises, we look forward from a year prior to the NAICS changeover at businesses in threedigit SIC codes that correspond closely to our selected NAICS industries. Table 9.5 provides summary information for our selected four-digit NAICS industries. Legal Services has the largest number of employer businesses, almost 150 thousand. It also has the biggest employment and labor costs, with more than one million workers and more than 58 billion dollars in payroll. The highest-revenue industry for employers is Gasoline Stations, at 187 billion dollars.5 Other Personal Services has the largest number of nonemployer businesses, more than 800 thousand. The highestrevenue industry for nonemployers is Real Estate Agents and Brokers, at almost 23 billion dollars. Table 9.6 provides information about industry shares of aggregate business revenues and the relative size of the employer and nonemployer segments within industries. Nonemployer revenue shares range widely. At the upper end, nonemployers account for more than two-thirds of revenue in Independent Artists, Writers, and Performers, and they account for at least 30 percent of revenues in a dozen industries. At the lower end, nonemployers account for a mere 24 hundredths of a percent of revenue in Software Publishers, even though 20 percent of all businesses in this industry are nonemployers. Although extreme, the basic pattern in this industry is not uncommon; it reflects the enormously skewed size distribution of activity in many industries. 9.3.2 Business Age and Size Distributions Figures 9.1 to 9.4 provide information about the age and size distributions of businesses in our selected industries. Age is measured in years 5. High revenue in Gasoline Stations mainly reflects the cost of gasoline. The administrative data in the Census business registers typically does not include information on gross margins or material costs. Such data are included in the economic censuses and various annual surveys.

Firms 1.5 36.4 52.3 27.6 24.5 23.2 1.1 52.2 9.4 20.7 5.1 2.3 6.7 101.7 50.0 39.5 13.7 147.7 76.9 74.9 76.5 21.3 115.2 120.5

Animal production support activities Painting and wall covering contractors Carpentry and floor contractors Roofing, siding, and sheet metal contractors Concrete contractors Printing and related support activities Ship and boat building Gasoline stations Book, periodical, and music stores Florists Taxi and limousine service Couriers Software publishers Agencies and other insurance-related activities Offices of real estate agents and brokers Activities related to real estate Consumer goods rental Legal services Accounting, tax prep, bookkeep, payroll service Computer systems design and related services Management, scientific, and tech consulting services Travel arrangement and reservation services Services to buildings and dwellings Offices of physicians 156 6,274 9,637 8,143 5,092 14,226 3,788 12,282 2,713 1,645 1,206 15,654 23,009 30,448 11,504 15,052 3,483 58,514 21,273 76,674 32,718 8,713 24,903 46,346

Employers payroll 7 226 383 264 178 420 105 837 222 123 65 550 249 745 297 484 236 1,055 765 1,194 729 292 1,407 1,043

Employment

Summary statistics for selected industries in the integrated business universe, 2000

Industry description

Table 9.5

664 17,469 35,398 27,315 18,185 41,613 16,896 187,841 12,577 6,417 3,451 18,610 49,988 90,461 46,826 38,058 11,391 149,400 42,205 141,900 67,277 45,546 59,490 102,651

Revenue 37.7 213.5 389.6 86.5 42.6 26.7 0.4 9.4 28.9 22.7 117.6 1.1 1.7 308.2 476.6 356.3 16.0 206.3 294.5 249.4 355.2 31.8 538.9 149.6

Nonemployers firms

1,462 7,443 16,722 5,047 2,567 1,486 47 1,682 1,008 869 3,419 111 120 14,849 22,952 18,274 768 11,626 6,059 9,688 16,796 1,776 11,294 11,664 (continued )

Revenue

(continued)

5,443.40

35.0 74.7 21.1 39.2 1.8 8.9 1.5 91.9 94.4 11.2 44.3 94.8 22.4 59.8 30.8 16.0

Firms

3,773,003

7,268 8,814 9,411 9,747 502 2,272 182 20,908 17,164 8,314 3,753 12,533 2,337 4,109 6,613 3,688

Employers payroll

Note: Firms and employment in thousands. Payroll and Revenue in millions.

Total Economy

Offices of dentists Offices of other health practitioners Individual and family services Child day care services Agents, managers for artists, and public figures Independent artists, writers, and performers Rooming and boarding houses Full-service restaurants Limited-service eating places Special food services Drinking places (alcoholic beverages) Automotive repair and maintenance Personal and household goods R and M Personal care services Drycleaning and laundry services Other personal services

Industry description

Table 9.5

113,658

273 346 478 691 10 38 13 1,863 1,743 563 360 583 98 303 373 173

Employment

16,871,471

19,865 31,356 11,919 14,125 1,549 4,461 717 47,763 53,707 17,466 12,468 46,945 8,685 9,770 17,731 12,288

Revenue

15,536.07

29.4 235.2 65.3 516.6 25.0 465.1 9.7 29.0 36.8 68.5 21.3 251.2 247.9 552.4 33.0 835.8

Nonemployers firms

711,264

1,907 9,053 1,096 6,263 867 9,631 281 3,308 3,086 2,014 1,466 11,570 6,147 11,776 1,650 17,609

Revenue

1152 2352 2355 2356 2357 3231 3366 4471 4512 4531 4853 4921 5112 5242 5312 5313 5322 5411 5412 5415 5416 5615 5617

NAICS code

Table 9.6 Industry revenues, percent of aggregate business revenue 0.01 0.14 0.30 0.18 0.12 0.25 0.10 1.08 0.08 0.04 0.04 0.11 0.28 0.60 0.40 0.32 0.07 0.92 0.27 0.86 0.48 0.27 0.40

Industry description

Animal production support activities Painting and wall covering contractors Carpentry and floor contractors Roofing, siding, and sheet metal contractors Concrete contractors Printing and related support activities Ship and boat building Gasoline stations Book, periodical, and music stores Florists Taxi and limousine service Couriers Software publishers Agencies and other insurance-related activities Offices of real estate agents and brokers Activities related to real estate Consumer goods rental Legal services Accounting, tax prep, bookkeeping, payroll service Computer systems design and related services Management, scientific, and tech consulting services Travel arrangement and reservation services Services to buildings and dwellings

Industry revenue shares and business-type shares within industries, 2000

31 70 68 84 88 97 100 99 93 88 50 99 100 86 67 68 94 93 87 94 80 96 84

Industry revenues, percent accounted for by employers

96 85 88 76 63 54 26 15 75 52 96 32 20 75 91 90 54 58 79 77 82 60 82 (continued )

Percent of business entities in industry that are nonemployers

6211 6212 6213 6241 6244 7114 7115 7213 7221 7222 7223 7224 8111 8114 8121 8123 8129

NAICS code

Table 9.6

Industry description

Offices of physicians Offices of dentists Offices of other health practitioners Individual and family services Child day care services Agents, managers for artists, and public figures Independent artists, writers, and performers Rooming and boarding houses Full-service restaurants Limited-service eating places Special food services Drinking places (alcoholic beverages) Automotive repair and maintenance Personal and household goods R and M Personal care services Drycleaning and laundry services Other personal services

(continued)

0.65 0.12 0.23 0.07 0.12 0.01 0.08 0.01 0.29 0.32 0.11 0.08 0.33 0.08 0.12 0.11 0.17

Industry revenues, percent of aggregate business revenue 90 91 78 92 69 64 32 72 94 95 90 89 80 59 45 91 41

Industry revenues, percent accounted for by employers

55 46 76 76 93 93 98 86 24 28 86 32 73 92 90 52 98

Percent of business entities in industry that are nonemployers

Fig. 9.1

Age distribution of business numbers and revenues within each universe

346

S. Davis et al.

since first appearance by a business entity in its respective business universe. For multi-unit firms, business age is defined as the age of the firm’s oldest establishment. We use revenue measures to compare size distributions across the two universes, because revenue is the only activity measure available for both universes. As seen in figure 9.1, older firms dominate economic activity among employers. Firms that are at least eight years old account for almost 70 percent of employer revenues and more than 40 percent of employer businesses in our selected industries. In contrast, older businesses play a much smaller role in the nonemployer universe. Firms that are at least eight years old account for only 40 percent of nonemployer revenues. Very young nonemployers account for a larger share of business units and even revenue than, say, five-year-old nonemployers. As seen in figure 9.2, most nonemployer businesses are quite small. Roughly a third of nonemployer businesses generate less than 6,000 dollars in annual revenue. While large in numbers, these very small nonemployers account for a tiny fraction of revenues. Somewhat larger business units account for much of the revenue generated by nonemployers. For example, nearly one-quarter of nonemployer revenue in our selected industries is generated by businesses with annual revenue in the range of 120 to 360 thousand dollars. In contrast, the size distribution of revenues has a very diﬀerent shape in the employer universe. Almost 70 percent of employer revenue is generated by firms with more than three million in annual revenue. The tremendous variation in size across nonemployer and employer businesses exhibited in figure 9.2 serves as a caution when drawing inferences about the behavior of small and young businesses. The wide size distribution reminds us that many nonemployer businesses are extremely small and often represent a secondary or supplemental source of income to the household. Analyzing the dynamics of such businesses alongside much larger businesses is a challenge. In what follows, we often report results for both the share of business units and the share of revenue. The former provides more insights about the very small and more prevalent businesses, while the latter provides more insights into the contribution of larger businesses. Figures 9.3 and 9.4 display the share of revenues and business units accounted for by young businesses (zero to three years old) and small businesses (less than $90,000 in annual revenue) in our selected industries. Figure 9.3 reveals wide variation across industries in the revenue and number shares of young and small businesses in the nonemployer universe. Figure 9.4 shows a similar pattern with respect to the revenue and numbers share of young businesses in the employer universe and with respect to the numbers share of small businesses. However, with the exception of Personal Care Services (NAICS 8121), employers with less than $90,000 in annual revenue account for very small revenue shares, typically less than five percent.

Fig. 9.2

Size distribution of business numbers and revenues within each universe

348

S. Davis et al.

Fig. 9.3 Share of revenues and business numbers accounted for by young firms, 0 to 3 years old, in selected four-Digit NAICS industries Source: Own Calculations from ILBD.

9.4 Ownership Links and Transition Dynamics 9.4.1 Backward Links of Employers to Nonemployers We turn now to ownership links between the employer and nonemployer business universes. We first consider all employer businesses in our selected industries in the 2000 cross section. Table 9.7 reports size and age distributions for these businesses in columns (2) and (3). These columns show familiar patterns: the number of active businesses declines with age and size,

Measuring the Dynamics of Young and Small Businesses

349

Fig. 9.4 Share of revenues and business numbers accounted for by small firms, less than $90,000 annual revenue, in selected four-digit NAICS industries Source: Own Calculations from ILBD.

but the bulk of activity—here measured by payroll—is concentrated in older and larger businesses. The more novel elements of table 9.7 appear in the last four columns, which provide information about ownership links between the employer and nonemployer universes. The at-risk population for this analysis is all employer businesses in our selected industries with positive payroll in 2000. For these employers, we consider all ownership links to nonemployers in the current and previous eight years, that is, in 1992 or 1994 to 2000. Out of the

1,416,292 432,027 242,636 139,368 36,886 16,426 4,332 3,103

2,291,070

Number of firms in age group (2)

213,292 190,690 170,091 153,400 134,315 117,723 199,523 1,112,036

2,291,070

a) 1–4 b) 5–9 c) 10–19 d) 20–49 e) 50–99 f) 100–249 g) 250–499 h) 500+

Total

Firm age in 2000, years since first appearance in employer universe (1)

0 1 2 3 4 5 6–7 8+

Total 100.00

2.28 3.27 3.40 3.44 3.40 3.10 5.61 75.50

Percentage of payroll in age group (3)

100.00

9.81 9.02 10.58 13.43 8.56 8.83 7.16 32.62

Percentage of payroll in size class (3)

265,738

35,082 31,314 27,197 25,244 20,675 20,615 24,823 80,788

Number of firms with backward links to nonemployers (4)

265,738

200,252 34,590 16,656 9,229 2,526 1,362 489 634

Number of firms with backward links to nonemployers (4)

11.60

16.45 16.42 15.99 16.46 15.39 17.51 12.44 7.26

Percentage of firms in age group with backward links (5)

11.60

14.14 8.01 6.86 6.62 6.85 8.29 11.29 20.43

Percentage of firms in size class with backward links (5)

100.00

13.20 11.78 10.23 9.50 7.78 7.76 9.34 30.40

Percentage of all backward links to nonemployers (6)

100.00

75.36 13.02 6.27 3.47 0.95 0.51 0.18 0.24

Percentage of all backward links to nonemployers (6)

100.00

11.01 9.08 7.95 6.46 5.49 4.48 10.81 44.74

Percentage of nonemployer pre-link revenues (7)

100.00

46.73 13.09 12.38 8.62 6.18 4.48 2.32 6.20

Percentage of nonemployer pre-link revenues (7)

Notes: Column (7) reports the pre-link percentage distribution of nonemployer revenues for nonemployers that link to firms in the employer universe. To calculate this distribution, we first express nonemployer revenues in 2000 dollars using the GDP deflator for all goods and services. Then, for each nonemployer that links to the employer universe, we take the value of its deflated revenue in the year prior to its first link to a firm in the employer universe. We sum these values across all nonemployers that link to the 2000 LBD for our selected industries. The percentages reported in the table are based on this total value of year-prior-to-link nonemployer revenue.

Number of firms in size class (2)

Employer links to nonemployers by size and age of employer, employers in selected industries in 2000

Firm size in 2000, number of employees (1)

Table 9.7

Measuring the Dynamics of Young and Small Businesses

351

2.3 million employers in our selected industries, about 266 thousand have ownership links to the nonemployer universe within the current or previous eight years based on the matching algorithm described previously. Columns (4) and (5) in table 9.7 report the number and percentage of employer firms with ownership links to nonemployers by employer size and age. While most employers have no ownership links to the nonemployer universe, many do, and this pattern holds for all size and age categories. Among firms with one to four employees, 14 percent link to the nonemployer universe within the previous eight years. The propensity for links to the nonemployer universe is U-shaped in employer size. Among firms less than six years old, more than 15 percent link to the nonemployer universe. The propensity for ownership links to nonemployer businesses declines after age five, but this pattern may simply reflect our inability to identify ownership links in 1993 and prior to 1992. Column (7) in table 9.7 reports the pre-link distribution of nonemployer revenues for those nonemployers that link to the employer universe. To construct this distribution, we sum deflated revenues over the nonemployer records that link to the 2000 LBD. We use the nonemployer’s revenue value in the year prior to the link. For example, if a 1997 nonemployer record links to an employer that operates in 2000, we use the deflated 1996 nonemployer revenue in the computation. Because there can be multiple dynamic links between employer and nonemployer, we take the oldest link and count each nonemployer at most once. Column (6) is constructed in the same manner as column (7), except that each nonemployer record receives a unit weight. Comparing columns (6) and (7) in the top panel yields the inference that relatively large nonemployers tend to link to larger employers. To see this, note that employers with at least 500 workers account for 6.2 percent of pre-link revenues among linked nonemployers, but only 0.24 percent of the linked nonemployers. Similarly, over 75 percent of the employer businesses with links to a nonemployer business have fewer than five employees, but they link to less than 48 percent of pre-link nonemployer revenues. When thinking about the process of business formation and growth, we anticipate a pattern whereby some businesses start as nonemployers, grow over time, and eventually transition to employer status, perhaps continuing to grow thereafter. This pattern holds for many businesses in the ILBD, but it is certainly not the only linkage pattern that arises. This point is evident in the lower panel of table 9.7, specifically in columns (6) and (7). More than 30 percent of nonemployer firms that link to the employer universe—and more than 45 percent of pre-link revenues—involve links to employers that are at least eight years old as of 2000. All of these cases involve nonemployer firms that link to previously established employer businesses. That is, they do not involve a nonemployer business that evolves into a new employer business.

352

S. Davis et al.

Figure 9.5 highlights this point in greater detail by plotting the frequency distribution of the age diﬀerence between employers in the 2000 cross section and their linked nonemployers (again using the first link to the employer). Recall that we know the exact age in years for employers that first appear in the employer universe after 1975. For nonemployers, we construct an age measure based on first appearance in the Nonemployer Business Register. Because we only begin observing nonemployers in 1992, examining this age diﬀerence for nonemployers observed first in 1992 is clouded by left censoring issues. As such, for figure 9.5, we examine all 2000 employers with links to nonemployers where the nonemployer first appeared after 1992 so that we have an accurate age for the nonemployer.6 For each of these linked nonemployers, we then compute the diﬀerence between its age and the age of the employer to which it links. According to figure 9.5, 60 percent of nonemployers are older than the employer to which they link. These cases are consistent with the standard pattern whereby a nonemployer business evolves into a new employer business. The pronounced mode at a one-year age diﬀerence reflects businesses that transition to employer status one year after inception as a nonemployer. Many other businesses operate in nonemployer mode for a few years before transitioning to employer status. In addition to these standard cases, figure 9.5 shows a large number of linkages in which the employer business predates the nonemployer business. These nonstandard linkage cases reflect other types of ownership relations between the two business universes. For example, an individual who owns a business with employees may also generate consulting income in a nonemployer business. As another type of example, a corporate business with employees may establish nonemployer subsidiaries for legal, financial, or tax reasons. 9.4.2 Nonemployer Transitions To continue our exploration of linkages between the two business universes, we now conduct an analysis of transitions. We first examine transitions from the nonemployer universe. In particular, we consider the population of 1994 nonemployer businesses in our selected industries and classify their operational status three years later in 1997. Figure 9.6 summarizes the three-year transition dynamics for the population of nonemployer businesses in one of our selected industries.7 The atrisk population is all nonemployer businesses with revenues in 1994. We classify outcomes into six categories:

6. We have also examined a version of figure 9.5 with the left censored cases included and the results are very similar to those reported. 7. The basic patterns for these transition dynamics are very similar over a six-year horizon, although the magnitudes change in the expected way (e.g., the share of activity accounted for by exits rises substantially).

Fig. 9.5

Age diﬀerence between matched employer and nonemployer records

354

S. Davis et al.

Fig. 9.6 Three-year transitions for the 1994 population of nonemployers in selected industries

• Exits: Businesses with positive revenue in the nonemployer universe in 1994, no revenue in the nonemployer universe in 1997, and no payroll in the employer universe in 1997. • Migrants: Businesses with positive revenue in the nonemployer universe in 1994, positive payroll in the employer universe in 1997, no revenue in the nonemployer universe by 1998,8 and the matched employer does not predate the nonemployer with respect to the age of its oldest establishment. • Other Transits: Businesses with positive revenue in the nonemployer universe in 1994, positive payroll in the employer universe in 1997, no revenue in the nonemployer universe by 1998, and the matched employer predates the nonemployer business. • Continuers: Businesses with positive revenue in the nonemployer universe in 1994 and 1997, and no payroll in the employer universe in 1997. • Persistent Duals: Businesses with positive revenue in the nonemployer universe in 1994, 1997, and 1998, and positive employer payroll in 1997 and 1998. • Other Duals: Businesses with positive revenue in the nonemployer universe in 1994 and 1997, positive payroll in the employer universe in 1997, no revenue in the nonemployer universe in 1998, and no payroll in the employer universe in 1998. We compute the share of the 1994 nonemployer analysis population that falls into each category on an unweighted and revenue-weighted basis. 8. We require no revenue in the nonemployer universe by 1998, rather than 1997, because a business that transitions during the 1997 calendar year will typically record positive 1997 revenues in both universes.

Measuring the Dynamics of Young and Small Businesses

355

Continuers account for 62 percent of nonemployer businesses and 58 percent of their revenues in 1994. Exits account for another 38 percent of nonemployers and 26 percent of their revenues. Migrants account for only 3 percent of nonemployer businesses but 9 percent of their 1994 revenues. While three percent is a small share of the population at risk, there are 7.4 million nonemployer businesses in our selected industries. In terms of raw numbers, approximately 220,000 nonemployers in 1994 migrate to employer status by 1997. Other Transits, Persistent Duals, and Other Duals account for very small shares of businesses and revenue. Figure 9.7 shows that Migrants and Exits occur with greater frequency among EIN cases than SSN cases.

Fig. 9.7 Three-Year Transitions for the 1994 Population of Nonemployers, SSN and EIN Cases

356

S. Davis et al.

9.4.3 The Contribution of Migrants to Young Employers We now quantify the contribution of Migrants to the economic activity of young employers with positive revenue in 1997 but no paid employees prior to 1995. Thus, young employers have had paid employees for at most three years. As reported in table 9.8, Migrants account for 28 percent of young employers and 20 percent of their revenues in our selected industries. Several industries stand out for the large role of Migrants, including Farm Labor and Management Services, Landscape and Horticultural Services, Painting and Paper Hanging, Carpentry and Floor Work, Insurance Agents and Brokers, Real Estate Agents and Managers, Computer and Data Processing Services, Automotive Repair Shops, Legal Services, Child Day Care Services, and Accounting, Auditing, and Bookkeeping. Migration from nonemployer to employer status is an important feature of business formation and growth in these industries. 9.4.4 The Growth Pattern of Migrants Next, we compare the growth rates of Migrants to two other groups of nonemployers—Continuers plus Exits and Continuers Only. For this purpose, we restrict attention to Migrants with nonemployer status as of 1996 that transit to employer status in 1997. That is, we consider Migrants with positive revenue in the nonemployer universe in 1996, positive revenue in the employer universe in 1997, and no revenue in the nonemployer universe by 1998. As before, we also require that a Migrant predate its matched employer. For Migrants with positive activity in both universes in 1997, its 1997 revenue value is the sum of its nonemployer and employer revenues. We also limit Exits and Continuers in the comparison groups to those with positive revenue in the 1996 nonemployer universe. We measure the growth rate as the change in annual revenue from t – 1 to t divided by the simple average of revenue in t – 1 and t.9 We compute all summary statistics on a revenue-weighted basis. As reported in table 9.9, Migrants grow much faster than other nonemployers. In the year prior to transition, the mean (median) growth rate for Migrants is 31 (15) percent, compared to 20 (5) percent for Continuers. In the year of transition, the mean (median) growth rate for Migrants is 101 (102) percent, compared to 6 (3) percent for continuers. In short, Migrants are on a trajectory of rapid growth before and during the transition to employer status.

9. The growth rate measure is bounded, symmetric about zero and ranges from –200 to 200 percent, with endpoints corresponding to exit and entry. See Davis, Haltiwanger, and Schuh (1996).

Table 9.8

Migrants from the nonemployer population in 1997, selected industries First appearance in the nonemployer population

SIC

76 78 172 175 176 275 367 412 421 472 554 581 621 641 653 721 723 729 737 738 753 784 792 799 801 802 803 804 811 832 835 872 874

Industry description

1995

1996

Percentage of young employers accounted for by migrants from the nonemployer population Farm labor and management services 24.5 13.1 15.8 Landscape and horticultural services 25.9 8.6 6.4 Painting and paper hanging 24.0 8.1 6.9 Carpentry and floor work 23.1 7.9 6.4 Roofing, siding, and sheet metal work 19.1 7.9 7.3 Commercial printing 13.4 5.6 4.8 Electronic components and accessories 9.8 4.3 6.4 Taxicabs 15.7 7.4 7.7 Trucking and courier services, except air 20.4 7.5 6.8 Passenger transportation arrangement 14.6 6.6 6.9 Gasoline service stations 8.4 3.6 3.5 Eating and drinking places 8.5 4.5 4.3 Security brokers and dealers 17.8 6.4 6.6 Insurance agents, brokers, and service 25.1 7.3 5.9 Real estate agents and managers 22.3 7.2 6.8 Laundry, cleaning, and garment services 17.9 6.5 5.3 Beauty shops 19.2 8.4 6.2 Miscellaneous personal services 22.2 8.4 7.2 Computer and data processing services 9.9 5.8 6.8 Miscellaneous business services 16.5 6.9 6.7 Automotive repair shops 19.0 7.4 5.5 Video tape rental 16.0 7.1 6.3 Producers, orchestras, entertainers 15.5 6.2 6.6 Misc. amusement, recreation services 12.3 5.4 5.4 Offices and clinics of medical doctors 11.4 4.3 4.5 Offices and clinics of dentists 17.2 6.6 4.9 Offices of osteopathic physicians 14.2 5.0 4.5 Offices of other health practitioners 18.4 6.8 5.9 Legal services 18.1 6.4 4.9 Individual and family services 10.6 4.0 3.7 Child day care services 21.9 8.0 6.0 Accounting, auditing, and bookkeeping 21.8 6.4 5.1 Management and public relations 12.3 5.6 6.6 All Selected Industries

76 78 172 175 176 275 367

1994 or earlier

16.1

6.3

5.7

All

53.5 40.9 39.0 37.4 34.2 23.8 20.4 30.8 34.6 28.0 15.6 17.3 30.8 38.3 36.4 29.7 33.8 37.8 22.5 30.1 31.9 29.5 28.2 23.1 20.2 28.7 23.7 31.1 29.4 18.3 35.8 33.2 24.5 28.1

Percentage of young employer revenues accounted for by migrants from the nonemployer population Farm labor and management services 20.3 18.9 19.2 58.4 Landscape and horticultural services 19.3 5.8 4.6 29.7 Painting and paper hanging 17.7 6.3 6.1 30.0 Carpentry and floor work 16.6 7.0 5.7 29.2 Roofing, siding, and sheet metal work 13.1 5.3 6.4 24.7 Commercial printing 7.3 3.7 4.1 15.0 Electronic components and accessories 4.0 1.0 3.3 8.3 (continued )

358 Table 9.8

S. Davis et al. (continued) First appearance in the nonemployer population

SIC

Industry description

1994 or earlier

1995

1996

All

412 421 472 554 581 621 641 653 721 723 729 737 738 753 784 792 799 801 802 803 804 811 832 835 872 874

Taxicabs Trucking and courier services, except air Passenger transportation arrangement Gasoline service stations Eating and drinking places Security brokers and dealers Insurance agents, brokers, and service Real estate agents and managers Laundry, cleaning, and garment services Beauty shops Miscellaneous personal services Computer and data processing services Miscellaneous business services Automotive repair shops Video tape rental Producers, orchestras, entertainers Misc. amusement, recreation services Offices and clinics of medical doctors Offices and clinics of dentists Offices of osteopathic physicians Offices of other health practitioners Legal services Individual and family services Child day care services Accounting, auditing, and bookkeeping Management and public relations

18.2 16.6 12.0 6.5 5.6 9.7 16.0 18.6 11.9 15.4 15.3 8.2 12.1 12.0 10.9 9.8 8.4 7.9 13.8 10.6 12.9 11.5 5.9 12.8 12.5 7.7

7.3 7.9 5.1 2.8 3.0 4.2 5.0 10.8 5.0 6.4 15.5 4.1 4.9 5.2 4.3 4.8 5.7 3.4 5.1 3.0 5.3 4.2 2.6 4.5 3.7 4.7

9.0 7.1 6.8 2.6 2.8 5.6 6.0 8.9 3.4 4.8 6.1 5.7 7.7 3.4 4.1 6.8 4.4 4.6 3.8 2.1 5.1 3.7 4.1 4.1 7.0 7.3

34.5 31.6 23.9 11.8 11.4 19.5 27.1 38.3 20.3 26.5 36.9 17.9 24.7 20.6 19.3 21.4 18.4 15.9 22.8 15.7 23.2 19.3 12.7 21.4 23.2 19.8

All Selected Industries

10.4

4.8

5.2

20.4

Note: “Young Employers” in 1997 are businesses that first hire one or more paid employees in 1995, 1996, or 1997. In this sense, they are 0 to 3 years of age as of 1997.

9.4.5 Employer Transitions Figure 9.8 summarizes three-year transition dynamics for the 1994 population of employers in our selected industries. As before, we group businesses in the at-risk population into six categories based on their status three years later. The categories mirror the ones considered previously. For example, Migrants now refer to businesses with positive revenue in the employer universe in 1994, positive revenue in the nonemployer universe in 1997, no revenue in the employer universe by 1998, and the employer predates its matched nonemployer. Continuers—businesses with positive revenue in the employer universe in both years and no ownership links to the nonemployer universe in

Measuring the Dynamics of Young and Small Businesses Table 9.9

359

Summary statistics for nonemployer revenue growth rates by transition status Time interval

Migrants (%)

Continuers and exits (%)

Continuers only (%)

Mean Median 10th percentile 90th percentile 90–10 differential

1995–1996 1995–1996 1995–1996 1995–1996 1995–1996

31 15 –26 141 167

23 5 –51 193 244

20 5 –44 139 183

Mean Median 10th percentile 90th percentile 90–10 differential

1996–1997 1996–1997 1996–1997 1996–1997 1996–1997

101 102 –5 200 205

–14 0 –197 70 267

6 3 –59 76 135

Notes: The analysis population contains all nonemployers in our selected industries with positive revenue in 1996 that are classified as either Migrants, Continuers, or Exits based upon their 1996–1997 transitions. Table entries report summary statistics for the distribution of annual revenue growth rates from 1995 to 1996 and from 1996 to 1997. The revenue growth rate is measured as the change in annual revenue from t – 1 to t divided by the simple average of revenue in t – 1 and t. All statistics are computed on a revenue-weighted basis.

Fig. 9.8 Three-year transitions for the 1994 population of employers in selected industries

1997—account for 68 percent of all employers and 74 percent of their 1994 revenues. Exits account for 19 percent of employer revenues and 27 percent of employer businesses. The exit figures point to high death rates for employers in our selected industries, but they are considerably smaller than exit rates for nonemployers (fig. 9.6). There are approximately 39,000 Migrants from the 1994 employer universe to the 1997 nonemployer universe, which amounts to about 2 percent of employers and 2 percent of their 1994 revenues. Other Transits account for 1 percent of employer businesses and

360

S. Davis et al.

revenues. Persistent Duals account for 1 percent of employer businesses and 3 percent of employer revenues. 9.5 Revenue Growth and Dispersion by Age and Size There is a vast body of research on the relationship of business growth patterns to business size and age. Almost all of this research restricts attention to businesses with employees, and much of it considers a subset of employers that meet a minimum size threshold or that include only publicly traded companies.10 Using the ILBD, we can systematically analyze and compare the dynamics of employer and nonemployer businesses. We consider business growth from 1999 to 2000 with attention to mean growth rates by size and age and the dispersion of growth rates within size and age categories. Our dispersion measure is the excess revenue reallocation rate: gross revenue gains at expanding units plus gross revenue losses at contracting units minus the absolute value of the net revenue change, all divided by aggregate revenue for the units under consideration. The excess reallocation rate is equivalent to the average absolute deviation of growth rates about zero, confirming its interpretation as a measure of crosssectional dispersion in growth rates.11 We consider all employers in our selected industries but limit attention to Continuers, Exits, and Migrants for the nonemployers. These three groups account for the vast majority of nonemployers and their revenues (fig. 9.6 and 9.7). For Migrants from nonemployer to employer status, our 2000 revenue measure includes nonemployer revenue, if any. In principle, we could treat Migrants from employer to nonemployer status symmetrically, but we ignore the matter as unimportant. Figure 9.9 shows that the mean growth rate of employers drops oﬀ very rapidly by age two and displays no clear relationship to age among older employers when we include Exits. The drop in mean growth with age is even more rapid among nonemployers. Indeed, mean nonemployer growth rates are negative for all ages beyond zero (i.e., beyond the year of entry). Conditional on survival, mean growth is positive at all ages and shows a clear tendency to decline with age. Figure 9.10 shows that excess revenue reallocation rates tend to decline with business age, especially for nonemployers. Perhaps more important, the magnitude of excess revenue reallocation is very large for employers and nonemployers alike, and at all ages. Excess revenue reallocation exceeds 50 percent in all age groups for nonemployers. It exceeds 30 percent in all age groups for employers. These results underscore the tremendous amount of revenue expansion and contraction that takes place on a routine 10. Dunne, Roberts, and Samuelson (1989), Sutton (1997), Caves (1998), and Davis and Haltiwanger (1999) review various branches of this literature, which spans several decades. 11. See Davis and Haltiwanger (1999), who review the use of this measure in the literature on job flows.

Measuring the Dynamics of Young and Small Businesses

Fig. 9.9

Fig. 9.10

361

Annual revenue growth rates by business type and age

Excess revenue reallocation rates by business type and age

basis among U.S. businesses. In this respect, the results echo previous findings on the large magnitude of simultaneous job creation and destruction in Davis, Haltiwanger, and Schuh (1996) and other work. When we condition on survival, considerable revenue churning remains, but it does not show a strong relationship to business age. Figures 9.11 and 9.12 display revenue growth and excess reallocation rates by business size. The size categories are narrow at the lower end to reflect the revenue distribution among nonemployers. As seen in figure 9.11, mean growth rate for small revenue classes is highly sensitive to whether we restrict attention to survivors. Conditional on survival, very small businesses have very high net growth rates relative to their larger counterparts. When we include Exits, the relationship between net revenue growth and size is basically flat. Figure 9.12 shows that excess revenue reallocation rates decline sharply with size for employers and nonemployers. Excess reallocation is high for businesses of all sizes, exceeding 20 percent even for the largest businesses.

362

S. Davis et al.

Fig. 9.11

Mean revenue growth rates by business type and size

Fig. 9.12

Excess revenue reallocation rates by business type and size

Perhaps surprisingly, excess reallocation rates among businesses with less than $120,000 in annual revenue are greater for employers than nonemployers. However, we know from figure 9.2 that there is little revenue activity in the very small size classes. Once again, conditioning on survival has a profound eﬀect on the size of the relationship, although excess revenue reallocation remains high in all size categories. 9.6 Where Do We Go From Here? The preceding sections describe the employer and nonemployer business universes, relate our eﬀorts thus far to integrate the two universes, and

Measuring the Dynamics of Young and Small Businesses

363

present evidence on the dynamics of young and small businesses, including the migration of nonemployers to employer status. In this section, we describe several challenges that arise in further developing the ILBD. 9.6.1 Issues Common to Both Business Universes One issue common to both business universes is the conversion from SIC to NAICS industry codes following the 1997 Economic Censuses. A related but larger set of problems involves the reliability and accuracy of industry codes in the ILBD. Large portions of the two business universes rely almost exclusively on administrative records for source data on industry codes. As a rough generalization, the industry codes are less reliable and less precise for nonemployers and for smaller employers. Geographic identifiers also tend to be less accurate for nonemployers and smaller employers. In general, there are fewer sources of information for business-level records that derive entirely from administrative sources, as compared to those that rely on administrative and survey sources. Another common issue pertains to the interpretation of business revenue measures and their consistency over time. In particular, revenue measures can be aﬀected by changes over time in income tax rules. This issue merits study. 9.6.2 Issues in the Employer Universe Multi-unit businesses above a size threshold are surveyed by the Census Bureau in the annual Company Organization Survey (COS). However, the list of such businesses is drawn from the prior economic census. These procedures mean that a firm’s transition from single-unit to multi-unit status often goes undetected until the next economic census. In addition, new establishments operated by small multi-unit firms not covered by the COS are detected only at the economic censuses. In both cases, the economic activity measures for these new establishments are included with older establishments of the parent company in the intercensal years. Hence, the delayed recognition of some new establishments in intercensal years leads to inaccurate establishment counts and, possibly, to an initially incorrect geographic and industrial classification for these new establishments. These issues are not critical for this chapter because our unit of analysis is the firm, but they are important for the development of the LBD and ILBD. Turning to another issue, the Census Bureau has made considerable progress in developing and maintaining longitudinal establishment identifiers for employer businesses, but the development of firm-level longitudinal identifiers remains an open area for research and development. Standard firm-level identifiers automatically change when a business undergoes certain types of reorganization, such as a change in its legal form of organization or a merger. In the previous analysis, we dealt with this issue by equat-

364

S. Davis et al.

ing firm age to the age of the oldest establishment operated by the firm. However, we identified entry and exit of firms based on changes in standard firm-level identifiers in the ILBD. 9.6.3 Issues in the Nonemployer Universe Some data issues unique to the nonemployer universe reflect the relatively recent availability and development of annual nonemployer files at the Census Bureau. For example, we cannot trace the inception of nonemployer businesses to years before 1992. In addition, as mentioned previously, the nonemployer data rely very heavily on administrative sources. The construction of longitudinal links for nonemployer business units also raises several challenges, and our work in this area is at a relatively early stage of development. There is room for improving the longitudinal and cross-sectional linkages via name and address matching, the treatment of joint returns for proprietorships (where there are separate firm identifiers for the filer and his or her spouse), and the reliability of employer identification information for nonemployer proprietors. 9.6.4 Integrating the Two Business Universes Studies of the integrated business-level data also face other challenges. First, the standards for classification by industry and geography diﬀer between the two universes. These diﬀerences limit our ability to isolate narrowly defined industries and regions. Second, at the most basic level, the unit of observation diﬀers between the two universes. For employers, the fundamental unit of observation is typically an establishment. For nonemployers, it is a tax return that reflects economic activity at the home or other locations. Our current analysis also aggregates tax filers with multiple Schedule C forms into a single nonemployer entity, even when each Schedule C involves quite diﬀerent business activities. Third, some firms with employees create aﬃliated business entities with no employees in order to shelter income from taxation or limit legal exposures and financial risks. We deliberately sought to sidestep the complex measurement issues associated with these special purpose business entities through our choice of industries. A thorough treatment of this issue for all industries is likely to require careful study of the legal framework and economic incentives governing the creation of special-purpose business entities. 9.6.5 Integrating Employee Records with the ILBD An exciting direction for future research is the integration of employee data with the ILBD. Using the longitudinal matched employer-employee data from the LEHD program at the Census Bureau, demographic and earnings data for the universe of employees can be integrated with the

Measuring the Dynamics of Young and Small Businesses

365

ILBD files.12 Integration of the employer, nonemployer, and employee data will provide an unprecedented opportunity to study firm, job, and worker dynamics. For example, it will be possible to follow someone who first works as an employee in a specific industry, then starts a small nonemployer business on the side, and later opens an employer business. More generally, for questions about where and how employer businesses originate, it will be very useful to know the demographic characteristics of business founders and their previous histories as employees and business owners. 9.7 Concluding Remarks It is tempting to think of the nonemployer business universe as a vast nursery for employer businesses. According to this view, many nonemployers evolve into employers and a few eventually grow into giant corporations that generate thousand of jobs. However, as our results confirm, most nonemployer business are quite small and never become employers. Indeed, it is misleading to think of all records in the nonemployer universe as businesses in the usual sense. Many nonemployer records reflect side jobs, hobby businesses, or occasional consulting engagements that generate extra income for households that depend primarily on wages. One goal of our research is to help the Census Bureau develop algorithms that can distinguish hobby businesses, for example, from other types of nonemployer businesses, including entrepreneurial undertakings that might evolve into larger businesses with employees. There is, however, a kernel of truth in the vast nursery view. Our evidence shows that migrants from the nonemployer universe account for a sizable share of young employers in the industries we study. These Migrants make up 28 percent of young employers (zero to three years old) and account for 20 percent of their revenues. Their importance varies considerably across industries. Among young employers, Migrants account for 38 percent of revenues in Real Estate Agents and Managers, 35 percent in Taxicabs, and 30 percent in Painting and Paper Hanging and Landscaping, but only 11 percent in Eating and Drinking Places. These figures probably understate the role of Migrants because of our conservative matching algorithms. In any event, the results indicate that a significant fraction of employers originate as nonemployer businesses. On the data front, this study takes important strides in developing an Integrated Longitudinal Business-level Database. Considerable work lies ahead, but the ILBD is already yielding useful information about the dy12. Another important direction for future work is the integration of the Characteristics of Business Owners (CBO) and Survey of Business Owners (SBO) data sets into the ILBD. Holmes and Schmitz (1995), among others, have shown the rich analysis that can be conducted with the CBO.

366

S. Davis et al.

namics of young and small businesses. Major strengths of the ILBD include comprehensive industry and geographic coverage, longitudinal links for establishments and firms, linkability to the large number of business surveys housed at Census, and an integrated treatment of employer and nonemployer business. The ILBD makes it possible to examine the behavior over time of virtually all businesses in the U.S. economy, employers and nonemployers alike, with robust samples and even entire populations.

References Boden, R., and A. Nucci. 2004. Business dynamics among the smallest and newest of businesses: Preliminary analyses of longitudinally matched nonemployer and Business Register data. Paper presented at Annual Meetings, Eastern Economic Association. 19–24 February, Washington, D. C. Caves, R. E. 1998. Industrial organization and new findings on the turnover and mobility of firms. Journal of Economic Literature 36 (4): 1947–82. Davis, S. J., and J. Haltiwanger. 1999. Gross job flows. In Handbook of labor economics ed. Orley Ashenfelter and David Card, 2711–2805. North Holland: Elsevier. Davis, S. J., J. C. Haltiwanger, and S. Schuh. 1996. Job creation and destruction. Cambridge, MA: MIT Press. Dunne, T., M. Roberts, and L. Samuelson. 1989. The growth and failure of U.S. Manufacturing Plants. The Quarterly Journal of Economics 104 (4): 671–98. Holmes, T., and J. Schmitz. 1995. On the turnover of business owners and business managers. Journal of Political Economy 103 (5): 1005–38. Jarmin, R., and J. Miranda. 2003. The longitudinal business database. Center for Economic Studies Working Paper no. 02-17. Nucci, A., and R. Boden. 2003. Demography of nonemployer businesses—preliminary evidence from the United States. Paper presented at the Comparative Analysis of Enterprise (micro) Data Conference. 15–16 September, London, England. Sutton, J. 1997. Gibrat’s legacy. Journal of Economic Literature 35 (1): 40–59.

Comment

Thomas J. Holmes

Over the past two decades, much progress has been made in the analysis of business dynamics. For the most part, the business population studied in this literature is the population of employers as opposed to nonemployers. Statistical agencies, like the U.S. Census Bureau, place the vast bulk of their data collection energies on employers and so that is the vast bulk of data available to researchers. One particularly important employer data set Thomas J. Holmes is the Curtis L. Carlson Professor of Economics at the University of Minnesota, a visiting scholar at the Federal Reserve Bank of Minneapolis, and a research associate of the National Bureau of Economic Research.

Measuring the Dynamics of Young and Small Businesses

367

is the Longitudinal Research Database (LRD), which matches annual survey data of employers with the quinquennial economic census information. The analysis of the LRD, including work by the some of the authors of this chapter, has had a major impact on the profession. Nonemployers are businesses operated by owner-managers (i.e., the subset of self-employed or partnership businesses with no employees). There exists a fairly large literature on self-employment. This literature is based mainly on surveys of individuals, such as the Current Population Survey (see for example, Evans and Leighton [1989]). But it also includes work based on federal income tax filings. Individuals who file a Schedule C (Proprietorship), a Partnership return, and a special kind of corporate return (S Corporation) when filing their personal income tax forms all are business owners of some kind. Some work uses the tax forms directly (Holtz-Eakin, Joulfaian, and Rosen 1994) or in conjunction with Census surveys, such as the Characteristics of Business Owner’s Survey that use the tax form population as the underlying business universe to sample from (Holmes and Schmitz 1995). This research team is developing the Integrated Longitudinal Business Data Base (ILDB) and the chapter reports its initial findings. The project makes two data contributions. First, it links the tax-based information about nonemployer businesses over time. The chapter does not emphasize this first contribution much, but there is potentially significant research potential in being able to follow nonemployers over time, even ones who never become employers. Second, and this is what the chapter emphasizes, it links the nonemployer records with records in the employer files, such as the records in the LRD. This is a very interesting project and the authors should be commended for engaging in this labor-intensive eﬀort. Given all the important work that has been done on firm birth and growth with the employer population, it is natural to try to find out information about a firm before it even hires its first employee and gets in the employer data set. Basically, the chapter finds that about 16 percent of all young employer firms can be traced as having a backward link in the nonemployer file. So in thinking about the life cycle of the firm, they do find some evidence of firms starting as a nonemployer and then later adding employees and entering the employer data set. But interestingly, (and perhaps reassuringly for those who want to focus on employer data sets) this happens only a small percentage of the time. The overwhelming majority of employer firms do not have nonemployer antecedents. Apparently, firms that are eventually going to employ somebody start out with employees in the vast majority of cases. The chapter shows that the links between the nonemployer and employer databases can form a complex web. As discussed previously, they find cases where a nonemployer in one year transits to employer status the following year, an easy case to understand. But they also find cases where

368

S. Davis et al.

large employer firms that have existed for many years have concurrent links with nonemployer firms. This can happen, for example, if a social security number for an owner of a nonemployer business is linked to an employer business. The employer business will have social security numbers in its file because business owners need to report them when initially filing for an Employer Identification Number (EIN). There are a number of ways these kinds of links can appear. First, an employer might want to organize a nonemployer subsidiary for tax or regulatory purposes. Second, an original founder of an employer business might leave and later start a new nonemployer business. There are a myriad of possibilities. Down the line, it would be useful if progress can be made in diﬀerentiating between some of the various factors accounting for the links between the employer and nonemployer universes. In conclusion, this research team has tackled a tough and messy project. It is messy because the underlying objects being studied are complex with complicated linkages. And there are more than the usual problems with linking data fields because of name changes, abbreviations, and so forth. While the problem is tough, I cannot think of a better team to take on the challenge. It is a first-rate team with extensive experience in this kind of data construction process. The team has put together an interesting data set that will be of use to researchers studying transitions from nonemployer status to employer status, as well as to researchers studying nonemployers for their own sake. While the project still seems in its early phases, the preliminary results are sensible. The results pass a “sniﬀ test” that the matching exercise is working. References Evans, D. S., and L. S. Leighton. 1989. Some empirical aspects of entrepreneurship. American Economic Review, American Economic Association 79 (3): 519–35. Holmes, T. J., and J. A. Schmitz, Jr. 1995. On the turnover of business firms and business managers. Journal of Political Economy 103 (October): 1005–38. Holtz-Eakin, D., D. Joulfaian, and H. S. Rosen. 1994. Entrepreneurial decisions and liquidity constraints. RAND Journal of Economics 25 (2): 334–34.

10 Producer Dynamics in Agriculture Empirical Evidence Mary Clare Ahearn, Penni Korb, and Jet Yee

10.1 Introduction The U.S. farm sector is characterized by a great deal of heterogeneity. This heterogeneity has been well-documented through Censuses of Agriculture (beginning in 1840) and a variety of surveys (such as the annual U.S. Department of Agriculture [USDA] farm household surveys, initiated in 19841). A major indicator of the heterogeneity within the farm sector is the size distribution of farms. One reason for the heterogeneity in farm sizes are the multiple objectives of producers, which in addition to profitmaking, include a high-quality rural lifestyle. More than three-quarters of farms have gross farm sales less than $50,000 and, on average, lose money farming. Other common indicators that exhibit extensive heterogeneity in the structure of the industry include the type of commodity specialization, the extent of commodity diversification, and various farm household characteristics, such as major occupation of the farm operator. While the traditional aggregate indicators capture the heterogeneity of agriculture, they also provide a picture of relative stability over time. According to the most recent Census of Agriculture (2002), there were about 2.1 million farms in the United States (USDA 2004a). That count is only 8 Mary Clare Ahearn is an agricultural economist with the Economic Research Service, U.S. Department of Agriculture. Penni Korb is an agricultural economist with the Economic Research Service, U.S. Department of Agriculture. Jet Yee was an economist with the Economic Research Service, U.S. Department of Agriculture, when this research was done. The authors would like to thank Spiro Stefanou for his helpful comments. The views expressed are those of the authors and do not necessarily represent the policies or views of USDA. 1. Since 1996, the survey is the Agricultural Resource Management Survey. Previously, it was the Farm Costs and Returns Survey (USDA 2004b).

369

370

Mary Clare Ahearn, Penni Korb, and Jet Yee

percent less than thirty years ago, and, in fact, there has been a slight increase in farm numbers in the last decade. Similarly, the average acres in a farm in 1978 were 449, compared to 441 in 2002. However, this relative stability in the number of farms and average size in acres masks a great deal of dynamics of entry and exit of farms, as well as changing of farm size for the new and continuing farms. Unlike the cross-sectional heterogeneity and the aggregate changes in farm characteristics over time, relatively little has been known about the dynamics of the changing U.S. farm sector. As is true for other industries in the economy until recently, this is largely because of the lack of access to a panel data set. Early research in the United States on farm turnovers relied on county-level estimates of farm exits (e.g., Goetz and Debertin 2001) and national analysis (e.g., Barkley 1990). Research that relies on panel data to examine turnover has largely focused on small geographical areas. For example, following the strained financial conditions of the 1980s, the USDA supported small area studies of farm exits in Wisconsin and Kentucky (Bentley et al. 1989; Bentley and Saupe 1990; Wu 1997). Foltz (2004) is a more recent example of a small area study, which focused on farm exits in Connecticut and the role of a government program to support dairy producers. The longitudinal file we employ in this chapter has its roots in a joint eﬀort during the 1980s by the Bureau of the Census and the Economic Research Service to link 2 Censuses of Agriculture. Early applications using this data file focused on forecasting future changes in the structure of farming (e.g., Edwards, Smith, and Peterson 1985). A recent study using the Census of Agriculture longitudinal file focused on explaining the determinants of exits at the national level (Hoppe and Korb 2005).2 The annual USDA farmer surveys and the Censuses of Agriculture were not designed to be panel data sets. However, in this chapter we use the Longitudinal Census of Agriculture file that was constructed over time by linking individual farm record data for the five censuses between 1978 and 1997 to document the extent of exit, entry, and growth in U.S. agriculture. Our analysis parallels a number of studies of firm or plant turnover for manufacturing industries in the United States (e.g., Dunne, Roberts, and Samuelson 1988). These studies of various nonagricultural U.S. industries are based on panel data built from economic censuses. In section 10.2, we describe the way in which the agricultural industry is defined and important characteristics of the industry that have implications for the framework to evaluate the dynamics of the industry. In section 2. Panel data for farms have been available in Canada and Israel for some time and determinants of turnover have been compared for those two countries (Kimhi and Bollman 1999). We would expect the results for Canada to most closely resemble the results for the United States. They found that a major factor explaining exits in Canada was farm size; the larger the farm the less likely to exit. Other determinants of exit were oﬀ-farm work (negatively related) and age of the operator.

Producer Dynamics in Agriculture: Empirical Evidence

371

10.3 we consider how well the general theoretical frameworks for considering industry dynamics apply to the agricultural industry. Section 10.4 provides a description of the longitudinal data file for agriculture and considers measurement of farm size. Section 10.5 of the chapter presents the empirical analysis of the longitudinal data, summarizing the findings regarding farm entry, exit, and reallocation over the period among surviving farms. We use the term turnover to capture trends on entry and exit of farms (and their inputs and outputs) and the term mobility to capture the trends in the reallocation of inputs and outputs of surviving farms. Finally, the chapter ends with conclusions and suggestions for future work. 10.2 Defining the Agricultural Industry In the United States, a farm is defined (in the Agricultural Census) as any place from which $1,000 or more of agricultural product was produced and sold, or normally would have been sold, during the year (USDA 2004a). Hence, it is a very liberal definition and one that assures a very diverse group of establishments will be counted in the farm population. It includes farms operated by households that are retired or attracted to farming for reasons not primarily related to production, such as the rural lifestyle or investment opportunities. In addition, since the definition is dollar-based, it is aﬀected as price levels change. Although changing the definition is regularly discussed, a liberal definition of a farm is very popular with many for a variety of reasons. For example, some Federal program dollars are distributed to states in part based on the farm population in a state (e.g., agricultural extension funds). The farm sector is a unique sector of the economy in a number of ways. The uniqueness can aﬀect the dynamics of exit, entry, and reallocation of the farm production industry, relative to other industries in the U.S. economy. First of all, most farms are closely-held businesses that combine the production and household choices in one decision-making unit. This requires that economic analyses take into consideration the utility of the household members over the life cycle, as well as profit-maximizing motivations. Secondly, a major input of farms—farmland—is considered fixed and immobile. This has technological, policy, and social implications. For example, the benefits of cost-reducing technologies and the massive government subsidies to the industry generally accrue to the owner of the farmland, who is not always the farm operator. The fixed supply of arable farmland is a focus of those concerned with long-term sustainability as well. Moreover, it could be argued that the family labor is also considered as immobile. A third unique feature of agriculture is its high total factor productivity (TFP) growth relative to most other sectors of the U.S. economy, in large part because of the public investments in Research and Development (Ahearn et al. 1998; Fuglie et al. 1996). While aggregate pro-

372

Mary Clare Ahearn, Penni Korb, and Jet Yee

ductivity growth is high, there is a great deal of heterogeneity in productivity across farms. Not unrelated to this heterogeneity is the primary importance of the farm as a residence to the majority of producers. Fourthly, agricultural commodity markets are generally characterized by instability, largely as a result of weather and low price elasticity of demand for food. And, finally, there is an unusually high level of interest in preserving the farming way of life, even by the general population. Public opinion polls have consistently revealed that the U.S. public has an interest in protecting the family farm from the vagaries of the marketplace, and this support has often translated into the transfer of subsidies to the agricultural sector and special treatment in the tax code. Some of the support is likely diﬃcult to separate from the public’s support of the environment and scenic vistas since approximately half of the 2 billion acres of U.S. land is in some type of agricultural use. Federal agricultural-specific policies have long been concerned with the restructuring and reallocation of outputs and inputs across agricultural producers. Current farm policies have their roots in a time around the Great Depression, when farm households were significantly worse oﬀ than most households. This condition and its cause in low farm prices due to surpluses of commodities are commonly referred to as the farm problem (Gardner 1992). A reason behind commodity surpluses is technological advances. The standard undergraduate agricultural economics models for understanding the relationship between innovation, surpluses, and reallocation of outputs and inputs are the treadmill and the farmer cannibalism models described by Cochrane (1958). Evidently, Cochrane was greatly influenced by Schumpeter’s Theory of Economic Development (1934). The dynamic nature of Schumpeter’s process of creative destruction, in particular, was key in Cochrane’s development of his models of innovation and reallocation for the farm industry (Levins 2003, p. 28). 10.3 Relevance of Theoretical Frameworks for Agriculture Schumpeter’s early work concerning the role of reallocation, in combination with an emerging literature that seeks to account for the heterogeneous performance across firms, forms the theoretical underpinnings of the current empirical work using micro-level firm data. The emerging theoretical models of industry dynamics are carefully reviewed in a number of sources, including Caves (1998). In his review article, Caves links the traditional industrial organization framework with new findings on exit, entry, and mobility of individual firms. The emerging theoretical models include Jovanovic (1982), Lambson (1991), Hopenhayn (1992), and Ericson and Pakes (1995). What the models have in common is that they assume firms have heterogeneous productive eﬃciency and are subject to various sources of uncertainty. These assump-

Producer Dynamics in Agriculture: Empirical Evidence

373

tions allow the models to explain the divergent paths of entry, exit, and reallocation that characterize the observed firm-level data. These common assumptions of the industry dynamics literature are very consistent with key characteristics of farm firms, as described previously. Namely, micro-level analysis of farms shows extensive cross-sectional productivity and cost diﬀerences (e.g., USDA 2004c) and farms are subject to numerous shocks, in particular the classic weather shocks. Numerous applications have considered the role of productivity in the dynamics of nonagricultural industries. Foster, Haltiwanger, and Krizan (2001) provide a recent review of the microeconomic evidence on productivity dynamics. What the emerging theoretical models do not adequately capture for farming is the role played by the most unique characteristic of farm firms, namely, the dual residence and business objectives of the majority of farm households. As mentioned, the majority of farms are small farms that usually lose money farming when returns only consider before-tax cash costs and returns. Many of these farm households likely receive a variety of returns from farming that are not captured in their before-tax cash income. For example, farm households may simply enjoy farming as a lifestyle and in general these households would have a shadow value of family labor that is less than their opportunity cost. Farm work may even be considered as a leisure activity by these households. The single most powerful trend in resource allocations of farm households during the past several decades is the allocation of household time to the oﬀ-farm labor market. More than 70 percent of U.S. farm households have someone in the household working oﬀ the farm. This high rate of oﬀ-farm labor participation is true, even in very rural areas of the United States. The unique relationship that a farm household has with the farm business means that micro decisions of farm businesses must be modeled along with micro decisions of farm households in a household production model. Farm households provide most of the labor on the farm and have a tripartite choice of time allocation (farm, oﬀ-farm, and leisure hours). The household production model is an extension of the basic labor-leisure model (e.g., Becker 1965). The conceptual model combines the decisions of agricultural households relating to production, consumption, and labor supply into a theoretically consistent model (e.g., Strauss 1986). The individual is assumed to allocate time to farm work, oﬀ-farm work, and leisure in such a fashion that the optimal allocation is achieved when the marginal values of time devoted to the activities are equal. Because of the dependence of farm households on oﬀ-farm income sources and the fixed supply of household labor, an important component of this literature is the empirical literature on estimating oﬀ-farm labor participation and supply (e.g., El-Osta and Ahearn 1996; Hallberg, Findeis, and Lass 1991; Mishra and Goodwin 1997). The household production model provides demands for farm household

374

Mary Clare Ahearn, Penni Korb, and Jet Yee

labor in farming, for leisure time (including personal maintenance time), and oﬀ-farm work. One of the possible solutions for the farm household is to provide no labor to the farm business; that is, to exit agriculture entirely. Recent work links agricultural productivity and state-level exit rates in explaining how various Federal policies have aﬀected structure, including participation in oﬀ-farm work, over a recent period in the agricultural industry (Ahearn, Yee, and Korb 2005). 10.4 Measurement Issues: Data Source and Farm Size 10.4.1 The Longitudinal File The Census of Agriculture has been conducted for over 150 years. In 1997, responsibility was transferred from the Bureau of the Census to National Agricultural Statistics Service, USDA. The Census of Agriculture Longitudinal file is currently a subset of the Census files, developed by combining individual farm operator records for five censuses (1978, 1982, 1987, 1992, and 1997) into one continuous record. Each record represents one individual farm operator’s responses about a farm operation to all and/or any censuses. Thus, farms can be followed for a twenty-year period. The file contains 4.5 million observations (records) and eighty-five analysis variables, such as the farm size, economic details about commodities produced, government program participation, county location of the farm, and demographic characteristics of the farm operator.3 One obvious weakness of the data for examining turnover and mobility is that the censuses are taken every five years (or four years for the 1978 to 1982 subperiod). Hence, yearto-year changes are likely underestimated. The longitudinal file attempts to follow farm operations that are tied to the farm land rather than follow individual farm operators. This is done using the Census File Number (CFN). The CFN identifies a farm operation for a particular census, and may follow a farm operation through subsequent censuses (up to five on the longitudinal file). If the farm continues from one census to the next, and the farm operator responds to the census using the same CFN, the information reported by that farm for that census period is appended to the longitudinal file using the same CFN. If the operation changes hands, either through sale or inheritance, the CFN may continue, it may change, or it may be terminated. For example, if an operator dies and leaves a farm to a surviving family member who continues to farm it, then a CFN should continue. However, if a surviving family mem3. Analysts can also add any additional variables that are collected on the censuses. The survey instruments are available at the back of each printed census volume and at the NASS website, www.nass.usda.gov

Producer Dynamics in Agriculture: Empirical Evidence

375

ber merges the inherited farm with another existing operation with that operation’s CFN, then the CFN may change. Farms that are split up may have a portion of their operation continue under the old number and the rest under new number(s), or all parcels of the operation may receive new numbers. In the case where a farm is sold for nonagricultural purposes, then the CFN is terminated. A farm is defined as going out of business when either the questionnaire is returned with the indication that it is no longer operating as a farm, or there is no response to repeated requests for information. The absence of a farm in a particular census year is represented in the longitudinal file by zeros for all the variables for that observation for that year. We consider a farm to be out of business (an exit) when zeroes in the CFN field indicate that the farm has been discontinued. When a particular CFN is classified as an exit through the process described previously, it is not possible to determine if the exit was the result of a merger with another farm or the result of the exiting farm being used for a nonfarm purpose. Likewise, a farm operation with a CFN that is not matched or linked to a previous longitudinal record would be considered a new business and added to the longitudinal file as a new record. This is an entry. A farm which has a CFN for both a beginning and an ending census time period in its record is considered to be a survivor. Most observations on the longitudinal file represent only themselves and are assigned a nonresponse weight of one. Some farms have a weight greater than one, meaning they represent themselves and other farms (or portions of farms) that did not respond to the Census. 10.4.2 Farm Size: Measurement and Aggregate Distribution There are a variety of ways in which farm size is measured, and the topic is occasionally reevaluated in agricultural economics (e.g., see Hanson, Stanton, and Ahearn 1989; Sumner and Wolf 2002; Yee and Ahearn 2005). Two of the most common ways to measure farm size in statistical publications are in terms of farm acres (an input measure) and the dollar value of gross sales (an output measure). The advantage of the acre-based measure is that land is generally viewed as a key production input in farming and an acre is a clearly defined unit of measurement. It can be considered as the counterpart to a measure of employment in manufacturing or service industries. The relative proportion of land as a production input varies considerably by technology, and the quality of an acre of land varies considerably over space. Output-based measures, such as the gross sales measure, avoid the major disadvantages of the acre-based measure. However, outputbased measures can interject biases as a result of the diﬀerences across commodities in farm value-added and in the changing value of the dollar over time, not to mention transitory output variations (Stanton et al. 1992). In this chapter, we focus largely on the acreage-based size.

376

Mary Clare Ahearn, Penni Korb, and Jet Yee

As mentioned, the average farm acreage has been quite stable in recent decades. The average acreage was 449 in 1978, compared to 441 in 2002.4 Average farm acreage is significantly greater in western states than in the East, but farm sizes are highly dispersed in all regions. This is in part because of the shape of the cost curve. It is generally thought to be L-shaped, with the low-cost plateau occurring at a relatively small farm size. Again, the role of technology is important here because there are diﬀering lowcost technologies available for farms of diﬀerent sizes. The prevalence of oﬀ-farm income is another reason we observe a significant share of small farms. While small farms may be eﬃcient from a technical viewpoint, they nevertheless usually do not generate enough cash income to support a family. Oﬀ-farm work opportunities allow many farm families to be engaged in farming and its lifestyle amenities. Both the small farms and the very largest farms are increasing as a share of the total farms during the period 1978 to 1997. The size distribution of farms is heavily skewed toward the small farms, while the production in agriculture is largely concentrated on the large farms. Approximately 2.4 percent of the largest farms (or 46,000 farms) accounted for half of all product in 1997.5 The increased concentration in production is only expected to continue in the future. However, it is still clear that with its more than 2 million farms, agriculture is not in danger of losing its poster child status for an industry characterized by many producer/sellers. 10.5 Empirical Evidence, 1978 to 1997 As mentioned, the traditional indicators of farm structure are widely available and document the extensive heterogeneity in farm structure. We focus our description of farm structure on the largely unavailable statistics of dynamic change for the 1978 to 1997 time period.6 We first consider turnover—exit, entry, and volatility—for all farms and by farm size. We next consider the mobility of surviving farms. 10.5.1 Turnover Exit, entrant, and surviving farm rates vary by inter-census time period (fig. 10.1). Many farms go out of business and many new farms come into business. In the 1978 to 1982 period and the 1992 to 1997 period, the number of farms that entered the farm sector exceeded the number of farms 4. The original 1997 census estimate was 487 acres per farm, and the revised estimate is 431 acres per farm. 5. By the year 2002, there were 34,000 (or 1.6 percent of farms) that accounted for half of the product. 6. The 2002 Census of Agriculture has not yet been added to the longitudinal database. With the release of the 2002 Census the number of farms in 1997 has been revised upward due to an adjustment in weights based on a survey of undercoverage. However, the revision to the 1997 Census will not be incorporated into the longitudinal file.

Producer Dynamics in Agriculture: Empirical Evidence

Fig. 10.1

377

U.S. Farm dynamics, 1978–1997

Source: Compiled by the Economic Research Service from Census of Agriculture data.

that exited. In the two intervening census periods, the opposite was true. For example, in 1997 62 percent of the farms that existed in 1992 were still in existence, and 38 percent of the 1992 farms had exited. However, slightly more farms entered farming during the period as exited. Contrast those significant changes to the slight increase in the net number of total farms between the 1992 Census of Agriculture and the 1997 Census. When annualized, the entry and exit rates are somewhat greater than those reported by Dunne, Roberts, and Samuelson (1988) for U.S. manufacturing industries in an earlier period. They report a 7.7 percent annual entry rate and a very similar 7.0 percent annual exit rate for the 1963 to 1982 period for manufacturing in general, but they also report significant variation across manufacturing industries. Another statistic used to characterize the turnover in an industry is the volatility of the industry. Volatility is defined as the sum of the entry and exit rate minus the absolute value of the net entry rate. It can be interpreted as a measure of the amount of producer turnover that is in excess of the amount needed to account for the change in industry size (Dunne and Roberts 1991). Given both the high rates of entry and exits, the volatility rate in farming is about double the entry and exit rates. The positive correlation between entry and exit rates is consistent with the findings reported for other industries (Caves 1998).

378

Mary Clare Ahearn, Penni Korb, and Jet Yee

Turnover statistics are sometimes calculated for output or employment. For farming, it makes sense to calculate entry, exit, and volatility rates for output (value of sales) and acres of farmland.7 The total land in agriculture is relatively stable over time, but there is also some shifting of land. Land moves to and from agricultural and nonagricultural uses, for example, between agricultural and forest uses. Land also shifts among agricultural uses, such as pasture and cropland. Much of the land operated by the farms that exit agriculture is subsequently purchased or rented by existing farms to expand their operation. In general, entry, exit, and volatility rates were lower for acres than they were for farm firms during these periods. Entry rates for farmland acres exceeded the exit rates in the first two subperiods and the opposite was true for the last two Census subperiods. Only in two Census subperiods was there a positive relationship between turnover in farm firms and turnover in farmland (table 10.1). Entry rates for value of sales were similar to those rates for acres of farmland, but exit rates, and hence, volatility rates, were lower for value of sales than they were for acres of farmland. This indicates that those acres that went out of production contributed less to value of sales per acre than the acres that continued in production. At the end of each of the four subperiods considered, farms that had entered during the period accounted for 28 to 32 percent of all sales of the farming sector. Turnover as exhibited by these data has been largely uncorrelated with market conditions. For example, the early 1980s were known as a diﬃcult financial time for some farms, especially those specializing in rice, cotton, and certain cash grains, such as corn and soybeans. Although we did not find strong evidence of Shumpeter’s creative destruction at work when we look at all farms combined, when we examine turnover by the type of commodity in which farms specialize we find some evidence that market conditions do play a role. During the 1982 to 1987 Census subperiod, more farms exited than entered for some cash grain, cotton, and rice producers. In addition, during this period, the average sales of the exitors exceeded the average sales of the continuing farms in these specialties. 10.5.2 Turnover by Farm Size In general, the average farm size in acres was larger for surviving farms than it was for either exiting or entering farms during the four subperiods (table 10.2). Also, the average size of surviving farms increased over the full 1978 to 1997 period by thirty-three acres, from 495 acres to 528 acres, although there was no change in the average size of farms during the last two subperiods. Exiting farms were larger in the initial subperiod and in the 7. This is not unlike the tracking of jobs in manufacturing industries; tracking of jobs in agriculture is problematic since much of the labor is unpaid and many workers are multiplejob holders.

Producer Dynamics in Agriculture: Empirical Evidence Table 10.1

Entry, exit, and volatility rates for farms, acres of farmland, and value of sales, for Census subperiods Entry rate (%)

Farms 1978–1982 1982–1987 1987–1992 1992–1997 Acres 1978–1982 1982–1987 1987–1992 1992–1997 Value of sales 1978–1982 1982–1987 1987–1992 1992–1997

Table 10.2

379

Exit rate (%)

Volatility (%)

Subperiod

Annual

Subperiod

Annual

Subperiod

Annual

37 33 32 39

11 9 8 10

33 40 38 37

10 10 10 9

66 66 63 74

20 18 17 19

28 34 26 30

8 8 6 7

26 30 30 32

8 7 7 8

53 61 51 60

15 14 12 15

31 32 29 35

8 8 6 8

20 33 24 24

5 8 5 6

39 63 47 49

11 15 11 11

Average farm size in acres of entering, exiting, and surviving farms, 1978–1997

1978–1982 1982–1987 1987–1992 1992–1997

Entrants

Exits

Surviving

344 391 373 380

359 355 357 428

495 498 528 528

1992 to 1997 subperiod than were the entering farms. Recall that it was in these two subperiods where entry rates slightly exceeded exit rates. That is, during the periods when more farms were entering farming than leaving farming, the average size of the entering farms was less than the average size of exiting farms. In the 1992 to 1997 period, the entry rate of farms exceeded the exit rate, but more total acres left agriculture than entered agricultural uses because the average farm that exited the sector was larger than the average entering farm, while at the same time the average size of the continuing farm remained constant from the previous period. Unlike the textbook explanation that holds that new firms enter at the optimal size, we find farms entering at all sizes. However, exit rates, entry rates, and survival rates vary considerably by size of farm. We have calculated entry and exit rates for farms and farmland for various size classes for the four subperiods (tables 10.3 and 10.4). Exit and entry rates are higher

Table 10.3

Exit rates by farm size (measured in acres) for farm firms and farmland

Acre class

1978–1982

1982–1987

1987–1992

1992–1997

1–49 50–99 100–179 180–259 260–499 500–999 1,000–1,999 2,000 plus

42 35 32 29 27 26 26 26

Exit rate for farms (%) 51 41 38 35 34 33 31 31

49 40 37 34 32 30 27 27

47 38 36 33 31 30 28 30

1–49 50–99 100–179 180–259 260–499 500–999 1,000–1,999 2,000 plus

38 35 32 29 27 26 26 25

Exit rate for acres of land (%) 47 41 38 35 34 32 31 31

45 40 37 34 32 30 27 28

43 38 35 33 31 29 28 34

Table 10.4

Entry rates by farm size (measured in acres) for farm firms and farmland

Acre class

1978–1982

1982–1987

1987–1992

1992–1997

1–49 50–99 100–179 180–259 260–499 500–999 1,000–1,999 2,000 plus

51 35 31 28 27 27 27 28

Entry rate for farms (%) 48 34 32 29 28 29 30 31

46 34 31 28 27 25 25 27

49 41 36 32 29 27 27 29

1–49 50–99 100–179 180–259 260–499 500–999 1,000–1,999 2,000 plus

45 35 31 28 26 27 27 28

Entry rate for farmland (%) 42 34 31 29 28 29 30 31

42 34 31 28 26 25 25 26

47 41 36 32 29 27 27 32

Producer Dynamics in Agriculture: Empirical Evidence

381

for small farms and decline steadily until farms reach a midsize range of 260 acres or more. The exit and entry rates flatten out considerably for the large size classes. In general, there are not large gaps between exit and entry rates over time. However, at the beginning of the period, entry rates slightly exceeded exit rates for large farms and the very smallest farms. (This was true for acres of land operated by those sizes of farms, too.) The early 1980s was the beginning of a time of significant financial stress for U.S. agriculture and the beginning of the consequent adjustment. There had been significant expansion in U.S. production prior to that period as international markets for U.S. products grew at a rapid rate, and then for a variety of reasons (including the contraction of the international demand for U.S. products) the United States had significant surpluses of agricultural commodities. The consequent adjustments are clear in the entry and exit rates. In the latter part of the period, 1992 to 1997, entry rates exceeded exit rates for the small farms. This reflects the growing demand for farms as high-quality rural residences, which continues today. It is also useful to consider relationships by farm size when size is measured by gross sales. The general trends in exit and entry rates by the gross sales size measure are similar to the acre size measure. By a gross sales measure, very small farms (with gross sales of less than $10,000) have the lowest survival rates. Survival rates are also low for the next size of farms ($10,000 to $99,999 in gross sales), but somewhat higher than for the smallest farms. These two smallest categories of farms represent about 85 percent of all U.S. farms, but only 10 percent of farm output. Prior to 1987, the survival rates of the midsized farms ($100,000 to $249,999) were on par with the largest farms in the sector, but since that time have been somewhat below the survival rates of the larger farms. Across the United States, since small farms are more likely to exit farming than large farms, we see the highest exit rates in those states that have large proportions of small farms. Small farms often require oﬀ-farm employment opportunities for their survival, and these are more likely to be available in or near metropolitan areas. The South and the East have the highest share of operators working oﬀ their farm full-time (200 or more days per year). In contrast, large farms require high quality agricultural resources—land and climate—and for some commodities, are recipients of government support. Farms in metropolitan areas are more likely to change ownership than farms in more rural areas for a variety of reasons, including the higher probability that farming is a secondary occupation of the operator and that the land is in higher demand for urban conversion. 10.5.3 Mobility of Surviving Farms We consider several aspects of the mobility of surviving farms in the results presented in tables 10.5 through 10.7. In table 10.5, we examine the survival rates, market shares (in sales and acres), and average farm size (in sales

382

Mary Clare Ahearn, Penni Korb, and Jet Yee

Table 10.5

Survival rates, market shares, and average farm sizes of entry cohorts by year 1978

1978 farms 1982 entry cohort 1987 entry cohort 1992 entry cohort 1997 entry cohort

1.000

1978 farms 1982 entry cohort 1987 entry cohort 1992 entry cohort 1997 entry cohort

1.000

1978 farms 1982 entry cohort 1987 entry cohort 1992 entry cohort 1997 entry cohort

1.000

1978 farms 1982 entry cohort 1987 entry cohort 1992 entry cohort 1997 entry cohort 1978 farms 1982 entry cohort 1987 entry cohort 1992 entry cohort 1997 entry cohort

1982

1987

Survival rate (farms) 0.703 0.482 1.000 0.446 1.000

Market share (sales) 0.725 0.508 0.275 0.169 0.323

Market share (acres) 0.730 0.543 0.270 0.157 0.299

1992

1997

0.331 0.279 0.486 1.000

0.228 0.189 0.307 0.482 1.000

0.377 0.129 0.212 0.282

0.265 0.097 0.152 0.183 0.303

0.427 0.118 0.193 0.261

0.312 0.090 0.135 0.159 0.304

Average size of surviving farms relative to all farms (sales) 1.000 1.110 1.058 1.055 0.793 1.020 1.147 0.912 1.139 0.820 Average size of surviving farms relative to all farms (acres) 1.000 1.117 1.131 1.194 0.780 0.948 1.053 0.846 1.038 0.760

1.069 1.260 1.280 1.098 0.778 1.258 1.170 1.142 0.950 0.780

and acres) of entry cohorts over time. The 1978 base period will include all farms. (This explains why the proportion is 1.0 in the 1978 column.) The cell that corresponds to the 1982 column of data and the 1978 farms row is for all farms that existed in 1978, regardless of when they entered, and survived through 1982. Hence, the 1978 row is not exactly comparable to the other rows where we can identify the entry cohort year. Notice that the 1-period survival rate for the 1978 farms was much higher (0.703) than the 1-period survival rate for the subsequent entry cohorts (i.e., 0.446, 0.486, 0.482). New entrants have lower survival rates than farms with more experience. The twenty-year survival rate for farms that existed in 1978 was about 23 percent. Also, the farms that existed in 1978 and survived over the periods were

Producer Dynamics in Agriculture: Empirical Evidence

383

all larger than the industry average at each period. In spite of their greater size, their market shares declined as a group because many of the farms that existed in 1978 exited the sector over time. The 1982 to 1997 entry cohorts had market shares from 28 to 32 percent, and controlled a comparable 26 to 30 percent of farmland. The bottom two sections of table 10.5 indicate the size of entry cohorts (in sales and acres) relative to all farms at the time period. In the initial census entry year, the entry cohort is smaller than all farms in the sector. However, by the next census period, these newest entry cohorts in the agricultural industry are at least 95 percent as large as all farms, but usually larger. For example, the entering farms in 1982 were 79 percent the size of the average farm. However, by 1997, the size of the surviving members of the 1982 cohort were 126 percent the size of the average farm. The average size of surviving farms can increase as the cohort ages because the smaller farms exit and/or the surviving farms increase in size. Supporting evidence suggest that both phenomena take place. It is also worth noting that the 1987 entry cohort stands out among the entry cohorts as being somewhat larger and capturing a higher market share than other years. This finding is consistent with the results we reported above on entry and exit rates by farm size for the 1982 to 1987 period. The results in table 10.5 are similar to those of Dunne, Roberts, and Samuelson (1988) who use Census of Manufactures data for 387 manufacturing industries for the 1963 to 1982 period. They also found that, on average, there is a decline in the market share of each entering cohort as it ages. This reflects a decline in the number of firms in the cohort that more than oﬀsets the rise in the average size of the surviving members of the cohort relative to all firms in the industry. Survival rates are lower in manufacturing than in agriculture. This could be true for a number of reasons described earlier (e.g., favorable government policies and the strong household link to the business). Dunne, Roberts, and Samuelson also reported lower market shares of entrants than we found for agriculture, perhaps because of the large share of farms that serve more as rural residences for the farm family than as money-making business ventures. Mobility of continuing farms in the shares of output (or acres of farmland) is usually measured by summing the absolute values of the diﬀerences between their output (acres) at t and t 1 and dividing by the sum of their output (acres) at t. Table 10.6 provides this measure of mobility for the four Census subperiods for both output and acres of farmland. When we consider all surviving farms, the mobility in output varies from a high of 72 percent in the first subperiod to a low of 51 percent in the 1982 to 1987 subperiod. The 1982 to 1987 period again stands out among the four subperiods. It had the smallest sum of (absolute value in) diﬀerences in output from the beginning to the end of the period and had the smallest share of farms that increased their sales during the period—less than half

384

Mary Clare Ahearn, Penni Korb, and Jet Yee

Table 10.6

Mobility indicators for output and acres of farmland for surviving farms, by Census subperiod Value of producta (%)

Acres (%)

Farms (%)

72 51 68 68

26 31 35 37

100 100 100 100

1978–1982 1982–1987 1987–1992 1992–1997

Value of product increased b 87 68 84 86

26 36 36 39

67 47 63 59

1978–1982 1982–1987 1987–1992 1992–1997

Value of product decreased b 41 41 43 43

29 28 34 36

33 53 37 41

All surviving farms 1978–1982 1982–1987 1987–1992 1992–1997

a

Value of product is the total value of sales and the value of product removed under contract in 1997 dollars. b Excludes less than 1 percent of farms that had no value of sales in either period.

of continuing farms increased their output during this period. This is a reflection of the financial stress in agriculture during that particular period. The mobility in acres is much less than for output for each subperiod, but shows a consistent increase over the subperiods. In agriculture, we expect more variation in output over time, given the vagaries of weather and market prices, in contrast to the input of farmland. Baldwin (1995) highlighted the diﬀerences in mobility in employment by dividing Canadian manufacturing firms into those that gained and those that lost employment. In a similar fashion, we divided farms into those that had increasing value of product and those that had decreasing value of product over the subperiods, and then examined their mobility in output and acres of farmland. Farms that increased their output had significantly greater mobility than those that decreased their output. Again, for acres we see lower mobility and no diﬀerence by whether or not sales increased or decreased over the subperiods. In table 10.7, we consider how surviving farms change size over time. We calculated detailed transition matrices for each of the four census subperiods by size of farm, where size is measured in acres. We present the share of those remaining in the size class at the end of the period as a share of those that started in the size class at the beginning of the period, averaged over the four time periods. (The shares do not sum to 1.0 because exits and entrants are excluded.) The majority of surviving farms stay in the same size class (i.e., are along the diagonal of the tables). The smallest farms (1

Producer Dynamics in Agriculture: Empirical Evidence Table 10.7

385

Transition matrix for surviving farms and acres, by acreage size class (four-period average) Ending acre class

Beginning acre class

1–49

50–99

100–179

1–49 50–99 100–179 180–259 260–499 500–999 1,000–1,999 2,000+

0.429 0.087 0.039 0.026 0.017 0.011 0.008 0.007

0.050 0.407 0.071 0.031 0.016 0.007 0.004 0.002

0.025 0.077 0.415 0.104 0.043 0.016 0.009 0.005

1–49 50–99 100–179 180–259 260–499 500–999 1,000–1,999 2,000+

0.411 0.076 0.036 0.024 0.016 0.010 0.007 0.005

0.064 0.379 0.062 0.029 0.014 0.007 0.004 0.002

0.027 0.077 0.387 0.095 0.039 0.015 0.008 0.003

180–259

260–499

500–999

1,000–1,999

2,000+

Number of farms 0.009 0.009 0.021 0.016 0.061 0.040 0.366 0.114 0.069 0.434 0.018 0.104 0.007 0.026 0.004 0.011

0.004 0.005 0.011 0.024 0.094 0.451 0.114 0.024

0.001 0.001 0.003 0.005 0.013 0.088 0.466 0.086

0.001 0.001 0.001 0.002 0.003 0.011 0.085 0.576

Total acreage 0.009 0.008 0.021 0.015 0.061 0.040 0.343 0.111 0.060 0.407 0.016 0.090 0.007 0.024 0.002 0.006

0.003 0.005 0.011 0.023 0.099 0.425 0.099 0.013

0.001 0.001 0.003 0.005 0.013 0.095 0.441 0.036

0.001 0.001 0.001 0.001 0.003 0.012 0.093 0.599

to 49 acres) and the larger farms have the highest share of farms remaining in their size class for each subperiod. Not surprisingly, the surviving large farms with an unconstrained size category were most likely to stay in the large category, since they had nowhere to go, except to contract. Midsized farms were the least likely to remain in their size class, and were somewhat more likely to contract in size than to expand. Given the stability in the farms along the diagonal, the size distribution (by these aggregated acreage classes) of all surviving farms has changed little over the subperiods. The stability of the size distribution of firms has been observed for nonagricultural industries as well. The stability in the size distribution of the surviving farms underscores the role played by new entrants and exiting farms in aﬀecting the aggregate size distribution. The results in table 10.7 indicate when surviving farms changed their size category. However, many farms stay within an arbitrary size category and change their acres operated to a smaller extent. We found that, over any given period, the majority of farms change size. For example, during the 1992 to 1997 period, only about 30 percent of the surviving farms did not expand or reduce their acres operated at all. And small farms are less likely to expand, while large farms are more likely to get larger. This result diﬀers for the result generally found for nonfarm industries, where the average growth rate of continuing firms generally declines with firm size. This

386

Mary Clare Ahearn, Penni Korb, and Jet Yee

Fig. 10.2 Surviving farm changes in acreage by farm specialization, 1992–1997 Source: Compiled by the Economic Research Service from Census of Agriculture data.

diﬀerence for farming is likely the result of the unique dual role played by farms as both business and residence, oftentimes with the residence being the predominant motivation for being in farming. There is some variation in the tendency of surviving farms to change size based upon the type of commodity in which they specialize. Some commodity specializations require more fixed investment than others; therefore, we would expect those specializations to exhibit more stability in farm size and turnover. This is certainly the case for farms that specialize in fruit and tree nuts and horticulture specialties in 1992 to 1997 (fig. 10.2). One might also expect relative stability in dairy operations, given the large fixed investment of that specialization. However, we see just the opposite. This is likely because of the increased economic pressure to increase dairy farm size. Restructuring of dairy production has been going on for some time, including the movement of production out of the traditional midwestern dairy states towards southern climates. California is now the largest dairy state in the nation.8 8. In fact, California is the top producer for one-third of the twenty-five major commodity groups.

Producer Dynamics in Agriculture: Empirical Evidence

387

The least developed area of the United States is the mid-region of the country; in contrast, the two coasts are the most developed. While there is significant variation in level of development within a state, the link between level of metropolitan development and farm survivability is even crudely evident at the state level. The lowest survivability of farms is along the two coastlines, which have the most developed land and the greatest proportion of small farms. The more populated areas of the United States have both greater entry and greater exit rates than the less populated areas. The midsection of the country clearly has the greatest farm survivorship. The midsection of the country has some of the highest rates of farm expansion, along with areas of the Southeast and the Southwest. The joint distribution of these two indicators of dynamics shows that in some areas, when farms exit, the farm resources are largely used to expand existing farms. This is the case for the Southeast. In others areas, like the Northeast, when farms exit, entering farmers are operating farms of approximately the same size as those that exiting farmers operated. The strong farm economy of the midsection of the country is evident from both high survivability levels and farm expansion. An important factor in the farm economy in this region is the high level of government farm subsidies. For example, in 2000, government farm payments were nearly $23 billion, and seven states in the midsection of the country exceeded $1 billion. Those states were Illinois, Indiana, Iowa, Minnesota, Nebraska, North Dakota, and Texas. Combined, they received half of the total subsidies in that year. 10.6 Summary and Conclusions This chapter draws on a unique panel data set, the longitudinal Census of Agriculture data for 1978 to 1997, to examine turnovers and mobility in U.S. agriculture. The data are widely used in their published aggregate form for individual censuses, but only recently used in a panel file. The microanalysis of turnovers shows considerable structural change underlying the traditional aggregate indicators of farm structure. For example, the data show that in 1997, 62 percent of the farms that existed in 1992 were still in existence, and 38 percent of the 1992 farms exited. In contrast, the net change reflected in the aggregate statistics reported a net change of 1 percent. This analysis for agriculture is in the same vein as work that has occurred for other industries in the United States and elsewhere, often focused on manufacturing. For example, Caves (1998) provides a review of these studies. Many of the stylized facts that have emerged from the literature synthesized by previous studies are relevant to our findings for agriculture. Our major findings for agriculture are: • The rate of entry and exit varies somewhat over time, and their generally positive correlation is consistent with the findings reported for

388

•

• • •

• • •

•

Mary Clare Ahearn, Penni Korb, and Jet Yee

other industries; however, entry and exit rates in agriculture are somewhat higher than in manufacturing. The entry and exit of farms are involved in the growth-farm size relationships. Although farms enter and exit at all farm sizes, entry and exit are more likely to occur among small farms. The entry, exit, and volatility rates were lower for farmland acres (a key input in farming) than they were for farm firms. Turnover and resource mobility have been largely uncorrelated with market conditions. The exception to this is for certain specializations during the stressful subperiod of 1982 to 1987. There is large-scale reallocation of outputs and inputs in agriculture. Entering farms, for example, account for about 30 percent of all output over the time period studied, which is higher than those shares reported elsewhere for manufacturing. Approximately 23 percent of the farms that existed in 1978 survived to 1997. Surviving farms are larger on average than either exiting or entering farms. The mobility of surviving farms in output varies significantly. The 1982 to 1987 subperiod had the lowest mobility and the smallest share of farms that increased their sales during the period—less than half of continuing farms increased their output during this period. The majority of surviving farms change their farm size, and in some cases the changes are large enough to put them in a larger acreage class, thereby aﬀecting the aggregate size distribution. Small farms are less likely to expand, while large farms are more likely to get larger. This result diﬀers for the result generally found for nonfarm industries, where the average growth rate of surviving firms generally declines with firm size. The diﬀerence for farming is likely the result of the unique dual role played by farms as both business and residence.

There is not a clear view regarding the most dominant factors in the structural change process in agriculture. Our description of farm structural change underlying the traditional aggregate indicators (i.e., the turnover and reallocation indicators) underscores the challenge in drawing simple generalizations about the process. The dearth of empirical applications of models is likely a result of the complexity of factors that are related to structural change and the importance of identifying their separate roles. The structural change process is a macro event that occurs rather slowly over time as a result of micro-level decisions. Hence, it is important to empirically consider both micro-level paths of firms and to measure the body of interactions over time to gain insight into the key determinants of productivity growth and change. Unlike for agriculture, the panel data set in manufacturing allows for an analysis of total factor productivity and the role

Producer Dynamics in Agriculture: Empirical Evidence

389

turnover and reallocation of output and inputs plays in the level of an industry’s aggregate total factor productivity. Another barrier to total factor productivity measurement with the Census of Agriculture data are the complexities associated with the variation in the natural resource input across farms and the fundamental importance of site-specific commodity mix. Although agriculture accounts for less than 2 percent of the GDP, the sector has special policy significance. One of those interests is the structure of the farming industry. In order to understand the sources of structural change, the relationships must be considered over time because of the lengthy lags of their impacts, policy and otherwise. In addition, structural change must be considered in the context of the whole farm sector because of the extensive linkages in the marketplace for land, inputs, and outputs, agricultural and otherwise. Our results support the view that the structural change process is a complex one, involving the interplay among technological change, market forces, and public policies. Consequently, policies designed to impact a single target, such as productivity or family farm survivability, will likely have reverberating structural implications, perhaps even counterintuitive or unwanted eﬀects.

References Ahearn, M., J. Yee, E. Ball, and R. Nehring. 1998. Agricultural productivity in the United States. Agriculture Information Bulletin No. 740 (January) Washington, D.C.: U.S. Department of Agriculture, Economic Research Service. Ahearn, M., J. Yee, and P. Korb. 2005. Eﬀects of diﬀering farm policies on farm structure and dynamics. American Journal of Agricultural Economics 87 (5): 1182–89. Baldwin, J. 1995. The dynamics of industrial competition. Cambridge: Cambridge University Press. Barkley, A. P. 1990. The determinants of the migration of labor out of agriculture in the United States, 1940–85. American Journal of Agricultural Economics 72 (3): 567–73. Becker, G. S. 1965. A theory of the allocation of time. Economic Journal 75 (299): 493–517. Bentley, S. E., P. F. Bartlett, F. L. Leistritz, S. H. Murdock, W. E. Saupe, D. E. Albrecht, B. L. Ekstrom, et al. 1989. Involuntary exits from farming: evidence from four studies. Agriculture Economic Report No. 625 (November). Washington, D.C.: U.S. Department of Agriculture, Economic Research Service. Bentley, S. E., and W. Saupe. 1990. Exits from farming in southwestern Wisconsin, 1982–1986. Agriculture Economic Report No. 631 (February). Washington, D.C.: U.S. Department of Agriculture, Economic Research Service. Caves, R. E. 1998. Industrial organization and new findings on the turnover and mobility of firms. Journal of Economic Literature 36 (4): 1947–82. Cochrane, W. 1958. Farm prices: Myth and reality. Minneapolis, MN: University of Minnesota Press.

390

Mary Clare Ahearn, Penni Korb, and Jet Yee

Dunne, T., and M. Roberts. 1991. Variation in producer turnover across U.S. manufacturing industries. In Entry and market contestability, ed. P. A. Geroski and J. Schwalbach, 187–203. Oxford, U.K.: Basil Blackwell Ltd. Dunne, T., M. Roberts, and L. Samuelson. 1988. Patterns of firm entry and exit in U.S. manufacturing industries. RAND Journal of Economics 19 (4): 495–515. Edwards, C., M. G. Smith, and R. N. Peterson. 1985. The changing distribution of farms by size: A markov analysis. Agricultural Economics Research 37 (4): 1–16. El-Osta, H., and M. Ahearn. 1996. Estimating the opportunity cost of unpaid farm labor for U.S. farm operators. U.S. Department of Agriculture, Economic Research Service Technical Bulletin No. 1848, March. Ericson, R., and A. Pakes. 1995. Markov-Perfect industry dynamics: A framework for empirical work. Review of Economic Studies 62 (1): 53–82. Foltz, J. 2004. Entry, exit, and farm size: Assessing an experiment in dairy price policy. American Journal of Agricultural Economics 86 (3): 594–604. Foster, L., J. Haltiwanger, and C. J. Krizan. 2001. 2000. Aggregate productivity growth: Lessons from microeconomics evidence. In New developments in productivity analysis, ed. C. R. Hulten, E. R. Dean, and M. J. Harper, 303–63. Chicago: The University of Chicago Press. Fuglie, K., N. Ballenger, K. Day, C. Klotz, M. Ollinger, J. Reilly, U. Vasavada, and J. Yee. 1996. Agricultural research and development: Public and private investments under alternative markets and institutions. Agricultural Economics Report 735. U.S. Department of Agriculture, Economic Research Service. Gardner, B. L. 1992. Changing economic perspectives on the farm problem. Journal of Economic Literature 30 (1): 62–101. Goetz, S., and D. Debertin. 2001. Why farmers quit: A county-level analysis. American Journal of Agricultural Economics 83 (4): 1010–23. Hallberg, M., J. Findeis, and D. Lass. 1991. Multiple job holding among farm families. Ames, IA: Iowa State University Press. Hanson, G., B. F. Stanton, and M. C. Ahearn. 1989. Alternative measures of farm output to classify farms by size. U.S. Department of Agriculture, Economic Research Service Technical Bulletin No. 1749. Hopenhayn, H. 1992. Entry, exit, and firm dynamics in long-run equilibrium. Econometrica 60 (5): 1127–50. Hoppe, R., and P. Korb. 2005. Understanding U.S. farm exits. U.S. Department of Agriculture Economic Research Report No. 21. Jovanovic, B. 1982. Selection and the evolution of industry. Econometrica 50 (3): 649–70. Kimhi, A., and R. Bollman. 1999. Family farm dynamics in Canada and Israel: The case of farm exits. Agricultural Economics 21 (1): 69–79. Lambson, V. 1991. Industry evolution with sunk costs and uncertain market conditions. International Journal of Industrial Organization 9 (2): 171–96. Levins, R. 2003. Willard Cochrane and the american family farm. Lincoln, NE: University of Nebraska Press. Mishra, A., and B. Goodwin. 1997. Farm income variability and the supply of oﬀ farm labor. American Journal of Agricultural Economics 79 (3): 880–87. Schumpeter, J. A. 1934. Theory of economic development. Cambridge, MA: Harvard University Press. Stanton, B. F., J. Jinkins, M. C. Ahearn, and G. Hanson. 1992. Perspectives on farm size and structure provided by value-added measures. Journal of Agricultural Economic Research 44 (2): 36–44. Strauss, J. 1986. The theory and comparative statics of agricultural household models: A general approach. Agricultural household models: Extensions, applica-

Producer Dynamics in Agriculture: Empirical Evidence

391

tions, and policy, In I. Singh, L. Squire, and J. Strauss, eds. 71–94. Baltimore, MD: Johns Hopkins University Press. Sumner, D., and C. Wolf. 2002. Diversification, vertical integration, and the regional pattern of dairy farm size. Review of Agricultural Economics 24 (2): 442–57. U.S. Department of Agriculture. 2004a. Agricultural Resource Management Survey Economic Research Service. (ARMS) Briefing Room. Available at http:// www.ers.usda.gov/Briefing/ARMS/ ———. 2004b. 2002 Census of Agriculture. Vol. 1, Geographic Area Series, U.S. and State Report, http://www.agcensus.usda.gov/Publications/2002/Volume_1, _Chapter_1_US/CenV1US1.tx ———. 2004c. Characteristics and Production Costs. Economic Research Service, Statistical Bulletin No. SB974. Available at http://www.ers.usda.gov/publications/ sb974/ Wu, H. 1997. An analysis of the farming decision: To farm full-time, part-time, or to exit farming. Academia Economic Papers 25 (1): 1–19. Yee, J., and M. Ahearn. 2005. Government policies and farm size: Does the size concept matter? Applied Economics 37 (19): 2231–38.

Comment

Spiro E. Stefanou

Ahearn, Korb, and Yee (AKY) embark on an interesting and challenging task in the first attempt to assemble and characterize a panel of farms using the U.S. Census of Agriculture. The authors undertake the goal of starting to organize the data to obtain a farm-level picture of the evolution of the farm size and structure. The measure of the agricultural activity is presented in terms of number of farms, average farm size, and value of farm activity over the panel. The period of focus, 1978 to 1997, is arguably the most interesting and relevant for measuring farming activity. The decline in the number of farms from over 5.65 million in 1950 had leveled oﬀ to just over 2 million by the beginning of the panel where it still hovers. The policy focus in the last quarter of the twentieth century has been on the restructuring and organization of agricultural production. Ahearn, Korb, and Yee (AKY) note the challenges with matching up the farm as a manufacturer with the manufacturing plants found in the Census of Manufacturing. The starkest contrast is with the arbitrary and static definition of a farm as an entity tied to a parcel of land that from which at least $1,000 of agricultural products were produced and sold (could have been sold) during the census year. The case can be made that the diﬀerences between agricultural commodity production (farm firms) and manufacturing production units (nonfarm firms) involve both broad and subtle diﬀerences. An analysis of producer dynamics in agriculture needs to reflect on these diﬀerences. Spiro E. Stefanou is a professor of agricultural economics at Pennsylvania State University.

392

Mary Clare Ahearn, Penni Korb, and Jet Yee

As we review AKY’s findings, I suggest there are five major forces distinguishing the nature of the farm from the nonfarm firm. Role of Nature: With production stages in farming tending to be short and not often automated, Allen and Lueck (1998) argue that the benefits of specialization are restricted. Only when farmers can manage the eﬀects of nature by abating the eﬀects of seasonality and random shocks to output does farm organization start to resemble a factory processes. For example, changes in livestock and greenhouse technologies have largely eliminated nature, allowing factory production to dominate. The panel that AKY have assembled can investigate a hypothesis that such firms are similar to nonfarm firms. Role of Land: Ahearn, Korb, and Yee (AKY) focus on how the census data are anchored to an asset, namely, land, which is a driving force in determining the potential production activities. While manufacturing plants do locate with a view toward input sourcing (e.g., in the case of fresh vegetable processing) or distribution networks, agricultural production facilities are often confined to locations for reasons due to family legacy. In addition to the production aspect of the land, the social ties can lend even less mobility in a farm firm locating to another site. Even still, the census denotes the land parcel as the equivalent to the manufacturing plant in the Census of Manufacturing protocol. On a broader scale, geography plays a role here similar to the case of manufacturing plants, with the reception that suburbanization pressures impact farms more acutely since they often must be more land extensive. Some manufacturing facilities are benign as far as the community is concerned (e.g., a furniture manufacturing facility or an apple orchard); still others are less welcomed (e.g., paper processing facility or a larger scale swine production facility) for reasons of eﬄuent and odor management. Role of Technologies: The nature of the farm firm is tied to production activity that is necessarily constrained by assets. Ahearn, Korb, and Yee (AKY) note that these technologies are often characterized by L-shaped average costs over significant ranges leading to the viability of a wide range of farm sizes. In addition, many agricultural production activities involve sunk investment and unrecoverable transactions costs leaving farmers less likely to either enter or exit the industry or respond to increases in output price levels with more capital investment (Chavas 1994). Taken together, one expects to see a wide range of farm sizes surviving and exits taking longer to be realized, ceteris paribus. Role of Government: There has been an active presence of U.S. policy in the farm sector since the 1930s to date, ranging from public R&D supportspecific commodity production activities to subsidies available to individual farming units. This policy has been historically scale neutral. The importance of the direct government payments to farmers is quite vari-

Producer Dynamics in Agriculture: Empirical Evidence

393

able, averaging about 22 percent of direct farm income with a coeﬃcient of variation of 65 percent (USDA 2008). The ownership of land is central with direct government payments being assigned to the landowner, who may or may not participate in rural life. However, the eﬀects of sunk costs in farm sectors with high price volatility can be alleviated, ceteris paribus, when price support programs focused on reducing output price uncertainty are introduced. Role of Lifestyle Choice: Agricultural production is arguably the last case of family production in the United States. From the perspective of the Census of Agriculture and even farm policy, the farm household is not so distinct from the nonfarm household. Farms that manage nature well enough to become factory-like enterprises are guided by the economics of decisions in calculating when to expand, when to extend operations, and so forth. But for the entry decisions, AKY note that these entrants are smaller farms, where noneconomic variables tend to drive these decisions, with Foltz (2004) oﬀering some micro data evidence. However, exits can be driven by the economic variables related to land and output price volatility. Having already mentioned the family ties to land as a reason for immobility, the identification for farming as a way of life is no small consequence in this complex and features largely in the entry decisions. The confluence of mitigating seasonal forces, price risk, sunk costs, and government programs for decision makers managing a way of life can almost surely lead to inconclusive theoretical predictions. The government policy tends to deemphasize programs to support farms that have managed to control nature as they evolve to a factory process and thus act as nonfarm firms. Hence, further analysis of a panel as AKY have assembled is an important first step and resource to address how structure, productivity, and government policy connect. Hypotheses explaining the diﬀerences need to focus on the impact of the diﬀerences between farm and nonfarm firms. As is the case with most census data studies, the skewness is substantial, with 1.6 percent of the farms accounting for half of the agricultural production in 2003. At the same time, 38.8 percent of all farms are drawing some form of direct government support (USDA 2008). One of the directions for the future is that analysts cannot be encumbered by the politically expedient, but economically trivial, hurdle that a farm is defined by $1,000 production value sold (or potentially sold). At some point, decisions must be made to streamline the meaningful set of units to be studied. For a start, farms being typified as rural residences comprise 66.2 percent of all farms in 2003 and can be eliminated from the set (USDA 2008). While farm input prices are rising over time, farm output prices remain flat. As a result, farms must grow and become more productive to maintain

394

Mary Clare Ahearn, Penni Korb, and Jet Yee

profitability. With food demand in the United States growing with population, growth beyond this rate must come at the expense of competitors of which one route is consolidation. The trend toward a bimodal distribution of farm size may be reflecting the consolidation of serious farming operations and the persistence of rural residences. The extent to which the gap between small and larger farms exists and persists will almost surely be tempered by introducing an alternative definition of a farm. When the picture of agriculture involves decision making units as households, productivity studies involving the entire farm sector should address the contributions to the family production unit, pecuniary and nonpecuniary. Given all these reasons why one may expect to see a much diﬀerent picture for farm and nonfarm firms, AYK find that agriculture presents an amazingly similar set of patterns to the manufacturing trends. The challenge is to look into the diﬀerences between farm and nonfarm firms’ entry/exit rates, turnover, and resource mobility pattern, and the emerging pattern of surviving farms tending to get larger over time, while nonfarm industries that exhibit the average growth rate of surviving farms decline with firm size. In summary, this chapter oﬀers an intriguing and useful work addressing the diﬀerences and similarities of the farm as a firm when compared to the manufacturing firm. Ahearn, Korb, and Yee (AKY) reveal the first glimpses of the micro data farm dynamics and set the stage for future discussion. I suspect that this eﬀort will push future research to make choices in defining farms for more focused debates connecting farm structure, productivity, and policy. References Allen, D. W., and D. Lueck. 1998. The nature of the farm. Journal of Law and Economics 41 (2): 343–86. Chavas, J. P. 1994. Production and investment decisions under sunk costs and temporal uncertainty. American Journal of Agricultural Economics 76 (1): 114–37. Foltz, J. D. 2004. Entry, exit, and farm size: Assessing an experiment in dairy price policy. American Journal of Agricultural Economics 86 (3): 594–604. U.S. Department of Agriculture, Economic Research Service. 2008. U.S. and State Farm Income Data. Available at http://www.ers.usda.gov/Data/FarmIncome/ finfidmu.htm

IV

Employer-Employee Dynamics

11 Ownership Change, Productivity, and Human Capital New Evidence from Matched Employer-Employee Data Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

11.1 Introduction In the 1990s, there was a substantial increase in the volume of assets transferred through mergers, acquisitions, and divestitures. This trend was especially pronounced outside of the United States. Gugler et al. (2003) report that the number of deals consummated in continental Europe increased from 986 during 1981 to 1990 to 8,609 during 1991 to 1998. The authors also note that the average value of these transactions rose from $186.1 million in 1991 to $414.1 million in 1998 (in constant dollars). This new wave of corporate restructuring has stimulated an important debate concerning whether these changes in ownership improve economic eﬃciency. Researchers typically address this question by analyzing the impact of ownership change on short-run stock prices (event studies), long-run stock prices, or accounting profits (e.g., Ravenscraft and Scherer 1987; Jensen 1988, 1993; and McWilliams and Siegel 1997). There are several problems with the use of such performance indicators. One problem with the use of Donald S. Siegel is a professor and dean of the School of Business at the University of Albany, SUNY. Kenneth L. Simons is an assistant professor of economics at Rensselaer Polytechnic Institute. Tomas Lindstrom is an economist at the Skandinaviska Enskilda Banken. Paper presented at the NBER/CRIW Conference on “Producer Dynamics: New Evidence from Micro Data,” Bethesda, MD, April 9, 2005. Comments from seminar participants at the University of Illinois-Urbana/Champaign, UC-Riverside, and the Network of Industrial Economists/ESRC/EPSRC Advanced Institute for Management Research Conference at the University of London, Rajshree Agarwal, John Baldwin, Eric Bartelsman, Nick Bloom, Tim Dunne, Rachel Griﬃth, John Haltiwanger, Rupert Harrison, Jonathan Haskel, Brad Jensen, Joe Mahoney, Pedro Martins, Mark Roberts, Anju Seth, and especially Judith Hellerstein, are greatly appreciated.

397

398

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

stock prices is that many economists question the validity of the eﬃcient markets hypothesis (see Shleifer 2001), which conjectures that changes in share prices following announcements of ownership changes reflect changes in future real economic performance. This is a critical issue, since market eﬃciency provides the basis for use of the event study methodology. Furthermore, McWilliams and Siegel (1997) have demonstrated that inadequate attention has been paid to research design issues in many event studies in the management literature. Accounting profitability has long been known to be an imperfect measure of economic performance (see Fisher and McGowan 1983). Policy decisions regarding the optimal level of ownership change should be based on analysis of the eﬀects of these transactions on economic eﬃciency. It is also important to note that many ownership changes involve privately-held companies or occur below the firm level (e.g., divisions of large, publicly-traded firms), which makes it virtually impossible to assess stock price or accounting profitability eﬀects, except for those transactions involving large, publicly-traded firms. The end result is that analyses of ownership changes based solely on information from public companies could yield misleading estimates of the antecedents and consequences of ownership changes. To overcome these limitations, several authors (e.g., Lichtenberg and Siegel 1987, 1990a, 1990b; McGuckin and Nguyen 1995; Maksimovic and Phillips 2001; Harris, Siegel, and Wright 2005) have asserted that a more desirable methodology is to assess the total factor productivity (TFP) of plants before and after ownership changes. Empirical evidence from the United States has been derived from the Census Bureau’s Longitudinal Research Database (LRD).1 The LRD is a plant-level file constructed by linking information from the quinquennial Census of Manufactures and the Annual Survey of Manufactures. Empirical evidence from the United Kingdom has been derived from the Annual Respondents Database (ARD). The ARD consists of plant-level records from the U.K. Annual Census of Production. Several of these studies have been based on restricted samples (e.g., a sample consisting mainly of long-lived plants or the universe of plants in one or two industries), which potentially limits the ability to generalize from these findings. From a welfare perspective, it is also interesting to examine how ownership change relates to characteristics of the workforce. Existing plant-level studies are constrained by having aggregate data on workers at the establishment and limited information on the composition of the workforce. That is, while authors can make inferences regarding the eﬀects of ownership change on the average worker at the plant, they cannot follow individ1. Excellent reviews of LRD-based studies are presented in Caves (1998) and Bartelsman and Doms (2000).

Ownership Change, Productivity, and Human Capital

399

ual employees at these establishments or analyze detailed changes in the composition of human capital that arise in the aftermath of ownership change. As a result, these studies have provided limited evidence on how changes in corporate control aﬀect the demand for diﬀerent types of workers. For example, policy makers might wish to consider the impact of ownership change on male versus female, Swedish-born versus immigrant, young versus old, and highly-educated versus non-highly-educated workers. The use of matched employer-employee data also enables us to directly explore the dynamics of workforce/human capital adjustment at the establishments involved in such transactions, in the sense that we can track the movement and relative compensation of employees who experienced an ownership change. The purpose of this study is to fill these gaps by relating productivity patterns associated with ownership change to numerous worker characteristics. Our empirical analysis is based on matched employer-employee data for over 19,000 Swedish manufacturing establishments, which constitute the majority of that nation’s population of manufacturing plants, for the years 1985 to 1998. Our sample also includes information on every worker in those plants, along with their complete work history during this fourteenyear period. We also assess whether there are diﬀerential eﬀects on productivity for diﬀerent types of ownership changes: partial and full acquisitions and divestitures, and unrelated and related diversification. Finally, we present the first plant-level findings from continental Europe and analyze more recent data on ownership change. The remainder of this chapter is organized as follows. Section 11.2 provides a review and critique of existing plant-level studies of the consequences of ownership change. A discussion of various theories relating to the impact of ownership change on economic performance is presented in section 11.3. Section 11.4 describes the construction of the micro data set and its salient characteristics. Section 11.5 outlines the econometric methodology. Section 11.6 presents empirical results. The final section, 11.7, contains preliminary conclusions. 11.2 Review and Critique of Plant-Level Studies of the Relationship between Ownership Change, Productivity, and Labor Demand 11.2.1 Productivity Table 11.1 presents a summary of plant-level studies of the relationship between ownership change and productivity. Several stylized facts emerge from this table. The first is that there have been no studies based on evidence from continental Europe. Second, most authors report that plants

U. S./annual data/mostly large continuous plants in the Longitudinal Research Database (LRD)

U. S./annual data/mostly large plants in the LRD

U. S./quinquennial Census of Manufactures/all plants

Canada/Census of Manufactures in 1970 and 1979/all plants

U. S./LRD/full sample/plant-level and divisional level

U. S./LRD matched to Compustat

U. K./Annual Research Database (ARD)/Full sample

Lichtenberg and Siegel (1990b)

McGuckin and Nguyen (1995) McGuckin, Nguyen, and Reznek (1998)

Baldwin (1998)

Maksimovic and Phillips (2001)

Schoar (2002)

Harris, Siegel, and Wright (2005)

Country/frequency/ nature of sample

Management buyouts

Diversification

Mergers and asset sales

Mergers and divestitures in the entire manufacturing sector

All ownership changes in the food manufacturing industry (SIC 20)

Leveraged and management buyouts (LBOs and MBOs) in the entire manufacturing sector

All ownership changes in the entire manufacturing sector

Type of ownership change

Plant-level studies of the eﬀects of ownership change on productivity

Lichtenberg and Siegel (1987)

Authors

Table 11.1

One-stage GMM estimation of augmented Cobb-Douglas production functions

Two-stage regressions of residuals from Cobb-Douglas production functions

Two-stage regressions of residuals from translog production functions

Regressions of nonparametric estimate of relative productivity (computed as value-added per worker) on ownership change dummies

Two-stage regressions of residuals from Cobb-Douglas production functions

Two-stage regressions of residuals from Cobb-Douglas production functions

Two-stage regressions of residuals from Cobb-Douglas production functions

Methodology

Plants involved in MBOs are less productive than comparable plants before the buyout; they experience a substantial increase in productivity after a buyout.

Plants that are acquired via diversification become more productive; however “incumbent” plants become less productive.

Acquired plants and divisions tend to be less productive; they experience an increase in productivity after the ownership change, the extent of which depends on whether the buying or selling division is “main” or “peripheral.”

Plants involved in ownership changes are more productive than comparable plants before the change in ownership; plants acquired by a firm in the same industry experience an increase in productivity.

Plants involved in ownership changes are more productive than comparable plants before the change in ownership; they experience an increase in productivity after the change in ownership.

Plants involved in LBOs and MBOs are more productive than comparable plants before the buyout; LBOs and especially MBO plants experience a substantial increase in productivity after a buyout.

Plants involved in ownership changes are less productive than comparable plants before an ownership change; they experience an increase in productivity after an ownership change.

Results

Ownership Change, Productivity, and Human Capital

401

involved in an ownership change experience an improvement in productivity after the change in ownership. Third, the magnitude of the productivity increase appears to vary by type of ownership change (e.g., leveraged buyouts versus management buyouts), which underscores the importance of disaggregating ownership change. Fourth, evidence on relative productivity before ownership change is mixed. Some authors report that plants involved in ownership changes are less productive than comparable plants before the change in ownership, while others report the opposite. These mixed results could be due to diﬀerences in the nature of the samples and the time frames of the analyses. Some authors have analyzed mostly large plants (e.g., Lichtenberg and Siegel 1987, 1990b), while others have focused on a single industry (e.g., McGuckin and Nguyen 1995). Several papers use quinquennial Census of Manufactures data, which makes it diﬃcult to analyze timing eﬀects with suﬃcient precision. This is potentially important since studies based on annual data indicate that major changes occur soon after the change in ownership. The first plant-level study of the relationship between ownership change and TFP was Lichtenberg and Siegel (1987), based on a (mostly) balanced panel of 20,493 U.S. LRD establishments in 450 manufacturing industries. In subsequent empirical work (Lichtenberg and Siegel 1990a, 1990b), the authors were able to analyze a more unbalanced sample of LRD plants. Their econometric analysis was based on the following two-stage approach. In the first stage, the authors computed residuals from within-industry (four-digit Standard Industrial Classification [SIC]) OLS regressions of log-linear Cobb-Douglas production functions of the following form (with error term suppressed): (1)

ln Qit 1 ln Kit 2 ln Lit 3 ln Mit

where Qit, Kit, Lit, and Mit refer to output, capital, labor, and materials, respectively, in plant i and year t. The residuals from equation (1) can be interpreted as an estimate of the relative productivity of each plant (i.e., relative to plants in the same industry). In the second stage of their model, the authors regressed the productivity residuals on a set of dummy variables denoting whether the plant had changed owners: (2)

RELPRODit f(OCits)

where RELPRODit is the productivity residual of plant i in year t, the error term is again suppressed, and OCits is a dummy variable that equals 1 if plant i was involved in an ownership change in year t s (where s can be negative or positive) or 0 otherwise. McGuckin and Nguyen (1995) conducted a similar analysis of the eﬀects of ownership change on economic eﬃciency, based on the complete population of plants in the food manufacturing industry (SIC 20) in the U.S. Census of Manufactures. They used the same method as in the previous

402

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

LRD-based studies to construct estimates of relative TFP, as well as labor productivity. However, they did not employ precisely the same second-stage approach, since they do not observe annual ownership changes, only those occurring between the quinquennial Census of Manufactures. Maksimovic and Phillips (2001) computed similar measures of relative TFP, based on the following translog production function (error term suppressed): (3)

ln Qit i 1 ln Kit 2 ln Lit 3 ln Mit 4 (ln Kit)2 5 (ln Lit)2 6 (ln Mit)2 7 ln Kit ln Lit 8 ln Kit ln Mit 9 ln Lit ln Mit 10AGEit t

where i is a plant-specific fixed eﬀect, t is a technology shift parameter, and AGEit denotes the age of the plant. Table 11.1 reveals that most authors have used a two-stage method to assess the antecedents and consequences of ownership change. In contrast, we estimate within industry (four-digit SIC), one-stage, augmented CobbDouglas production functions. We also experimented with using similar one-stage translog production functions, and found that this had little eﬀect on our econometric results. 11.2.2 Labor Demand Table 11.2 summarizes plant and firm-level studies of the impact of ownership change on employment and earnings. Much of the plant-level evidence seems to indicate that ownership change does not result in statistically significant declines in the employment and earnings of production workers at production establishments. In fact, the most comprehensive evidence, presented in McGuckin and Nguyen (2001), suggests that earnings and employment increase after ownership change. On the other hand, Lichtenberg and Siegel (1990a) find that employment and wage growth are lower in central oﬃce or auxiliary establishments in the aftermath of an ownership change, suggesting that white-collar workers suﬀer more than blue-collar employees when such transactions occur. Table 11.2 also reveals that these eﬀects vary by type of ownership change. For instance, Baldwin (1998) reported that mergers had a negative impact on the employment and compensation of nonproduction workers. Similar patterns emerge in the aftermath of leveraged and management buyouts. Bhagat, Shleifer, and Vishny (1990) find that 45 percent of the firms involved in hostile takeovers laid oﬀ workers, aﬀecting about 6 percent of the workforce. To the best of our knowledge, there have been no empirical studies of the impact of the eﬀects of ownership change on the demand for diﬀerent types of workers (e.g., men versus women, Swedish-born versus immigrant, younger versus older, highly-educated versus non-highly-educated). In the

U.S.

U.S.

U.S.

U.S.

U.S.

Canada

U.S.

U.K.

U. S. and Europe

U.K.

Brown and Medoﬀ (1988)

Bhagat, Shleifer, and Vishny (1990)

Lichtenberg and Siegel (1990a)

Lichtenberg and Siegel (1990b)

Baldwin (1998)

McGuckin and Nguyen (2001)

Conyon, Girma, Thompson, Wright (2001, 2002)

Gugler and Yurtoglu (2004)

Harris, Siegel, and Wright (2005)

Country

Plant

Firm

Firm

Plant

Plant

Manufacturing plants and auxiliary establishments

Plant and firm

Firm

Firm

Plant

Unit of observation

Management Buyouts (MBOs)

Mergers

Related and unrelated mergers

All ownership changes

Related and unrelated mergers; spin-oﬀs

All ownership changes

Leveraged buyouts (LBOs) and management buyouts (MBOs)

Hostile takeovers

3 types: simple sales, assets-only, sale, merger

All ownership changes

Type of ownership change Results

Plants involved in an MBO experience a substantial reduction in employment

Mergers do not reduce labor demand in the U. S.; there is a 10 percent decline in labor demand in Europe in the aftermath of mergers

19 percent decline in employment for related mergers; 8 percent decline in employment for unrelated mergers

For representative plants, wages and employment increase after ownership change; eﬀects worse for workers in large plants

Mergers and spin-oﬀs had very little impact on labor costs; related mergers had a positive impact on wages; mergers had a negative impact on employment and compensation of nonproduction workers

Employment and wage growth is significantly lower in auxiliary establishments changing owners than in those not changing owners, but not for R&D employees; much smaller eﬀects at production establishments.

Employment and wages of nonproduction workers (but not production workers) declines after an LBO; No evidence of a post-LBO decline in R&D.

45 percent of the firms involved in hostile takeovers laid oﬀ workers (approximately 6 percent of the workforce).

Simple sale: 9 percent increase in employment, 5 percent decline in wages; Assets-only sale: 5 percent decline in employment, 5 percent increase in wages; Mergers: 2 percent increase in employment, 4 percent decline in wages.

Labor input growth rates were lower for plants changing owners than comparable plants before the transaction; slightly higher after the transaction.

Plant- and firm-level studies of the eﬀects of ownership change on employment and earnings

Lichtenberg and Siegel (1987)

Authors

Table 11.2

404

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

following section, we provide a brief summary of theories relating to the impact of ownership change on performance. 11.3 Brief Review of Theories of the Impact of Ownership Change on Economic Performance Scholars have proposed numerous theories relating to the antecedents and consequences of changes in ownership. Several authors have asserted that corporate control changes reduce firm performance. This class of theories usually focuses on several managerial flaws. For instance, Dennis Mueller (1969) hypothesizes that managers attempt to maximize growth instead of shareholder wealth. This leads corporate executives to adopt policies that benefit them financially and professionally, at the expense of profit or shareholder wealth maximization. Unfortunately, these actions could lead to expansion of the firm beyond an optimal point. In a similar vein, Richard Roll (1986) and Mathew Hayward and Donald Hambrick (1997) argue that the hubris of CEOs and other managers causes them to systematically overestimate their ability to manage the companies they wish to acquire. According to this view, overconfidence induces managers to overpay for target firms, resulting in a decline in the economic performance of the acquirer. Michael Gort (1969) advanced a theory predicting that ownership change has a neutral impact on economic performance. In his framework, ownership change is induced by divergent expectations between buyers and sellers regarding the future value of corporate assets. He also seeks to explain fluctuations in merger activity over time. Thus, his model predicts that the magnitude of diﬀerences in buyer and seller expectations of the share prices of target firms is likely to be higher during periods of economic disturbance, which he defines as periods of sustained share price increases or when firms experience rapid technological change. Note that Gort’s model is a variation of the familiar theme of stockholder wealth maximization. That is, he assumes that the market expects no gain to result from the merger because acquirers have diﬀerent expectations than the market. Therefore, the premium earned by the acquired firm is exactly oﬀset by a loss to the acquiring firm’s shareholders. Several theories predict that ownership change has a positive eﬀect on economic performance. James Meade (1968) asserted that takeovers are part of a process of natural selection, whereby eﬃcient managers are rewarded through survival, while ineﬃcient managers are punished via takeovers. According to Henry Manne (1965) and Michael Jensen (1988), the takeover threat constrains the self-serving behavior of managers, and induces them to pursue profit-maximizing strategies. Jensen (1993) extends this theory by noting that certain types of ownership changes (e.g., management buyouts) result in changes in governance and incentive structures

Ownership Change, Productivity, and Human Capital

405

that reduce agency costs. These agency costs could be substantial in large corporations, where there is considerable separation of ownership and control. Frank Lichtenberg and Donald Siegel (1987) advanced a matching theory of ownership change, in which the fit between heterogeneous plants and owners is reflected in productivity. The matching theory of ownership change borrows heavily from the theory of labor turnover or job separation proposed by Jovanovic (1979). In the Lichtenberg and Siegel framework, low productivity signals a bad match, which is the key determinant of the firm-level decision to maintain or relinquish ownership of a given plant. The model has two empirical implications. The first is that the lower the productivity of a plant, relative to average productivity in its industry, the higher is the probability of an ownership change. A second implication of the model is that when an ownership change occurs, even an average match can be expected to lead to above average productivity growth because a better match will result. Thomas Holmes and James Schmitz (1990) outlined an equilibrium model of ownership change (or business transfer) that pertains mainly to smaller firms. In their model, high quality managers buy companies that implement high quality projects based on new ideas. Jovanovic and Rousseau (2002) also assert that high quality projects and high quality managers are complements. Moreover, they assert that mergers and takeovers play a role in spreading new technologies and reallocating capital to more eﬃcient uses and to better managers. Thus, according to the authors, ownership change plays a role that is similar to the eﬃciency-enhancing, dynamic adjustment associated with entry and exit. The notion that technological change and ownership change are complements suggests that these transactions should result in a decline in employment and skill upgrading.2 Many of the previously mentioned theories of ownership change do not have obvious implications for changes in the workforce below the top management level. Indeed, for ownership changes that occur according to the logic of the theories advanced by Mueller, Roll, Hayward and Hambrick, and Gort, we might be surprised to observe any substantial changes in a plant’s workforce in the aftermath of ownership change. In contrast, at least some of the theories that predict an improvement in productivity suggest that this enhancement may arise due to new managers’ changes to the workforce. Indeed, the theory of Jovanovic and Rousseau specifically addresses such changes. Thus, evidence on the degree to which workforce changes occur may shed light on the relative merit of the aforementioned theories of ownership change. 2. There is considerable evidence in the literature on skill-biased technological change (see Siegel [1999] and Link and Siegel [2007] for comprehensive reviews of this literature) that technological change is associated with downsizing and skill-upgrading of the workforce.

406

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

In the following section, we describe the data set that allows us to assess the impact of ownership on performance and workforce characteristics. 11.4 Data Our empirical analysis is based on a special file that links detailed information on Swedish workers and the establishments that employ them. This file has data on the output and inputs of these plants, which enables us to construct estimates of total factor productivity (TFP). It also contains information on a wide variety of worker characteristics, such as level of education, age, gender, and national origin. 11.4.1 Plants The primary unit of observation in our study is the plant. Following conventional international standards, the plant or establishment is defined as a physically independent unit within a firm. It is assumed that each plant focuses on just one line of business (i.e., one activity). If a company is involved in multiple activities at the same physical address, the firm is asked to report separate figures for each activity. Each figure is then tied to a separate plant or sometimes recorded under the plant’s largest activity. In most cases, however, firms focus on a single activity, implying that the local units are seldom split into several plants. Plants that were considered to be nonactive and help plants, such as sales oﬃces (or what would be considered auxiliary establishments in the United States), were also excluded from the data. According to Swedish law, each business is required to report information to Statistics Sweden on an annual basis. In 1946, the certainty criterion for inclusion in the annual survey of manufacturing plants was established at a minimum of five employees and 10,000 SEK (about 1,300 U.S. dollars) in production value. In 1990, this certainty threshold was raised to a minimum of ten employees, while a sampling procedure is applied to the smaller plants. In 1997, the certainty threshold oﬃcially was raised to a minimum of twenty employees, but as will be seen shortly, evolving sampling procedures for smaller plants meant that this change had little eﬀect.3 Tables 11.3 and 11.4 compare our sample of 19,010 plants to the population of Swedish manufacturing establishments. Table 11.3 contrasts the size distribution of our sample (top panel) with the corresponding size distribution for the population of Swedish manufacturing plants (bottom panel) in 1986, 1990, and 1995. These figures reveal that our sample is not completely representative in terms of size, since it is more heavily weighted towards plants with more than ten employees. On the other hand, table 3. We have a small number of mining plants in our sample. The threshold increases in 1990, and 1997 only aﬀected manufacturing plants.

14.4 6.6 5.9

0.4 5.2 5.8

33.1 36.0 38.5

1986 1990 1995

1986 1990 1995

22.4 22.6 23.4

5–9 Employees

5 Employees 50–99 Employees

Sample of Swedish manufacturing plants 28.0 13.2 29.9 13.0 30.8 12.7

20–49 Employees

Population of Swedish manufacturing plants 16.1 13.8 6.7 15.8 12.8 5.9 14.8 12.2 5.3

28.3 30.0 30.6

10–19 Employees

4.0 3.6 3.0

8.0 7.7 7.1

100–199 Employees

2.3 2.2 1.9

5.1 5.0 4.9

200–499 Employees

1.3 1.0 0.8

2.7 2.7 2.3

500 Employees

Comparison of size distribution of sample of Swedish manufacturing plants to population of Swedish manufacturing plants (Percentages)

Year

Table 11.3

408

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

Table 11.4

Sample plants (N 19,010) relative to population of Swedish manufacturing plants

Variable % of plants with more than 20 employees included in our sample % of total employment in plants with more than 20 employees included in our sample % of plants with more than 10 employees included in our sample % of total employment in plants with more than 10 employees included in our sample % of plants with more than 5 employees included in our sample % of total employment in plants with more than 5 employees included in our sample

1986

1990

1995

85.6%

91.2%

94.5%

92.0%

95.7%

98.6%

79.8%

84.3%

87.5%

89.7%

92.4%

94.7%

63.6%

62.9%

62.4%

84.9%

87.0%

90.7%

11.4 indicates that the sample constitutes a large fraction of economic activity in the manufacturing sector, especially for plants with more than ten employees. 11.4.2 Ownership Change Table 11.5 presents statistics on the incidence of ownership change. Over the entire sample period (1985 to 1998), 5.1 percent of plants experienced at least one ownership change. These rates of plant turnover appear to be slightly higher when they are weighted by value-added and employment (columns 2 and 3). An analysis of the annual figures reveals that the incidence of ownership change appears to have risen during the late 1980s, reaching a peak in the early 1990s. In table 11.6, we present evidence on the incidence of several types of ownership change involving our sample of plants during the sample period (1986 to 1998). We can identify whether an acquisition or divestiture involves the buying or selling of an entire firm. Note that the overwhelming majority of such changes are full acquisitions or divestitures, although the relative importance of such transactions diminishes when they are weighted by value-added or employment (columns 2 and 3). The full and partial acquisition categories indicate whether all or part of a firm is acquired: 4.2 percent of plants in a year changed owners as part of a full-firm takeover, and 0.9 percent changed owners through a part-firm takeover, summing to the total annual figure of 5.1 percent. The full and partial divestiture categories indicate whether the original owner ceded ownership of all versus some plants in a firm, regardless of the new owner(s) of those plants. We have also identified whether the buyer had existing plants in the same (four-digit) industry, which we refer to as a related acquisition.

Ownership Change, Productivity, and Human Capital Table 11.5

409

Incidence of ownership change for 19,010 Swedish manufacturing plants (1986–1998)

Year

% of plants involved in an ownership change

% of value-added involved in an ownership change

% of employment involved in an ownership change

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998

3.2% 4.3% 5.5% 5.0% 4.8% 4.8% 5.6% 6.0% 4.6% 3.9% 3.9% 3.7% 3.2%

3.1% 5.2% 8.3% 5.1% 7.7% 7.8% 5.0% 4.7% 7.3% 6.0% 2.1% 4.7% 2.3%

3.3% 5.7% 7.5% 5.6% 8.2% 7.4% 5.7% 5.2% 6.7% 5.3% 3.1% 3.8% 3.0%

Entire period

5.1%

5.4%

5.6%

Table 11.6

Incidence of ownership change for 19,010 Swedish manufacturing plants (1986–1998) by type of ownership change

Type of ownership change All ownership changes Full acquisition Partial acquisition Full divestiture Partial divestiture Related acquisition Unrelated acquisition Change in ownership involving a single firm

% of plants involved in a particular type of ownership change

% of value-added involved in a particular type of ownership change

% of employment involved in a particular type of ownership change

5.1% 4.2% 0.9% 4.5% 0.7% 0.9% 0.6%

5.3% 2.7% 2.6% 3.3% 2.0% 1.2% 0.8%

5.6% 3.2% 2.4% 3.9% 1.8% 1.4% 0.6%

3.7%

3.4%

3.7%

11.4.3 Capital A critical issue in the calculation of total factor productivity (TFP) is construction of a capital measure. Some researchers avoid analyzing TFP and instead compute labor productivity (LP), which is easier to measure. We will present econometric results based on both TFP and LP. We calculated estimates of the capital stock as follows: initial values of capital were estimated in 1989, based on the assumption of a constant capital-to-sales

410

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

ratio across all plants in each two-digit SIC industry. Using these initial estimates, capital is constructed using the usual perpetual inventory algorithm, Kcit (1 – c)Kcit–1 tIcit, where i denotes a plant, t denotes a year, c is either machinery or buildings and land, K denotes capital, I denotes investment, denotes the depreciation rate, and denotes an investment deflator.4 The capital estimates for machinery, plus buildings and land were summed to create a single combined capital stock measure, Kit. 11.4.4 Employees Matched employee-level data come from a database on the jobs and earnings of every employee in Sweden. The data are based on tax filings and hence record each employee’s annual earnings, which distinguish yearlong work-related earnings from other earnings.5 Employment is recorded in November of each year, and the records match employees (with a tiny percentage of missing cases among our manufacturing employees) to specific plants, firms, and (five-digit SIC) industries. Because the database covers all employees as needed for relevant tax records, we are able to infer that any employee whose record is missing in a given year was not employed in Sweden in that year. The full database contains 36,398,617 records across the fourteen years of data from 1985 to 1998, for an average of 2.6 million workers per year, consistent with the Swedish population of close to 9 million. Among all of the records, 9,251,962 records pertain to cases in which a person is employed by a manufacturing plant in our sample during the relevant year. The employee data include information on the gender, national origin, year of birth, most recent year of education, and number of years of education for each employee. We use this information to construct measures at the plant and employee levels of workforce characteristics. At the plant level, we assess (in each year) the percentage of workers who are male versus female, the percentage who were born in Sweden versus immigrated, the mean age of employees, mean years of experience (as proxied by the number of years elapsed since their last year of education), and the percentage of employees with at least some college-level education. At the employee level, we assess (in each year) the gender of the employee, whether he or she was born in Sweden, was below or above the mean age, below or above the mean level of experience, and determine educational status based on the following four categories: (a) less than a high school educa4. The depreciation rate for machinery was allowed to diﬀer by three-digit SIC industry and was taken from figures of the OECD, while a constant figure of 0.0314 was used for buildings and land. Investments were deflated using manufacturing sector-wide annual investment deflators reported by Statistics Sweden. We rely on figures kindly provided by M. Carlsson and replicate and extend the methods he used. 5. The data do not include hours worked or hourly wages, only the employee’s annual total income from employment.

Ownership Change, Productivity, and Human Capital

411

tion (up to ten years apparently equivalent to U.S. grade 8), (b) high school education (up to fourteen years apparently equivalent to U.S. grade 12), (c) at least some college or technical school education, and (d) at least some PhD-level studies. For our analysis of employee turnover in the aftermath of ownership change (where do old employees go and where do new employees come from), we also assessed whether employees transitioned to jobs in the same four-digit SIC industry versus jobs in other manufacturing or nonmanufacturing industries, and whether each employee was working in Sweden in each year.6 11.5 Econometric Models 11.5.1 Types of Models In this version of the chapter, we estimate two types of models. For analyses of labor and total factor productivity, ln Qit fit it

(4)

where Qit denotes plant i’s output in year t, fit is the logarithm of plant i’s production function in year t, and it is an eﬃciency residual. The eﬃciency residual is assumed to be influenced by ownership change and other variables, as follows: 12

it ∑ sOCits xit εit

(5)

s13

where Σ sOCit–s parameterizes the relation to ownership change as discussed in the following, is a vector of coeﬃcients, xit is a vector of control variables for plant i in year t, and εit is the remaining eﬃciency residual. Rewriting (4) thus yields 12 s–13

12

(6)

ln Qit fit ∑ sOCits xit εit. s13

Other analyses, which are not based on estimation of a production function, assume the same form: 12

(7)

yit ∑ sOCits xit εit. s13

where yit is the dependent variable in question (e.g., employment or earnings), is an intercept parameter, and the other terms are as defined previously. 6. Employees’ four-digit SIC industries of employment were assessed using 1969 SICs where possible, for comparability with the plant-level analyses. However, 1969 SICs were not available in all years of data, so 1992 SICs were used to assess industry of employment when 1969 SICs were not available in both the year in question and the comparator year.

412

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

11.5.2 Ownership Change The treatment of ownership change in the econometric analysis requires careful consideration. In equations (5) through (7), s denotes the year relative to the year of ownership change, so that negative values of s signify years preceding ownership change, s 0 denotes the year during which the plant changed owners, and positive values of s pertain to years following ownership change. A dummy variable OCit–s equals 1 if plant i’s owner changed (with certainty) s years preceding the current year t for s 0, or |s| years following the current year for s 0, or 0 otherwise. Note that our sample allows us to identify each plant’s owner for the years 1985 through 1998, so a new owner can be identified in each year for 1986 through 1998. For a plant observed in 1985, we wish to know whether an ownership change will occur for up to thirteen years in the future, while for a plant observed in 1998, we wish to know whether an ownership change occurred up to twelve years in the past. This consideration of past and future ownership changes yields a possible range of leads and lags from –13 to 12. The relation of past and future ownership change to productivity, size, or workforce characteristics can then be assessed, at each value of s, by including in the model the terms Σ12 s–13 sOCit–s, where s parameterizes the relation to ownership change at lead/lag s. To avoid model specification bias, each s is unconstrained and is estimated over the full range of s from –13 to 12. The fitted terms of s provide estimates of the relationship of ownership change to productivity, size, and workforce characteristics in each year. 10.5.3 Avoiding Biases If just the ownership change dummies were included as regressors, the estimates would be subject to sample selection and measurement error biases. Sample selection bias would result because for large positive or negative values of s, the ownership change variable OCit–s equals one only if the plant survived a large number of years (at least –s 1 years for s 0 or at least s 2 years for s 0). Any characteristics of surviving plants, such as higher productivity, would thus be partially attributed to ownership change. Measurement error bias would also result, given that ownership changes are unmeasured when they occur outside the sample time frame. For example, for s –13, OCit–s can equal one only if t 1985 (so t – s 1998); for other values of t information about ownership changes is unavailable (since t – s 1998, the last year of data), causing, by definition, OCit–(–13) 0. Similarly, for s –12, OCit–s can equal 1 only if t 1986; . . . ; for s –1, OCit–s can equal 1 only if t 1997; for s 0, OCit–s can equal 1 only if t 1986; . . . ; for s 12, OCit–s can equal 1 only if t 1998. If observations are evenly dispersed across years and the probability of ownership change remains constant at p over time, the expected value of OCit–s would equal

Ownership Change, Productivity, and Human Capital

413

1/14 p for s –13 (as it is artificially 0 in 13 of 14 years of data), 2/14 p for s –12, . . . , 13/14 p for s –1 or s 0, . . . , 1/14 p for s 12. Thus, values of OCit–s would constitute error-ridden indicators of ownership change, with the error greatest for the largest (absolute) values of s. If these ownership change measures are uncorrelated with each other and with all other regressors, the resulting coeﬃcient estimates would be biased toward zero, with the greatest bias for estimates at large (absolute) values of s. If the true coeﬃcients all equaled the same constant number c, the expected values of the estimates would follow a U-shape (if c 0) or inverted U-shape (if c 0). Hence, both sample selection and measurement biases could confound our analysis of the relationship between ownership change and plant performance. Such biases can be especially severe when researchers use a balanced panel (e.g., Lichtenberg and Siegel 1987), restrict the range of s (McGuckin and Nguyen 1995), or analyze pre- versus post-acquisition periods using a single coeﬃcient for each. For example, the use of a balanced panel imparts a strong selection bias because the analysis is based only on those plants that survived throughout the sample period. Restrictions on the range of s eﬀectively constrain s to equal zero outside of the range, yielding possible specification error. Pre- versus post-acquisition periods eﬀectively constrain s to be identical across values of s and hence constitute an additional source of specification error. Moreover, none of these approaches entirely gets rid of the sample selection and measurement biases pointed out previously unless all data points are dropped from analysis if they are within L 1 years of the start and L years of the end of the sample and the range of s is constrained to –L s L. There is a simple way to address this problem without excluding any observations. The intended comparison is between plants that experienced ownership change in year t – s and those that could have but did not experience ownership change in year t – s (not between plants that did experience, versus those that might have or could not have experienced, ownership change). For each s, we divide the observations into three types of establishments: (a) plants that did experience ownership change in year t – s, (b) plants that could have but did not experience ownership change in year t – s, and (c) plants that did not exist or those for which it is unknown whether they experienced ownership change in year t – s. To ensure that the coeﬃcients s describe the diﬀerence between categories (a) and (b), it is suﬃcient to introduce into the model a dummy variable NDit–s that equals 1 for any observations meeting condition (c) in year t – s and 0 for all other observations. This gives rise to one additional variable for each s, yielding the sum Σ12 s–131sNDit–s, comparable to the ownership change term in the models. NDit–s 1 implies either no data about whether ownership change occurred in year t – s, or nonexistence of the plant in year t – s. These controls remove a potentially important source of bias in the estimates.

414

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

This removal of bias may seem complex. However, it is merely a simple application of the use of dummy variables in regression models. A researcher with data on the earnings of Asian, black, and white workers could compare the mean earnings of Asians and blacks by regressing earnings on 0 – 1 dummies for black and white workers and then examining the estimated coeﬃcient on the black dummy. Were the researcher to omit the white dummy variable and include only the black dummy variable, the estimated coeﬃcient of black would no longer yield an unbiased estimate of the black-minus-Asian earnings diﬀerence (it would be unbiased only if Asians and whites had equal expected earnings). Letting A, B, and W denote the mean earnings of these three groups, the estimated coeﬃcient of black converges to B – A in the first regression but B – [hA (1 – h)W ] in the second regression, where h is the proportion of Asians among all Asians plus whites in the sampled population. The use of dummy variables is analogous here: for each s replace Asians with No Change in Owner s years ago, replace blacks with Change in Owner s years ago, and replace whites with Don’t Know whether Ownership Change Occurred s years ago, and let A, B, and W denote for these three groups the mean of the dependent variable after subtracting the eﬀects of all other independent variables. The dummy variables used for the latter two groups are OCit–s and NDit–s. Were a researcher to omit NDit–s from a regression, the estimated coeﬃcient of OCit–s would converge not to the diﬀerence between firms with versus without ownership change, but to B – [hA (1 – h)W ], where h is the proportion of No Change in Owner s years ago cases among the total of No Change in Owner s years ago cases plus Don’t Know whether Ownership Change Occurred s years ago cases in the sampled population. Indeed, in simulations we have found substantial bias without the NDit–s controls, but no bias once they are introduced. To reduce another possible bias, caused by cross-industry, cross-year, or cross-plant-age diﬀerences in both the probability of acquisition and the dependent variable (or productivity), additional controls are used. Fixed eﬀect dummies are included in all analyses for each year, four-digit industry (according to 1969 Swedish SICs), and plant age.7 In addition, production function parameters are each allowed to diﬀer by industry, eﬀectively, by including interaction terms that equal industry-specific dummies (Ikit 1 if plant i’s primary industry is equal to k or Ikit 0 otherwise) times each 7. Industries must be defined according to 1969 industries because only in the later years of the sample have plants been classified according to more recent industry definitions. Another limitation of the data is that they do not include plant ages, so plants are classified according to their minimum age (1, 2, . . .) if they existed in 1985 or their actual age if they entered after 1985. Fortunately, the employee-level data file indicated (for nearly all plants) whether each plant existed in each year, even if it was not present in the plant-level data; these additional data therefore allowed identification of plant age without sample selection in years 1985 to 1998.

Ownership Change, Productivity, and Human Capital

415

production function parameter.8 Use of these controls implies that the relations of ownership change are studied largely for plants of comparable industry and age at a comparable date. 11.5.4 Endogeneity and Descriptive Estimation It is common in the estimation of production functions to use instrumental variables to ensure consistent parameter estimates despite possible endogeneity. In our context, ownership change may be endogenous, since some theories conjecture that the sale of a plant or firm may be more likely when it has relatively low productivity. Just as instrumental variables methods can be used to estimate a supply or demand equation rather than a mixture of the two, such techniques could be used in an analysis of the consequences of ownership change. In this chapter, however, our aim is not to estimate either causes or consequences, but merely to describe patterns experienced on average before and after plants undergo ownership change. We focus on describing trends in several key variables before and after ownership change.9 10.6 Empirical Results 10.6.1 Descriptive Statistics for Plant-Level Variables Table 11.7 contains descriptive statistics for key variables used in the econometric analysis, presented separately for plants that experienced an ownership change and for those that did not. Plants involved in these transactions tend to be larger by about 60 percent.10 On average, they employed slightly smaller percentages of female workers and workers with at least a college education, and slightly more non-Swedish employees than plants that did not experience an ownership change. 11.6.2 Productivity, Output, and Employment as Related to Ownership Change We now consider changes in productivity and plant size associated with ownership change. Table 11.8 presents OLS estimates of four equations: labor productivity, total factor productivity, output, and employment. The equations we estimate are the following. 8. We also experimented with including detailed geographic region dummies in the TFP equation, and found that their inclusion had almost no eﬀect on our results. 9. The focus on observed empirical patterns matches the descriptive focus of the NBER conference for which this paper was developed, and accords with the wishes of the conference organizers. 10. The 60 percent greater size of plants experiencing ownership change than those not experiencing ownership change can be computed as exp(9.98 – 9.51) 1.6.

416

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

Table 11.7

Means and standard deviations of production function variables and worker characteristics

Variable Log gross output Log real value of plant and machinery capital stock Log plant employment Log materials Average age of employees Percentage of female employees Percentage of non-Swedish employees Percentage of employees with at least a college education Log earnings

All plants

Plants that experience an ownership change

Plants that do not experience an ownership change

9.68 (1.52) 9.21 (1.49) 3.36 (1.21) 8.80 (1.82) 39.49 (5.36) 25.77 (21.64) 9.24 (11.49) 2.81 (7.11) 11.86 (0.37)

9.98 (1.55) 9.60 (1.41) 3.67 (1.22) 9.18 (1.75) 39.57 (4.95) 25.18 (20.61) 9.62 (11.17) 2.45 (5.85) 11.85 (0.36)

9.51 (1.51) 9.01 (1.48) 3.18 (1.17) 8.59 (1.82) 39.45 (5.56) 26.08 (22.17) 9.04 (11.66) 3.00 (7.68) 11.86 (0.38)

Note: Standard deviations are in parentheses.

Labor Productivity: 12

12

s13

s13

(14) ln(Qijt) jt 1j ln(Lijt) ∑ sOCijts ∑ sNDijts Age Dummies Industry Dummies Time Dummies εijt Total Factor Productivity: (15)

ln(Qijt) jt 1j ln(Lijt) 2j ln(Kijt) 3j ln(Mijt) 12

12

s13

s13

∑ sOCijts ∑ sNDijts Age Dummies Industry Dummies Time Dummies εijt Output or Employment: (16)

12

12

s13

s13

yijt jt ∑ sOCijts ∑ sNDijts Age Dummies Industry Dummies Time Dummies εijt

Table 11.8

Coeﬃcient on: Labora Capitala Materialsa OCt13 OCt12 OCt11 OCt10 OCt9 OCt8 OCt7 OCt6 OCt5 OCt4 OCt3 OCt2 OCt1 OCt0 OCt–1 OCt–2 OCt–3 OCt–4 OCt–5 OCt–6

Parameter estimates from labor productivity, total factor productivity, output, and employment regressions for all ownership changes Labor productivity 1.054∗∗∗∗ (0.014) — — .347∗∗∗∗ — –0.52 (.035) –.042∗ (.025) –.010 (.020) –.027∗ (.017) –.036∗∗ (.015) –.035∗∗∗∗ (.013) –.039∗∗∗∗ (.011) –.041∗∗∗∗ (.010) –.041∗∗∗∗ (.010) –.041∗∗∗∗ (.009) –.042∗∗∗∗ (.009) –.035∗∗∗∗ (.008) –.050∗∗∗∗ (.009) –.022∗∗∗ (.008) –.009 (.008) –.004 (.009) .002 (.010) .002 (.010) .005 (.011) .019 (.012)

TFP (Cobb-Douglas) .375∗∗∗∗ (.014) .270∗∗∗∗ (.018) (.014) — — — — .009 (.018) .012 (.011) .011 (.011) –.018∗∗ (.008) –.031∗∗∗∗ (.008) –.028∗∗∗∗ (.006) –.039∗∗∗∗ (.006) –.036∗∗∗∗ (.005) –.056∗∗∗∗ (.006) –.066∗∗∗∗ (.006) –.032∗∗∗∗ (.005) –.024∗∗∗∗ (.005) –.013∗∗ (.006) –.020∗∗∗ (.006) –.014∗∗ (.007) –.006 (.007)

Output

Employment

—

—

—

—

— .092 (.093) .020 (.060) .049 (.045) .060 (.039) .114∗∗∗∗ (.035) .136∗∗∗∗ (.029) .162∗∗∗∗ (.026) .186∗∗∗∗ (.023) .205∗∗∗∗ (.022) .179∗∗∗∗ (.020) .178∗∗∗∗ (.019) .179∗∗∗∗ (.018) .153∗∗∗∗ (.018) –.001 (.017) .082∗∗∗∗ (.018) .092∗∗∗∗ (.019) .101∗∗∗∗ (.021) .095∗∗∗∗ (.023) .097∗∗∗∗ (.025) .142∗∗∗∗ (.028)

— .162∗ (.085) .058 (.052) .055 (.040) .083∗∗ (.035) .142∗∗∗∗ (.032) .163∗∗∗∗ (.026) .191∗∗∗∗ (.023) .216∗∗∗∗ (.020) .232∗∗∗∗ (.019) .208∗∗∗∗ (.017) .208∗∗∗∗ (.016) .202∗∗∗∗ (.015) .195∗∗∗∗ (.015) .022∗∗∗∗ (.015) .086∗∗∗∗ (.015) .087∗∗∗∗ (.016) .092∗∗∗∗ (.018) .088∗∗∗∗ (.019) .090∗∗∗∗ (.021) .120∗∗∗∗ (.024) (continued )

418

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

Table 11.8

(continued)

Coeﬃcient on:

Labor productivity

TFP (Cobb-Douglas)

Output

R2

.043∗∗∗ (.014) .030∗ (.013) .037∗∗ (.018) .040∗ (.022) .061∗∗ (.030) .024 (.054) Yes Yes Yes Yes 6.150∗∗∗∗ (0.053) 0.859

.017∗∗ (.008) .009 (.011) .021∗ (.012) .024∗ (.014) .046∗∗ (.021) .028 (.038) Yes Yes Yes Yes 2.084∗∗∗∗ (.143) 0.960

.187∗∗∗∗ (.033) .159∗∗∗∗ (.037) .139∗∗∗∗ (.041) .175∗∗∗∗ (.050) .201∗∗∗∗ (.069) .171∗∗∗∗ (.108) Yes Yes Yes Yes 10.610∗∗∗∗ (.059) 0.358

Number of plants Number of observations

18,495 124,381

15,946 82,307

18,513 124,441

OCt–7 OCt–8 OCt–9 OCt–10 OCt–11 OCt–12 “No-data” dummies Industry dummies Year dummies Age dummies Intercept

Employment .140∗∗∗∗ (.027) .119∗∗∗∗ (.030) .096∗∗∗∗ (.034) .124∗∗∗∗ (.041) .121∗∗∗ (.057) .104 (.091) Yes Yes Yes Yes 4.049∗∗∗∗ (.049) 0.301 18,962 125,416

Note: Standard errors in parentheses. Significance levels are two-tailed using robust standard errors, allowing for correlated (“clustered”) errors within plants. a Weighted means of industry-specific coeﬃcients at the detailed (four-digit SIC) industry level. ∗∗∗∗ Indicates significance at the 0.1 percent level. ∗∗∗ Indicates significance at the 1 percent level. ∗∗ Indicates significance at the 5 percent level. ∗ Indicates significance at the 10 percent level.

where Lijt, Kijt, and Mijt are labor, capital, and materials for plant i in industry j at year t, OCijt–s and NDijt–s are the ownership change and no-data dummy variables described earlier, and yijt is output or employment for plant i in industry j at year t.11 Recall that each regression is estimated with detailed industry level (four-digit SIC) fixed eﬀects. Thus, the coeﬃcients on the nonownership change variables (labor, capital, and materials) are weighted means of industry-specific coeﬃcients. The coeﬃcients on labor, capital, and materials in the two productivity 11. The equations deliberately do not include plant fixed eﬀects, only industry fixed eﬀects, because including plant fixed eﬀects would make it impossible to observe whether plants that experience ownership change tend to have persistently low or high productivity or, indeed, to know how these plants compare to their industry (and age and time) averages at all—all patterns that are important to be able to detect.

Ownership Change, Productivity, and Human Capital

419

models appear to be plausible. They are reasonably close to their respective factor shares and strongly suggestive of constant returns to scale. The total factor productivity (TFP) equation is estimated for a restricted sample of plants and years, because the capital measure is only available from 1989 onward and, in those years, for 92.3 percent of plants. Next, we focus our attention on the coeﬃcients on the ownership change dummies in table 11.8. For example, the value –.042 for the estimated coeﬃcient of OC–1 in the labor productivity equation signifies that plants experiencing an ownership change one year hence were 4.2 percent less productive, on average, than comparable establishments that did not change owners. Note that while the relative performance of plants changing owners was significantly worse before the transaction, relative eﬃciency appears to have improved after the ownership change, as these establishments converged to the average level of industry performance. The LP estimates indicate lower than average productivity before ownership change, followed by a rapid increase to the industry norm, and ultimately, higher productivity following the ownership change. The TFP estimates indicate productivity steadily deteriorating to a low of nearly 6 percent below average, followed by a steady return to average and higher productivity after the ownership change. The decline in TFP, but not LP, in the five years preceding ownership change implies that while labor was used ineﬃciently in these plants, it was in the application of equipment and materials that they experienced a gradual eﬃciency decline.12 The output and employment results, which are presented in the last two columns of table 11.8, help explain the productivity increase. Plants that changed owners apparently had higher output and employment than comparable plants both before and after ownership change. They reduced both output and employment after an ownership change.13 However, employment declined at a faster rate than output, which resulted in a productivity increase. In table 11.9, we present averages of the coeﬃcients on the ownership change dummies in the LP, TFP, output, and employment equations for five years before and five years after the transaction (we exclude year zero, which is the year of the acquisition). In the third row of each panel, we 12. The TFP results are based on a subsample of observations, since capital stock measures are unavailable before 1989. However, this is not the source of the diﬀerence between the LP and TFP results. Estimating the LP equation using the same subsample of observations as for the TFP regression yields nearly identical results to those reported for LP in the chapter. 13. Although the change in output could result either from a decision on the part of the new owners or a firm-specific shock that triggers ownership change, there is some evidence that the output reductions may in fact result from the decisions of the new management. First, similar declines in output do not coincide with the gradual reduction in productivity that precedes ownership change. Second, the same decline in output and employment are apparent after adding industry-year interactions to the model using four-digit SIC codes, thereby controlling for possible industry-specific demand shocks. This leaves open the possibility of other shocks that lead to ownership change and also to downsizing.

Table 11.9

Estimated eﬀects of ownership change on labor productivity (LP), total factor productivity (TFP), output, and employment for various types of ownership changes

Period

Pre-ownership change Post-ownership change Post-pre Pre-ownership change Post-ownership change Post-pre Pre-ownership change Post-ownership change Post-pre

LP

TFP

All ownership changes –0.042∗∗∗∗ –0.038∗∗∗ –0.001 –0.021∗∗∗ 0.041∗∗∗∗ 0.017∗∗∗ Full acquisitions –0.055∗∗∗∗ –0.038∗∗∗∗ –0.007 –0.024∗∗∗∗ 0.047∗∗∗∗ 0.014∗∗ 0.011 0.019 0.008

Partial acquisitions –0.037∗∗ –0.011 0.026∗

Output

Employment

0.179∗∗∗∗ 0.093∗∗∗∗ –0.086∗∗∗∗

0.209∗∗∗∗ 0.089∗∗∗∗ –0.120∗∗∗∗

0.061∗∗∗ –0.055∗∗∗ –0.116∗∗∗∗

0.111∗∗∗∗ –0.046∗∗∗ –0.156∗∗∗∗

0.633∗∗∗∗ 0.556∗∗∗∗ –0.076∗∗

0.584∗∗∗∗ 0.510∗∗∗∗ –0.073∗∗

0.092∗∗∗∗ –0.012 –0.104∗∗∗∗

0.137∗∗∗∗ –0.007 –0.144∗∗∗∗

Pre-ownership change Post-ownership change Post-pre

Full divestitures –0.052∗∗∗∗ –0.039∗∗∗∗ –0.005 –0.022∗∗∗∗ 0.047∗∗∗∗ 0.017∗∗∗

Pre-ownership change Post-ownership change Post-pre

0.019 0.018 –0.001

Partial divestitures –0.030∗ –0.014 0.015∗

0.684∗∗∗∗ 0.594∗∗∗∗ –0.090∗∗

0.628∗∗∗∗ 0.551∗∗∗∗ –0.077∗∗

Pre-ownership change Post-ownership change Post-pre

0.005 0.021 0.015

Related acquisitions –0.043∗∗∗∗ –0.009 0.034∗∗

0.429∗∗∗∗ 0.363∗∗∗∗ –0.066∗

0.402∗∗∗∗ 0.324∗∗∗∗ –0.079∗∗

Unrelated acquisitions –0.013 –0.034∗∗ 0.020 –0.010 0.033∗∗ 0.024

0.345∗∗∗∗ 0.234∗∗∗∗ –0.111∗

0.346∗∗∗∗ 0.208∗∗∗∗ –0.138∗∗∗

Change in ownership involving a single firm –0.059∗∗∗∗ –0.037∗∗∗ 0.087∗∗∗ –0.007 –0.024∗∗∗ 0.014 0.052∗∗∗∗ 0.013∗ –0.075∗∗∗∗

0.137∗∗∗∗ 0.020 –0.118∗∗∗∗

Pre-ownership change Post-ownership change Post-pre Pre-ownership change Post-ownership change Post-pre

Notes: Pre-ownership change is average 5 years before; post-ownership change is average 5 years after. Significance levels are two-tailed using robust standard errors, allowing for correlated (“clustered”) errors within plants. ∗∗∗∗ Indicates significance at the 0.1 percent level. ∗∗∗ Indicates significance at the 1 percent level. ∗∗ Indicates significance at the 5 percent level. ∗ Indicates significance at the 10 percent level.

Ownership Change, Productivity, and Human Capital

421

report the growth in the average coeﬃcient from the five years before to the five years after ownership change, and we formally test whether the postversus pre-ownership change eﬀects are statistically significant. Our findings are also presented separately in table 11.9 for various types of ownership change: full acquisitions and partial acquisitions, full divestitures and partial divestitures, related acquisitions, unrelated acquisitions, and changes in ownership involving a single firm. The “post-pre” results in the first panel of table 11.9 confirm our earlier finding that plants involved in an ownership change became more productive after the transaction. From the five years before to the five years after ownership change, LP is estimated to have increased by 4.1 percent ( p .001) and TFP by 1.7 percent ( p .01). We also find that output and employment were reduced after ownership change, with employment declining more than output. Output is estimated to have declined by 8.6 percent ( p .001) from the pre- to post-ownership change five-year periods, and employment is estimated to have decreased by 12.0 percent ( p .001). The growth in LP is estimated to have been much higher for full acquisitions and divestitures than for partial acquisitions and divestitures (panels 2 through 5 of table 11.9). In full-firm acquisitions and divestitures both, labor productivity grew an estimated 4.7 percent between the two five-year periods (both p < .001). In contrast, partial acquisitions were associated with only 0.8 percent growth in LP, while partial divestitures were associated with a decrease of 0.1 percent in LP (both changes are insignificantly diﬀerent from zero). The diﬀerence may stem from the fact that plants acquired through partial acquisition and divestiture had higher labor productivity to begin with, 1.1 percent above the norm for partial acquisitions and 1.9 percent above the norm for partial divestitures, versus labor productivity averaging 5.5 percent below the industry norm for full acquisitions and 5.2 percent below the industry norm for partial divestitures.14 Growth in TFP was much more similar across full versus partial acquisitions and divestitures. Partial acquisitions are estimated to have experienced slightly higher (2.6 percent) TFP growth between the two five-year periods than either full acquisitions (1.4 percent) or any type of divestiture (1.5 percent to 1.7 percent), but the diﬀerence is not statistically significant. All types of acquisitions and divestitures involved plants whose TFP was about 3.0 percent to 3.9 percent below the norm before ownership change. Labor Productivity (LP) grew more in the aftermath of unrelated acquisitions, as opposed to related acquisitions, and even more in ownership 14. These diﬀerences pre-ownership change might stem from higher labor productivity in larger plants or from cherry-picking by acquiring firms that purchase only some of a firm’s plants.

422

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

changes that did not involve a second (manufacturing) firm. Among these three types of ownership change, the increase in labor productivity was negatively related to initial productivity: single-firm ownership changes increased their LP an estimated 5.2 percent ( p .001) from 5.9 percent below the norm to 0.7 percent below the norm, while unrelated acquisitions saw LP increase an estimated 3.3 percent ( p .05) from 1.3 percent below the norm to 2.0 percent above the norm, and related acquisitions saw LP increase only an estimated 1.5 percent (statistically insignificant) from 0.5 percent above the norm to 2.1 percent above the norm. Total Factor Productivity (TFP) growth was substantial and significant, but was greatest for related acquisitions, with 3.4 percent growth ( p .05) from an initial base 4.3 percent below the norm, whereas the unrelated and single-firm ownership changes, respectively, experienced only 2.4 percent (insignificant) and 1.3 percent ( p .10) TFP growth from initial bases 3.4 percent and 3.7 percent below the norm. The finding that unrelated acquisitions enhanced plant productivity is consistent with U.S. evidence presented in Maksimovic and Phillips (2001) and Schoar (2002). The decline in output, and especially employment, was greatest in the aftermath of full acquisitions and divestitures, for which output declined 11.6 percent ( p .001) and 10.4 percent ( p .001), respectively, and employment declined 15.6 percent ( p .001) and 14.4 percent ( p .001) respectively. In contrast, partial acquisitions and divestitures respectively experienced an estimated 7.6 percent ( p .05) and 9.0 percent ( p .05) decline in output and 7.3 percent ( p .05) and 7.7 percent ( p .05) decline in employment. Partial acquisitions and divestitures tended to involve plants that were substantially larger to begin with, starting larger than the norm by an estimated 63.3 percent and 68.4 percent, respectively, versus only 6.1 percent and 9.2 percent for full acquisitions and divestitures. The declines in output and employment are estimated to have been greater for unrelated acquisitions than for related acquisitions, but there is no statistically significant diﬀerence between the related and unrelated (and singlefirm) acquisitions. Both the annual coeﬃcient estimates and the five-year means for all ownership changes combined are shown graphically in figures 11.1 through 11.4. These graphs make it easy to visualize the relation of the four variables to ownership change, and moreover, clarify annual patterns that are not evident in the five-year means. The horizontal axis in each graph spans a fifteen-year period, from seven years before ownership change to seven years after ownership change. The vertical axis corresponds to the values of the estimated coeﬃcients, and hence to the relation of ownership change to productivity, output, or employment at a given time relative to the year of ownership change. The curve drawn across the diagram shows the annually changing values of productivity relative to the industry (and age and year) norm. For each coeﬃcient estimate, its 95 percent confidence

Fig. 11.1 Graphs of the coeﬃcients on the ownership change dummies in the LP equation

Fig. 11.2 Graphs of the coeﬃcients on the ownership change dummies in the TFP equation

Fig. 11.3 equation

Graph of the coeﬃcients on the ownership change dummies in the output

Fig. 11.4 Graph of the coeﬃcients on the ownership change dummies in the employment equation

Ownership Change, Productivity, and Human Capital

425

band clarifies the range of error in the estimates. Dashed lines in the five years pre- and post-ownership change indicate 95 percent confidence bands for means before and after ownership change, and a 95 percent band drawn at year zero pertains to the change between the five-year periods pre- to post-ownership change. The graphs demonstrate that five-year averages can mask important dynamics related to the time of ownership change. For labor productivity, the five-year means provide an accurate summary, but for TFP, they hide a substantial and statistically significant decline in productivity up to the time of ownership change, followed by substantial growth in productivity that begins immediately after the year of ownership change. The pattern indeed looks as if typical plants had been losing TFP relative to the norm at a pace of about 1 percent per year before ownership change, with the new owners apparently managing to enhance productivity by about 4 percent within one year after ownership change and continuing to enhance productivity by about 0.5 percent per year thereafter. The graphs for real output and employment also indicate substantial disruptions in the year of ownership change, with plants formerly about 15 percent above the norm in output and 20 percent above the norm in employment suddenly falling to levels near their industry (and age and year) means. Under the new ownership, output (especially) and employment (somewhat) then grew immediately in the year following ownership change, with very slow increases in subsequent years. 11.6.3 Labor Force Characteristics as Related to Ownership Change In table 11.10, we present similar results for six labor-related dependent variables: the average age of employees at the plant, average experience, the percentage of female employees, the percentage of non-Swedish employees,

Table 11.10

Period

Estimated eﬀects of ownership change on age, experience, % female, % non-Swedish, % college-educated, and earnings

Age

Pre-ownership change 0.053 Post-ownership change 0.213∗∗∗∗ Post-pre 0.160∗∗

Experience

% Female

0.046 0.213∗∗∗∗ 0.167∗∗∗

0.766∗∗∗∗ 0.117 –0.649∗∗∗

% % CollegeNon-Swedish educated 0.360∗∗ 0.245 –0.114

–0.051 0.126 0.177∗∗

Earnings –0.009∗∗∗∗ 0.004∗ 0.013∗∗∗∗

Notes: Pre-ownership change is average 5 years before; post-ownership change is average 5 years after. Significance levels are two-tailed using robust standard errors, allowing for correlated (“clustered”) errors within plants. ∗∗∗∗ Indicates significance at the 0.1 percent level. ∗∗∗ Indicates significance at the 1 percent level. ∗∗ Indicates significance at the 5 percent level. ∗ Indicates significance at the 10 percent level.

426

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

the percentage of college-educated workers, and earnings. As before, the table includes estimates for the various types of ownership change. Figures 11.5 through 11.10 present graphical representations of the annual coeﬃcient estimates for all ownership changes. The findings in table 11.10 imply that plants involved in ownership change experienced estimated increases in average employee age by 0.16 year or about two months of age ( p .05), in experience by 0.17 year ( p .01), and in the percentage of employees with a college education by (an absolute amount of) 0.18 percent ( p .05). The age and experience results suggest some tendency for newer workers to be laid oﬀ or leave more often than older workers. The education result suggests that ownership change led to a reduction in the demand for less-educated workers. We also find that ownership change resulted in an increase in employees’ mean earnings (as always, relative to the industry and plant age and year norm) by 1.3 percent ( p .001) and a decline in the percentage of female workers by (an absolute amount of) 0.65 percent ( p .01). The increase in earnings is consistent with more experienced employees remaining, while newer workers left, since (as the employee-level data confirm) the older and more experienced workers received higher earnings. The decline in the percentage of female workers might have been related to women workers often having shorter

Fig. 11.5 Graph of the coeﬃcients on the ownership change dummies in the mean employee age equation

Fig. 11.6 Graph of the coeﬃcients on the ownership change dummies in the mean employee experience equation

Fig. 11.7 Graph of the coeﬃcients on the ownership change dummies in the percentage female workers equation

Fig. 11.8 Graph of the coeﬃcients on the ownership change dummies in the percentage of non-Swedish workers equation

Fig. 11.9 Graph of the coeﬃcients on the ownership change dummies in the percentage of college-educated workers equation

Ownership Change, Productivity, and Human Capital

429

Fig. 11.10 Graph of the coeﬃcients on the ownership change dummies in the mean earnings equation

job durations than men, and hence being more likely to lose a job because of short job tenure and experience. The annual estimated coeﬃcients in figures 11.5 through 11.10 reaﬃrm the previous conclusions, but suggest that at least some of the estimated changes were gradual processes. In particular, mean employee age and experience increased gradually, and the percentage of female workers declined gradually over a period of several years following ownership change. There are multiple possible interpretations of these patterns, but one interpretation involves employees not fully losing their connection with a plant even if they were laid oﬀ temporarily during the year of ownership change, and the new owners often gradually shifting to a workforce that suits their demands. Two additional stylized facts emerge from these figures. The percentage of college-educated workers may actually have increased in the year before ownership change, although the ranges of error involved leave some uncertainty in this conclusion. Also, earnings plummeted (significantly) relative to the norm in the year preceding ownership change. This decline in earnings is explained by a reduction in hours worked per employee, which fell below the norm by 3.2 percent in the year preceding ownership change

430

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

and 1.3 percent in the year of ownership change, but in other years was close to the norm in plants that experienced ownership change.15 11.6.4 Where Pre-Ownership Change Employees Went Next, we make use of the individual-level data, in order to track the movement and relative compensation of workers whose establishments were involved in an ownership change. In doing so, we attempt to answer two questions: (a) where did the old employees go? and (b) where did the new employees come from? Table 11.11 presents statistics on the destination (if any) of employees whose plants experienced ownership change. The top panel of table 11.11 pertains to all employees who left their plant, while the second panel pertains to all employees. Each panel indicates the percentage of employees who had the following characteristics: female, non-Swedish born, age above the mean of all manufacturing employees, experience above the mean of all manufacturing employees, and three categories of educational achievement (less than high school, at least some high school, and at least some college-level study). The bottom panels of the table show the corresponding sample sizes for employees who left their plant and for all employees.16 The last column is the base case, pertaining to employees whose plants did not change owner in the subsequent year. In this and other columns in the table, observations in which ownership change occurred in the previous or following year are excluded to avoid contaminating data in nearby years. Also, the observations considered are all employee-year combinations for whom their plant meets the required categorization regarding the type of ownership change it was about to experience. Table 11.11 confirms the findings of the plant-level analyses, concerning which types of employees were most likely to leave in the wake of an ownership change. For example, in plants experiencing ownership change, 27.45 percent of employees were female and 14.23 percent were nonSwedish born. Because a slightly higher prevalence of females, 28.07 percent, and non-Swedish born, 15.28 percent, occurred among leaving employees, the percentages of female and non-Swedish born employees was driven down slightly, consistent with the slight decreases observed in the preceding regressions. Similarly, the table shows a disproportionately low percentage of older and more experienced employees leaving plants that experienced ownership change. 15. This finding stems from a regression, like those presented in the chapter, in which the dependent variable is the logarithm of the average number of hours worked per employee at the plant. In other years relative to ownership change, the coeﬃcients of the ownership change dummies in this regression are near zero and are statistically insignificant at the .05 level. 16. Because the years of experience and education level variables are not available for some employees, sample sizes are also reported for the number of employees for whom values of these variables are available.

Ownership Change, Productivity, and Human Capital Table 11.11

431

Employees leaving plants in the aftermath of an ownership change: Statistics for workers at year T – 1 whose plants experienced an ownership change during year T

Employee type Employees leaving plants: % Female % Non-Swedish % Age above mean % Experience above mean % Education less than high school % Education to high school % Education some college All employees: % Female % Non-Swedish % Age above mean % Experience above mean % Education less than high school % Education to high school % Education some college Sample size for employees leaving plant: All With experience data With education data Sample size for all employees: All With experience data With education data

All ownership changes

No ownership change

28.07 15.28 45.35 40.04 38.75 47.82 13.43

28.50 14.12 39.15 36.45 37.30 48.91 13.79

27.45 14.23 49.12 49.72 41.00 47.35 11.64

26.38 12.60 47.81 50.98 39.40 48.27 12.33

131,495 52,952 125,363

1,782,949 802,542 1,699,355

352,094 135,380 338,494

6,586,368 2,690,062 6,358,300

The individual-level results qualify the plant-level results presented earlier. Table 11.11 reveals that even among plants that did not experience ownership, relatively high percentages of women and non-Swedish born employees left the plants. Thus, while ownership change may have resulted in substantial job loss for these workers, it appears as though they were not treated more unfairly in plants experiencing ownership change than in representative plants (in fact, the evidence suggests they were treated slightly more fairly, perhaps because of diﬀering job roles).17 In tables 11.12 and 11.13, we follow workers at the end of year T – 1 and measure their employment status and earnings growth, respectively, at the end of year T 1, cross-classified by a set of dummy variables denoting whether the plant that employed them during year T – 1 experienced an 17. Some care must be taken in comparing findings at the plant and employee levels not only because of the complexity of deciphering the flows of employees, but also because the plant-level analyses control for industry, plant age, and year eﬀects while the employee-level results simply present outcomes for average employees.

432

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

ownership change during year T.18 In each table, the top panel pertains to employees whose plants experienced ownership change, while the bottom panel pertains to the comparator case of no ownership change. Within each panel, rows pertain to employees’ future job status—workers could be employed at the same plant, at another plant owned by the previous owner, at another plant owned by the new owner, at another firm in the same industry, at another firm in a diﬀerent industry, in an unknown industry or plant (which likely includes workers who become self-employed or who are employed at an entrepreneurial startup), or they could be unemployed. The findings in table 11.12 are consistent with the plant-level results cited earlier, in the sense that ownership change appears to have been associated with an increase in worker turnover at plants and firms. For example, only 62.7 percent of the workers observed at the end of year T – 1 whose plants changed owners during year T were still employed at the same establishment at the end of year T 1. Turnover findings are also presented separately for diﬀerent types of workers. Not surprisingly, these results imply that females, non-Swedes, younger employees, and less experienced workers were less likely than representative workers to remain at the same plant in the aftermath of an ownership change. We also find that workers with the highest levels of education had the greatest mobility across firms. Two-year mean earnings growth for the same groups of workers is presented in table 11.13. A potentially interesting finding pertains to women’s salary growth compared to the norm. Regardless whether ownership change occurred, females who remained in the same establishment had higher average earnings growth than males, presumably because they switched more frequently from part-year to full-year (or part-time to fulltime) employment. The potentially interesting pattern is that women who remained working in an establishment involved in an ownership change experienced less earnings growth relative to the norm of no ownership change (in both absolute and percentage terms) than men. It is important to note, however, that we do not have individual-specific information on hours worked. Thus, one explanation for this finding, which cannot be ruled out on the basis of our empirical analysis, is that women (who may relatively often have worked part-time) may have worked more hours in the aftermath of an ownership change. 18. The focus on years T – 1 and T 1, rather than times separated by only one year, is necessitated by the timing of when ownership changes occur and when employee information is reported. Recall that employee information pertains to November, while ownership change can occur at any time during the reporting year. If years T – 1 and T were used, it would be possible that ownership change could have occurred after the employee data were received in year T (not to mention that new owners’ policies may take some time to come into eﬀect). If years T and T 1 were used, the employee’s initial status normally would be recorded after ownership change occurred rather than before.

Same plant Other plant owned by same firm Other plant owned by acquiring firm Same industry, other firm Other manufacturing industry, other firm Other nonmanufacturing industry, other firm Unknown industry or plant (could include self-employed and start-ups) Not employed in Sweden

Swedish Non-Swedish

Young Old

5.7 2.0 18.8

6.0 2.6 13.9

5.9

2.5 15.2

3.6

4.6 10.4

9.1

4.9 1.4 16.2

4.5 4.9 1.5 11.4

4.1

4.9

1.4 12.6

3.1

73.7 2.2 0.0 1.9

72.9 2.2 0.0 1.8

1.5 11.9

5.1

4.2

1.3 17.7

3.4

3.7

1.6 12.7

7.1

5.9

1.2 12.6

2.4

2.2

1.7 16.1

9.1

7.4

61.3 2.1 0.0 2.2

2.6 14.5

8.1

7.8

Plants not experiencing an ownership change 68.4 70.8 77.8 73.4 69.7 2.2 2.1 2.2 2.2 2.3 0.0 0.0 0.0 0.0 0.0 2.1 1.6 1.5 1.8 1.9

2.3 20.2

4.2

6.0

2.5 17.7

2.5 14.4

6.2

6.2

53.4 1.7 2.3 2.9

Less experience

2.3 16.0

4.4

6.9

6.2

Plants Experiencing an Ownership Change 59.9 61.8 65.5 63.1 59.9 2.0 1.9 2.7 2.3 2.4 2.3 3.1 2.6 2.5 2.2 3.0 2.4 2.6 2.8 2.8

Female

63.0 2.4 2.2 3.0

Male

62.7 2.3 2.4 2.8

All employees

1.2 7.6

4.3

3.7

78.7 2.6 0.0 1.8

1.9 9.9

6.0

5.8

68.5 2.0 3.3 2.6

More experience

1.4 14.3

3.4

3.0

74.7 1.7 0.0 1.5

2.4 17.2

4.0

4.7

65.0 2.4 2.0 2.4

Less than high school

1.4 11.7

5.4

4.5

72.9 2.2 0.0 1.8

2.5 14.2

6.5

6.9

62.6 2.2 2.3 2.9

High school

1.4 9.3

7.2

5.5

70.1 3.9 0.0 2.7

2.4 10.8

10.1

8.0

57.3 2.6 5.0 3.8

At least some college

Where employees go: Employment status at the end of year T 1 (in %) of workers at year T – 1 whose plants experienced an ownership change during year T

Same plant Other plant owned by same firm Other plant owned by acquiring firm Same industry, other firm Other manufacturing industry, other firm Other nonmanufacturing industry, other firm Unknown industry or plant (could include self-employed and start-ups) Not employed in Sweden

Employment Status at the end of year T 1

Table 11.12

Note: n.a. not available.

Same plant Other plant owned by same firm Other plant owned by acquiring firm Same industry, other firm Other manufacturing industry, other firm Other nonmanufacturing industry, other firm Unknown industry or plant (could include self-employed and start-ups)

1.585 1.146

1.293 1.018

1.369

1.046

Non-Swedish

Young Old

0.988

1.088 1.528

1.627

1.879 1.211

1.551 1.390 1.115

1.628

1.519

1.139

1.946

1.073 1.093 n.a. 1.202

1.126 1.168 n.a. 1.256

1.147

1.522

1.624

1.075

1.484

1.653

1.279

1.648

1.768

0.940

1.110

1.208

1.365

1.685

1.847

1.321 1.359 n.a. 1.472

1.216

1.533

1.593

1.375

0.992

1.411

1.606

1.311 1.311 1.285 1.449

Less experience

0.854

1.054

1.365

1.379

Plants experiencing an ownership change 1.078 1.236 0.984 1.111 1.176 0.982 1.068 1.053 1.423 1.009 1.228 1.032 1.317 1.010 1.181 1.142

Swedish

Plants not experiencing an ownership change 1.263 1.281 0.995 1.127 1.119 1.316 1.394 1.006 1.167 1.177 n.a. n.a. n.a. n.a. n.a. 1.407 1.431 1.028 1.261 1.228

1.608

1.363

1.410

1.237 1.121 1.497 1.315

Female

1.058 1.050 1.047 1.133

Male

1.106 1.066 1.203 1.176

All employees

1.097

1.328

1.343

1.158 1.196 n.a. 1.193

0.966

1.218

1.217

1.128 1.058 1.490 1.146

More experience

Mean two-year real earnings growth of employees classified by post-ownership change employment status

Same plant Other plant owned by same firm Other plant owned by acquiring firm Same industry, other firm Other manufacturing industry, other firm Other nonmanufacturing industry, other firm Unknown industry or plant (could include self-employed and start-ups)

Employment Status at the end of year T 1

Table 11.13

1.076

1.595

1.668

1.070 1.115 n.a. 1.224

0.995

1.466

1.486

1.059 1.016 1.030 1.129

Less than high school

1.177

1.495

1.583

1.157 1.179 n.a. 1.265

1.041

1.320

1.342

1.126 1.097 1.351 1.179

High school

1.197

1.456

1.594

1.175 1.207 n.a. 1.260

1.247

1.309

1.337

1.189 1.097 1.183 1.251

At least some college

Ownership Change, Productivity, and Human Capital

435

It is not only women who experienced this relative wage growth disadvantage associated with ownership change. Similarly, earnings growth relative to the norm was lower for non-Swedish born and younger employees, compared to Swedish born and older employees, who remained at the establishment. We also find that more experienced employees had lower earnings growth relative to the norm than did less experienced workers. Employees with at least a college education were the only group of staying employees to experience higher earnings growth with ownership change than without. Again, perhaps surprisingly, employees without a high-school education experienced higher earnings growth relative to the norm than those with a high-school education. These results should be interpreted with caution, since they may be influenced by shifts in work hour patterns. They also reflect diverse percentages of staying employees and are derived from aggregate statistics. Earnings growth was low, relative to the norm, for employees who left an establishment that experienced ownership change. For example, employees who moved to another firm in the same industry had only 17.6 percent average wage growth if they began in ownership change establishments, versus 25.6 percent average wage growth if they began in establishments that did not experience ownership change. The diﬀerence may reflect lower average human capital among leaving employees as well as possible diﬃculties of job changes triggered by ownership change. 11.6.5 New Hires by Plants That Experienced an Ownership Change Table 11.14 contains descriptive statistics on workers at plants that recently experienced an ownership change. Consistent with the format of table 11.11, we present percentages by employee type for employees hired by plants during the previous year (new hires) in the top panel, followed by the same percentages for all workers employed at these establishments in the second panel. Note that these figures once again are presented separately for plants involved in an ownership change and those that did not experience such an event. The evidence in the top panel implies substantial diﬀerences between these two types of plants, in terms of the kind and fraction of workers they hired. These substantial diﬀerences existed despite only small diﬀerences (second panel) in the composition of the workforce between ownership change and no ownership change plants. Plants that recently experienced an ownership change more often hired older, more experienced, and less educated workers. Also, these findings, in conjunction with the results in table 11.11 on the characteristics of employees who leave plants (we do not use the term fired as we cannot distinguish between voluntary and involuntary actions), suggest that job turnover was higher in ownership change establishments, especially for these types of workers.

436

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

Table 11.14

Recently hired employees: General statistics for workers at year T 1 whose plants experienced an ownership change during year T

Employee type Recently hired employees (between year T and T 1): % Female % Non-Swedish % Age above mean % Experience above mean % Education less than high school % Education to high school % Education some college All employees: % Female % Non-Swedish % Age above mean % Experience above mean % Education less than high school % Education to high school % Education some college Sample size for employees coming to plant: All With experience data With education data Sample size for all employees: All With experience data With education data

All ownership changes

No ownership change

28.13 12.53 37.09 41.94 31.72 52.56 15.72

27.99 13.16 25.45 33.12 26.86 55.01 18.13

27.70 13.08 49.58 57.09 38.10 49.11 12.78

26.14 12.33 48.75 57.16 36.62 49.84 13.54

123,789 65,410 120,071

1,507,679 910,481 1,445,262

351,269 156,689 343,979

6,239,988 2,858,277 6,090,512

In table 11.15, we reverse the analysis of table 11.12 by identifying workers at the end of year T 1 and then determining their employment status at the end of year T – 1. The results presented in table 11.15 imply that plants involved in an ownership change were more likely to hire new workers than plants that did not experience these transactions. There is also an element of consistency with the regression results presented earlier, which indicated that ownership change was associated with downsizing of the workforce. That is, while three quarters of workers (75.8 percent) at plants not experiencing an ownership change were still employed at the same plant, only two thirds (64.8 percent) of workers whose establishment changed owners in the previous year remained employed at the same facility. Table 11.16 presents the two-year mean earnings growth of workers identified at the end of year T 1. The high wage growth apparent for employees coming from other nonmanufacturing industries and from unknown industries or plants reflects the fact that employees from these sources previously earned much less than other employees, which stems

Same plant Other plant owned by same firm Other plant owned by acquiring firm Same industry, other firm Other manufacturing industry, other firm Other nonmanufacturing industry, other firm Unknown industry or plant (could include self-employed and start-ups) Not employed in Sweden

4.4 4.5 15.2

3.8 5.7 12.4

3.9

5.4 13.2

4.8 1.6 13.6

4.3 3.9 1.5 10.4

3.9

4.1

1.5 11.2

2.9

76.4 2.1 0.0 1.3

75.8 2.1 0.0 1.2 74.1 2.0 0.0 1.1

2.9

4.3

4.0

64.2 2.8 4.8 1.1

Non-Swedish

Young Old

2.5

2.6 7.0

6.4

1.5 10.8

4.3

4.0

1.6 14.3

2.8

3.5

1.7 19.0

5.8

5.5

1.4 3.1

2.4

2.3

1.7 30.9

7.4

6.7

50.3 1.6 0.0 1.5

5.2 20.2

5.4

5.3

Plants not experiencing an ownership change 64.9 87.4 76.1 74.2 1.9 2.3 2.1 2.3 0.0 0.0 0.0 0.0 1.1 1.4 1.0 1.2

4.1 14.7

2.6

3.5

43.5 2.5 3.3 1.2

Less experience

4.5 31.4

5.6 12.9

4.1

4.0

Plants experiencing an ownership change 66.2 56.0 73.6 64.5 2.9 4.1 3.6 3.0 3.7 4.5 4.0 4.8 1.3 1.1 1.2 0.9

Swedish

5.6 6.0

Female

65.0 3.8 3.8 1.2

Male

64.8 3.5 4.1 1.2

All employees

1.2 5.5

4.1

3.9

81.5 2.4 0.0 1.3

5.6 7.6

4.0

4.0

69.3 3.4 4.7 1.3

More experience

1.5 7.4

2.8

2.8

82.6 1.8 0.0 1.0

5.5 9.4

2.6

2.9

70.9 3.8 3.8 1.0

Less than high school

1.5 12.6

4.6

4.3

73.8 2.0 0.0 1.2

5.3 14.8

4.5

4.4

62.6 3.3 3.8 1.2

High school

1.4 14.0

5.8

5.2

68.2 3.4 0.0 2.0

5.8 15.3

5.6

4.9

57.1 3.3 6.4 1.7

At least some college

Where employees come from: Employment status at the end of year T – 1 (in %) of workers at year T 1 whose plants experienced an ownership change during year T

Same plant Other plant owned by same firm Other plant owned by acquiring firm Same industry, other firm Other manufacturing industry, other firm Other non manufacturing industry, other firm Unknown industry or plant (could include self-employed and start-ups) Not employed in Sweden

Employment Status at the end of year T 1

Table 11.15

Note: n.a. not available.

Same plant Other plant owned by same firm Other plant owned by acquiring firm Same industry, other firm Other manufacturing industry, other firm Other nonmanufacturing industry, other firm Unknown industry or plant (could include self-employed and start-ups)

Young Old

3.852

1.978

3.204 5.723

3.034 3.565

3.086

4.140

2.120

3.983

2.987

1.765

5.191

4.176

1.845

5.088

3.165

1.924

2.944

2.884

1.392

6.485

3.420

2.275

1.407 1.452 n.a. 1.775

2.427

2.790

1.172

3.137

2.503

3.279

1.702

1.555

1.917

3.605

1.613

1.420 1.438 1.376 1.636

Less experience

Plants not experiencing an ownership change 1.120 1.282 1.281 1.004 1.127 1.379 1.456 1.008 1.178 1.169 n.a. n.a. n.a. n.a. n.a. 1.512 1.485 1.135 1.362 1.336

2.912

1.522 3.083

1.692

1.691

1.975

Non-Swedish

Plants experiencing an ownership change 1.081 1.260 1.000 1.118 1.207 0.981 1.084 1.023 1.401 1.012 1.149 1.396 1.405 1.072 1.251 1.269

Swedish

2.844

1.774

3.255

3.128

1.657

1.073 1.084 n.a. 1.321

1.500

1.533

1.253 1.108 1.430 1.292

Female

1.126 1.176 n.a. 1.359

1.061 1.068 1.069 1.240

Male

1.113 1.077 1.187 1.253

All employees

4.774

2.770

1.498

1.157 1.246 n.a. 1.207

2.000

2.582

1.296

1.128 1.075 1.320 1.121

More experience

2.878

2.980

1.559

1.056 1.083 n.a. 1.274

1.599

3.514

1.424

1.050 0.997 1.031 1.185

Less than high school

Mean earnings growth of post-ownership change workers classified by pre-ownership change employment status

Same plant Other plant owned by same firm Other plant owned by acquiring firm Same industry, other firm Other manufacturing industry, other firm Other nonmanufacturing industry, other firm Unknown industry or plant (could include self-employed and start-ups)

Employment Status at the end of year T 1

Table 11.16

4.784

3.283

1.885

1.168 1.217 n.a. 1.408

2.188

3.310

1.604

1.149 1.120 1.309 1.283

High school

5.634

2.562

1.762

1.185 1.219 n.a. 1.362

2.247

2.110

1.498

1.195 1.178 1.187 1.260

At least some college

Ownership Change, Productivity, and Human Capital

439

from some combination of fewer working hours (some may have just entered the workforce) and lower salaries.19 The patterns in tables 11.15 and 11.16 indicate greater mobility and higher earnings growth for Swedish employees, younger and less-experienced workers, and employees with higher levels of education. We find that new employees in ownership change plants experienced lower earnings growth than employees being hired by other plants. Our findings also imply that this relative earnings diﬀerential was greater for women, non-Swedish born employees, young employees, experienced employees, and non-college-educated employees than for male, Swedishborn, young, inexperienced, and college-educated employees. 11.7 Conclusions In this chapter, we have generated some stylized facts concerning the average consequences of ownership change for productivity and investment in human capital. The empirical analysis is based on a rich matched employer-employee data set, containing information on 19,010 Swedish manufacturing plants for the years 1985 to 1998. As such, this chapter is the first plant-level study based on evidence from continental Europe and the first analysis of ownership change (in any country) using matched employer-employee data. In contrast to existing plant-level studies, we use more robust econometric methods that adjust for survivor and measurement error biases. The results support theories of ownership change predicting improved economic performance. Our findings are consistent with recent theoretical and empirical evidence (see Jovanovic and Rousseau 2002; Maksimovic and Phillips 2001, 2002) suggesting that takeovers and asset sales result in the reallocation of a firm’s resources to more eﬃcient uses and to better managers. We find that establishments averaged 7 percent lower total factor productivity, and 2 percent lower labor productivity, than comparable plants just before a change in ownership and steadily improved to normal productivity thereafter. Short-term patterns (five years before and after transactions) diﬀered from long-term patterns (ten years before and after 19. For example, among all employees who experienced ownership change, those coming from a nonmanufacturing industry in another firm earned in the base year only 30 percent of the inflation-adjusted mean earnings, and those from an unknown industry or plant earned in the base year only 52 percent of the inflation-adjusted mean earnings. These mean earnings are below those for all other employee source categories in the table. Employees from another manufacturing industry and firm earned in the base year 65 percent of the inflationadjusted mean earnings, employees from another firm in the same industry earned 89 percent, employees from the same plant earned 92 percent, employees from the acquired firm earned 94 percent, and employees from another plant of the acquiring firm earned 102 percent. As these percentages suggest, employees entering plants that experienced ownership change tended to earn slightly below the economy-wide mean when not controlling for industry- or occupation-specific components of earnings.

440

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

transactions) and varied substantially over individual years, a stylized fact that underscores the benefit of having a long annual panel. Plants involved in an ownership change had higher output and employment before the transaction. The increase in labor productivity after the transfer of ownership appears to be the result of a decline in output, combined with an even larger reduction in employment. These patterns emerge most strongly for full acquisitions and divestitures and unrelated acquisitions. We also find that plants involved in an ownership change experienced an upgrading in the quality of human capital. That is, we observe increases in average employee age, experience, and the percentage of employees with a college education. Ownership change also led to an increase in earnings and a reduction in the percentage of female workers. Our analysis of individual-level data allows us to directly track the movement and relative compensation of workers whose establishments were involved in an ownership change. Several stylized facts emerge from this analysis. It appears that ownership change resulted in substantial job loss for women and non-Swedish employees. However, a more comprehensive analysis reveals that turnover rates among such employees were also high in plants that did not experience an ownership change. We also find that employees who leave a plant in the aftermath of ownership change tend to be less experienced, younger workers. It is important to note that Sweden has influential unions, with a membership rate of about 70 percent. Collective agreements, which regulate some firing procedures, could lead to a last-in, first-out policy that might have driven the tendency for younger, less experienced workers to leave.20 Indeed, this pattern also explains the high job loss for women and non-Swedes, since these employees tend to be relatively inexperienced with shorter job tenure. Another stylized fact is that highly-educated workers appear to be the most mobile employees. The findings also imply that women, foreign-born, and young workers employed at plants involved in an ownership change experience greater job loss and reductions in earnings than comparable workers at plants that are not involved in such transactions. The latter result should be interpreted with caution because we do not have data on hours worked. In future work, we hope to implement the robustness tests outlined in Van Biesebroeck (2004) by employing nonparametric and semiparametric methods to compute productivity and then reestimating the various 20. It is rare for a company to opt out of a collective agreement. How rare that is is reflected in the case of the retail chain Toys R Us. When the firm opened its first stores in Sweden, the CEO refused to sign the collective agreement for retail workers. Their refusal to do so was front page news in Sweden for several weeks. This led to a boycott of the company by other unions, which meant that painters, carpenters, electricians, and other skilled workers refused to work for them. After losing money due to the boycott, they signed the agreement.

Ownership Change, Productivity, and Human Capital

441

econometric models. Given our large sample size, we can also analyze whether there are significant diﬀerences across industrial sectors in the impacts of ownership change on economic performance and human capital. Finally, it would be useful to discriminate among the three theories that predict a positive eﬀect of ownership change on economic performance: agency theory, matching theories of ownership change, and the capital upgrading theory of ownership change.

References Baldwin, J. R. 1998. The dynamics of industrial competition. Cambridge: Cambridge University Press. Bartelsman, E. J., and M. Doms. 2000. Understanding productivity: Lessons from longitudinal microdata. Journal of Economic Literature. 38 (3): 569–94. Bhagat, S., A. Shleifer, and R. W. Vishny. 1990. Hostile takeovers in the 1980s: The return to corporate specialization. Brookings Papers on Economic Activity, Microeconomics: 1–72. Brown, C., and J. L. Medoﬀ. 1988. The impact of firm acquisition on labor. In Corporate takeovers: Causes and consequences, ed. A. J. Auerbach, 9–32. Chicago, IL: University of Chicago Press. Caves, R. E. 1998. Industrial organization and new findings on the turnover and mobility of firms. Journal of Economic Literature 36 (4): 1947–82. Conyon, M., S. Girma, S. Thompson, and P. W. Wright. 2001. Do hostile mergers destroy jobs? Journal of Economic Behavior and Organization 45 (4): 427–40. ———. 2002. The impact of mergers and acquisitions on company employment in the United Kingdom. European Economic Review 46 (1): 31–49. Fisher, F. M., and J. McGowan. 1983. On the misuse of accounting rates of return to infer monopoly profits. American Economic Review 73 (1): 82–97. Gort, M. 1969. An economic disturbance theory of mergers. Quarterly Journal of Economics 83 (4): 624–42. Gugler, K., D. C. Mueller, B. B. Yurtoglu, and C. Zulehner. 2003. The eﬀect of mergers: An international comparison. International Journal of Industrial Organization 21 (5): 625–53. Gugler, K., and B. B. Yurtoglu. 2004. The eﬀect of mergers on company employment in the USA and Europe. International Journal of Industrial Organization 22 (4): 481–502. Harris, R. D. S. Siegel, and M. Wright. 2005. Assessing the impact of management buyouts on economic eﬃciency: Plant-level evidence from the United Kingdom. The Review of Economics and Statistics 87 (1): 148–53. Hayward, M. L. A., and D. C. Hambrick. 1997. Explaining the premiums paid for large acquisitions: Evidence of CEO hubris. Administrative Science Quarterly 42 (1): 103–27. Holmes, T. J., and J. A. Schmitz, Jr. 1990. A theory of entrepreneurship and its application to the study of business transfer. Journal of Political Economy 98 (2): 265–94. Jensen, M. C. 1988. Takeovers: Their causes and consequences. Journal of Economic Perspectives 2 (1): 21–48.

442

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

———. 1993. The modern industrial revolution: Exit and the failure of internal control systems. Journal of Finance 48:831–80. Jovanovic, B. 1979. Job matching and the theory of labor turnover. Journal of Political Economy 87 (5): 972–90. Jovanovic, B., and P. Rousseau. 2002. Mergers as reallocation. NBER Working Paper no. 9279. Cambridge, MA: National Bureau of Economic Research, October. Lichtenberg, F. R., and D. Siegel. 1987. Productivity and changes in ownership of manufacturing plants. Brookings Papers on Economic Activity 1987 (3): 643–73. ———. 1990a. The eﬀect of leveraged buyouts on productivity and related aspects of firm behavior. Journal of Financial Economics 27 (1): 165–94. ———. 1990b. The eﬀect of ownership changes on the employment and wages of central oﬃce and other personnel. Journal of Law and Economics 33 (2): 383– 408. Link, A., and D. S. Siegel. 2007. Innovation, entrepreneurship, and technological change. Oxford, U.K.: Oxford University Press. Maksimovic, V., and G. Phillips. 2001. The market for corporate assets: Who engages in mergers and asset sales and are there eﬃciency gains? Journal of Finance 56 (6): 2019–65. ———. 2002. Do conglomerate firms allocate resources ineﬃciently across industries: Theory and evidence. Journal of Finance 57 (2): 721–67. Manne, H. 1965. Mergers and the market for corporate control. Journal of Political Economy 73 (2): 110–20. McGuckin, R. H., and S. V. Nguyen. 1995. On productivity and plant ownership change: New evidence from the longitudinal research database. RAND Journal of Economics 26 (2): 257–76. ———. 2001. The impact of ownership change: A view from labor markets. International Journal of Industrial Organization 19 (5): 739–62. McGuckin, R. H., S. V. Nguyen, and A. P. Reznek. 1998. On the Impact of Ownership Change on Labor: Evidence from Food Manufacturing Plant Data. In Labor statistics measurement, National Bureau of Economic Research, Studies in Income and Wealth, Volume 60, J. Haltiwanger, M. Manser, and R. Topel, ed., 207– 46. Chicago, IL: University of Chicago Press. McWilliams, A., and D. Siegel. 1997. Events studies in management research: Theoretical and empirical issues. Academy of Management Journal 40 (3): 626–57. Meade, J. E. 1968. Is “The New Industrial State” Inevitable? Economic Journal 78 (310): 372–92. Mueller, D. C. 1969. A theory of conglomerate mergers. Quarterly Journal of Economics 83 (4): 643–59. Ravenscraft, D. J., and F. M. Scherer. 1987. Mergers, sell-oﬀs, and economic eﬃciency. Washington, D. C.: The Brookings Institution. Roll, R. 1986. The hubris hypothesis of corporate takeovers. Journal of Business 59 (2): 197–216. Schoar, A. 2002. Eﬀects of corporate diversification on productivity. Journal of Finance 57 (6): 2379–2403. Shleifer, A. 2001. Ineﬃcient markets. New York: Oxford University Press. Siegel, D. S. 1999. Skill-biased technological change: Evidence from a firm-level survey. W. E. Upjohn Institute for Employment Research, Kalamazoo, MI: W. E. Upjohn Institute Press. Van Biesebroeck, J. 2004. Robustness of productivity estimates. NBER Working Paper no. 10303. Cambridge, MA: National Bureau of Economic Research, January.

Ownership Change, Productivity, and Human Capital

Comment

443

Judith K. Hellerstein

Siegel, Simon, and Lindstrom’s chapter “Ownership Change, Productivity, and Human Capital: New Evidence from Matched Employer-Employee Data” provides us with another important example of the power of matched employer-employee data to describe and illuminate various aspects of employment dynamics. Although there are now many examples of matched data sets around the world,1 in many ways we are still just beginning to explore the dimensions over which they can be constructed and analyzed. Using a new matched data set from Sweden, Siegel, Simon, and Lindstrom’s chapter provides an important new contribution as the first paper to examine employment dynamics across heterogeneous workers that are associated with ownership change.2 At the start of the chapter, the authors provide a comprehensive review of the literature on the impact of ownership change on productivity and employment. One of the striking aspects in this literature is how diﬃcult it is to tease out the true eﬀects of ownership change. The first problem is with (mis)measurement: in particular, measurement error in inputs to production, such as capital, can be really problematic. A second and related problem is the treatment of ownership change as exogenous. Both of these problems manifest themselves into the existence of unobservables in the estimating equation that can lead to biased estimates of the impact of ownership change on outcomes. Given all of this, the authors of this chapter take the very reasonable approach of treating their results as descriptive in nature, rather than representing causal relationships. Indeed, there is much to be learned from the descriptive conditional correlations they present. For example, there actually is strong suggestive evidence that assuming that ownership change is exogenous, or uncorrelated with the error term in the estimation equation, is an untenable assumption. Ownership change does not happen randomly. This is seen most clearly in figure 11.2 and table 11.9, column (2), where it is clear that total factor productivity is lower pre-ownership change for plants that will change owners in the future relative to plants that will not change owners, and indeed, that this relative gap actually grows in the years prior to ownership change. The lack of valid instruments makes it hard to imagine how one would tease out the actual eﬀects of ownership change, Judith K. Hellerstein is an associate professor of economics at the University of Maryland and a research associate of the National Bureau of Economic Research. 1. For a description of early work using matched employer-employee data and a description of early data sets, see Abowd and Kramarz (1999). 2. Related papers using administrative data to examine worker turnover with mass layoﬀ (but not ownership change) include Jacobson, Lalonde, and Sullivan (1993) and Lengermann and Vilhuber (2002).

444

Donald S. Siegel, Kenneth L. Simons, and Tomas Lindstrom

but in some sense the first-order fact that needs to be known is that ownership change is associated with other dramatic changes in the establishment. The existence of matched employer-employee data such as is used in this chapter allows for an analysis not only of how the ownership change is associated with changes in the characteristics of workers in jobs in establishments, but also allows for worker-level analyses of the relationship between ownership change and job flows of specific workers in and out of plants that experience ownership change relative to those that do not. This chapter provides a nice analysis along both of these dimensions. Table 11.10 and figures 11.5 through 11.10 show clearly how ownership change is associated with changes in the characteristics of workers in these establishments. Interestingly, while there are statistically significant increases in the age, experience, and earnings of workers in establishments that experience ownership change, as well as increases in the percentage of workers who are male, these results are not quantitatively big, measured either relative to the mean of these variables in the sample or when thinking about them in terms of standard deviations. The process of labor adjustment, as measured in terms of composition of workers, appears to be a gradual process, which may reflect the impact of strong union contracts, something that the authors note as being important in the Swedish labor market, and something that is worth formal exploration in the future. Another aspect of this that is not analyzed fully in this chapter is labor adjustment on the hours margin. Hours adjustments may be easier to make than employment adjustments, but as the authors note in footnote 16, while there appears to be a big hours adjustment associated with ownership change, it is only apparent in the year prior to ownership change. This hours adjustment accounts fully for the dip in mean earnings in (only) that year. It is not clear whether this is a statistical anomaly or reflects something structural about the way that ownership change actually occurs. Tables 11.11 through 11.16 provide interesting and important first snapshots of the churning of workers in and out of establishments as measured at the individual level. The findings in these tables are not fully comparable to the establishment-level findings because these are raw figures rather than regression-adjusted figures,3 and they do not contain standard errors that allow for statistical inference. Nonetheless, they show how regressions based on what is happening at the level of the establishment have the potential to mask important churning by individual workers. For example, tables 11.11 and 11.14 suggest that education is associated with higher rates of job (establishment) mobility, and moreover, that ownership change appears to be associated with yet higher rates of both departure and ac3. It would be straightforward to include establishment-level controls in regressions at the individual level where the dependent variable was a binary indicator for mobility, and where standard errors were clustered at the establishment level. The results would then be directly comparable to the findings from the establishment-level regressions that the authors present.

Ownership Change, Productivity, and Human Capital

445

cession for more educated workers. Further research into the extent and type of churning using individual-level observations from matched data will surely lead to an even more robust understanding of job mobility in manufacturing generally in Sweden, and the relationship specifically between ownership change and heterogeneous churning. Siegel, Simon, and Lindstrom provide an excellent first look at how ownership change is associated with labor dynamics across diﬀerent types of workers. This type of analysis can only be done with rich longitudinal data on workers and the establishments in which they work; data that must include detailed information on the establishments themselves (including, obviously, the ownership structure), and information on the demographic characteristics of the workers. These are very large data requirements, indeed. The hope is that data sets as rich as the one in this chapter will continue to be constructed in many countries and utilized in research to better our understanding of productivity and heterogeneous labor in dynamic economies. References Abowd, J. M., and F. Kramarz. 1999. The analysis of labor markets using matched employer-employee data. In Handbook of Labor Economics, Vol. 3, Part B., ed. Orley Ashenfelter and David Card, 2629–2710. North Holland: Elsevier. Jacobsen, L. S., R. J. Lalonde, and D. G. Sullivan. 1993. Earnings losses of displaced workers. American Economics Review 83 (4): 685–709. Lengermann, P., and L. Vilhuber. 2002. Abandoning the sinking ship: The composition of worker flows prior to displacement. LEHD Technical Paper No. TP-2002-11, August.

12 The Link between Human Capital, Mass Layoﬀs, and Firm Deaths John M. Abowd, Kevin L. McKinney, and Lars Vilhuber

12.1 Introduction The fairly sizable economics literature on displaced workers has typically concentrated on the eﬀects of displacement on worker outcomes (Anderson and Meyer 1994; Bowlus and Vilhuber 2002; Fallick 1996; Jacobson, LaLonde, and Sullivan 1993b; Kletzer 1998; Kuhn and Sweetman 1998; Ruhm 1994; Schoeni and Dardia 1996; Stephens Jr. 2002), which is also an important subject in the field of Human Resource Management (Davis, Savage, and Stewart 2003; Grossman 2002). The analysis typically occurs at the level of a single plant or a sample of workers, for whom the displacement event itself is a given. A mostly separate and distinct literature considers the causes of firm or plant exit (death), and reductions in John M. Abowd is the Edmund Ezra Day Professor of Industrial and Labor Relations at Cornell University and a research associate of the National Bureau of Economic Research. Kevin L. McKinney is an economist in the Longitudinal Employer-Household Dynamics program at the U.S. Census Bureau, and an administrator of the California Census Research Data Center. Lars Vilhuber is a senior research associate at the Cornell Institute for Social and Economic Research and a senior research associate in the Longitudinal EmployerHousehold Dynamics program at the U.S. Census Bureau. The authors wish to thank Kristin Sandusky for providing us with the data extract from the Business Register. The authors acknowledge the substantial contributions of the staﬀ and senior research fellows of the U.S. Census Bureau’s Longitudinal Employer-Household Dynamics (LEHD) Program. This research is a part of the U.S. Census Bureau’s Longitudinal Employer-Household Dynamics Program (LEHD), which is partially supported by the National Science Foundation Grant SES-9978093 to Cornell University (Cornell Institute for Social and Economic Research), the National Institute on Aging Grant R01 AG018854, and the Alfred P. Sloan Foundation. The computations for this research were done while at the Census Research Data Centers in Los Angeles, CA, and Ithaca, NY. This research is partially supported by the National Science Foundation Information Technologies Research Grant

447

448

John M. Abowd, Kevin L. McKinney, and Lars Vilhuber

employment levels (downsizing) (Audretsch 1994; Bernard and Jensen 2002; Davis, Haltiwanger, and Schuh 1996; Dunne and Roberts 1990; Dunne, Roberts, and Samuelson 1988; Haltiwanger, Lane, and Spletzer 2007; McGuckin and Nguyen 1995, 2001). The usual explanatory variables for firm exits are size, age, innovations (Audretsch 1994; Dunne, Roberts, and Samuelson 1988), market structure, and eﬃciency (Foster, Haltiwanger, and Krizan 1998, 2002; Kletzer 1998). Few authors explicitly link the micro-level movement of workers with death and downsizing at either the plant or firm level. Notable exceptions are Abowd, Corbel, and Kramarz (1999), Lengermann and Vilhuber (2002), and Carneiro and Portugal (2003), which we will describe shortly. There are many reasons employers use mass layoﬀs. They are not synonymous with the death of an establishment or firm. In our data, 55 percent of firms that have one displacement event between 1993 and 1996 are still alive in 1997. Conversely, if the decline in size is gradual, a firm death may not result in a displacement event. Abowd, Corbel, and Kramarz (1999) have previously investigated worker and job flows for establishments with declining employment, and Lengermann and Vilhuber (2002) considered the distribution of worker skill levels in flows out of firms prior to displacement events. Both find that there are changes in such flows relative to alternate establishments (establishments with stable or increasing employment, or the same establishments in prior periods), but neither address the point specifically as a potential latent cause of displacement. In the absence of a direct measure of worker skills, the literature has used wages as both a proxy for skills and as a cost component. Dunne and Roberts (1990), Bernard and Jensen (2002), and Carneiro and Portugal (2003) consider the determinants of wages and the eﬀects on plant closures (Bernard and Jensen 2002; Dunne and Roberts 1990) or displacement events (Carneiro and Portugal 2003). Dunne and Roberts (1990) find that higher-paying firms have a significant, but economically small increase in the likelihood of plant failure. One postulated explanation for this small eﬀect is that plants with higher wages also have a more productive work-

SES-0427889, which provides financial resources to the Census Research Data Centers. This document reports the results of research and analysis undertaken by U.S. Census Bureau staﬀ. It has undergone a Census Bureau review more limited in scope than that given to oﬃcial Census Bureau publications. The views expressed herein are attributable only to the authors and do not represent the views of the U.S. Census Bureau, its program sponsors, Cornell University, or data providers. Some or all of the data used in this chapter are confidential data from the LEHD Program. The U.S. Census Bureau supports external researchers’ use of these data through the Research Data Centers (see www.ces.census.gov). For other questions regarding the data, please contact Jeremy S. Wu, Program Manager, U.S. Census Bureau, LEHD Program, Data Integration Division, Room 6H136C, 4600 Silver Hill Rd., Suitland, MD 20233, U.S.A., or visit http://lehd.did.census.gov

The Link between Human Capital, Mass Layoﬀs, and Firm Deaths

449

force. In contrast to Dunne and Roberts (1990), Bernard and Jensen (2002) found that plants paying above-average wages have a lower likelihood of exiting. One plausible explanation for this result is that these firms use above-average human capital. Our analysis directly addresses this issue by deriving a measure of general (nonfirm-specific) human capital and using it instead of, or in addition to, wages. Closest in spirit to our analysis is Carneiro and Portugal (2003), who estimate simultaneous plant failure and wage determination equations for Portugal. Their measures of human capital are the usual right-hand side variables in a Mincerian wage equation (education, age, tenure), but they do not include a measure of the physical capital of the firm. Furthermore, plant layoﬀ is estimated as an event at the individual level. The distribution of human capital within the firm is not taken into account. In this chapter, we correlate firm-level measures of human and physical capital (capital intensity), as well as measures of eﬃciency (sales per worker) with displacement events. We diﬀer from the literature in our use of a measure of human capital, rather than a direct measure of wages, and we consider the eﬀect of the distribution of human capital within a firm on both displacement and firm-death outcomes. Based on methods first developed in Abowd, Lengermann, and McKinney (2002), we estimate a measure of human capital, based on observed and unobserved worker ability. Using this measure, we estimate firmspecific distributions, which allows us to consider the impact of diﬀerences in the use of human capital across firms on outcome variables. We identify displacements from quarterly worker flows, and merge data on firm performance and capital from Economic Censuses. Firms active in our data in 1992 are classified as survivors or exiters, depending on their activity in 1997. This set of companies is then cross-classified by whether or not they have experienced a single mass layoﬀ, or multiple displacement events. To anticipate our results, single displacement events occur substantially more often in firms that employ more workers in the lowest quartile of the human capital distribution. Firm closures occur substantially more often in firms that disproportionately employ workers in the lowest quartile of the human capital distribution and less often in firms that employ relatively more workers in the highest quartile of this distribution. Conditioning the firm closure analysis on the displacement event, our analysis suggests that firms that disproportionately employ workers in the highest quartile of the skill distribution are less likely to close, even given a displacement event, than are other firms that experience displacement events. The chapter is organized as follows. Section 12.2 lays out the basic definitions of human capital and displacement as used throughout our chapter. Section 12.3 describes the data used, section 12.4 provides results, and section 12.5 concludes.

450

John M. Abowd, Kevin L. McKinney, and Lars Vilhuber

12.2 Definitions The definitions of economic activity, human capital, and mass layoﬀ are obviously crucial for our analysis. In this section we consider each concept in turn, state our definition, and relate our measure to alternatives that have been used by others. 12.2.1 Defining Economic Activity Economic activity of firms in our sample is defined by the observation of positive employment in the Longitudinal Employer-Household Dynamics (LEHD) Infrastructure files in 1992. Survivors are those firms that are still active, by the same definition, in 1997. Additional data are matched from the Economic Census in 1992 and 1997, when available. This universe is essentially the ES-202 establishments that were alive in 1992 (positive employment), which are the basis for the LEHD Infrastructure File known as the Employer Characteristics File (Abowd et al., chapter 5 in this volume). An alternative frame might have been either the Economic Census frame, in which case some establishments in-scope but missing from the LEHD Infrastructure Files would have missing human capital data. 12.2.2 Defining Human Capital We provide a brief overview of our approach in this section. For a more complete discussion of the definition of the within-firm human capital distribution, and of human capital itself, see Abowd, Lengermann, and McKinney (2002). Assume human capital Hit has a market-return (average rental rate) rt. The wage is wit rt Hit, where i indexes persons and t indexes time. Individual firms might deviate from rt, paying rt pj, with E [ pj ] 1, where j indexes employers. Assume that a person-specific component (i) and a general experience component (Xit) are important factors determining the accumulation of human capital. Then, taking logarithms, we have ln Hit i Xit. We thus obtain the following model of earnings (1)

ln wit ln rt ψj i Xit

where ψj ln pj. Deviating wit and Xit from the grand mean across individual and time periods produces the estimating equation:1 (2)

ln wit i ψj Xit εijt

where i is the person eﬀect, ψj is the firm eﬀect, and Xit are time-varying person characteristics (such as experience), and εijt is the statistical residual. To compute a measure of a person’s human capital, we combine the es1. See Abowd, Creecy, and Kramarz (2002) for details. We have not changed the notation for the wage rate or the experience variables since subtracting a constant is just a technique for imposing one of the identification requirements for the estimation of both person and firm eﬀects.

The Link between Human Capital, Mass Layoﬀs, and Firm Deaths

451

timated person eﬀect ˆi, the experience components (after restoring the mean of Xit) of person characteristics Xitˆ , and the reference constant to compute hˆit ˆi Xitˆ .

(3)

Because the estimated person eﬀect (ˆ i ) absorbs all the usual time-invariant explanatory factors (such as sex, education, and age at first entry) and also absorbs all unobserved (by the analyst) time-invariant factors, such as innate ability, hit corresponds to the concept of general human capital. Once hˆit is computed, we estimate firm-level kernel density estimates of its distribution, yielding a firm-specific distribution of human capital gjt(hˆit), and hˆ

Gjt(hˆ) gjt(x)dx

(4)

H

where H and H define the support of hˆ. To obtain discrete measures, we partition [ H, H ] into 4 subsets, and calculate the population quartiles q∗k implicitly defined by (5)

q∗k

G(q ) g(x)dx k · 0.25 ∗ k

H

where k 1, 2, 3, 4. For each firm, we then calculate the proportion of workers who have human capital within the ranges defined by the overall population quartile boundaries q∗k for for k 1, 2, 3, 4. (6)

jt(k) Gjt(q∗k ) Gjt(q∗k 1).

These employer-level measures summarize the complete distribution of workers’ human capital at the establishment. Similar measures (k) are computed for the experience (Xit ˆ ) and estimated person eﬀect (ˆi ) distributions within the firm. While our firm-level human capital measure is obviously related to wage rates, it is important to note that its distribution is diﬀerent from the distribution of wage rates at the establishment at a point in time. By removing the firm eﬀects and the idiosyncratic residuals from the labor market distribution of h, from which we measure the reference quartile points, between-firm diﬀerences in compensation policy, which might be due to specific human capital or other active compensation policies, are removed from the human capital measure. Generally, such eﬀects are included in within-firm wage dispersion measures used by other authors (e.g., Gibbons et al. 2005; Lluis 2005). 12.2.3 Defining Mass Layoﬀs In keeping with the previous literature on the impact on workers of mass layoﬀs (Bowlus and Vilhuber 2002; Jacobson, La Londe, and Sullivan 1993a, 1993b; Schoeni and Dardia 1996), we define a mass layoﬀ in period

452

John M. Abowd, Kevin L. McKinney, and Lars Vilhuber

t as a 30 percent single-period drop in employment from firm j ’s maximum employment level over the observed time period: (7)

Sjt Djt 1 if

0.3 Bmax j

where Bmax maxt Bjq over the time period that firm j is in the sample with j positive employment. Firm j ’s beginning-of-quarter employment, Bjt , is a point-in-time measure derived by summing over workers employed at the firm in both period t – 1 and t. Worker separations from firm j are Sjt (i.e., workers that worked for the firm in period t but are no longer observed on the payroll in period t 1). Note that in our analysis, all firm deaths are classified as displacements. However, they may not involve mass layoﬀs. On the other hand, among survivors, some firms experienced mass layoﬀs, and some did not. Due to some limitations of the administrative data used for this chapter, a naïve use of the mass layoﬀ equation (7) will overstate mass layoﬀs by some margin.2 In order to reduce the impact of spurious events, we take particular care to exclude firms that either change identity or who continue to operate, yet fail to file a firm report. The firm identifier underlying all of our analysis is a state-specific Unemployment Insurance account number, whose primary purpose is to facilitate the administration of a state’s unemployment insurance system. These account numbers can and do change for reasons such as a simple change in legal form or merger. In our analysis, the separation of a worker from a firm is identified by a change in the firm identifier on that worker’s wage records. If a firm changes account numbers, but makes no other changes, the worker would seem to have left the original firm, when in fact his employment status remains unchanged. Thus, a simple change in account numbers would lead to the observation of a mass layoﬀ at the firm associated with the original account number.3 To identify spurious employer birth and death events, we track large worker movements between firms. Benedetto et al. (2007) provide an analysis for one particular state of such an exercise using LEHD data. For this chapter, if we observe 80 percent of a firm (the predecessor) moving to a single successor, then we eliminate the displacement event. The assumption is that such a movement is associated not with a layoﬀ, but a reorganization, a takeover, or some similar event. Similarly, if we observe that 80 percent of a successor’s employment stems from the same predecessor, then a displacement event is also eliminated. 2. Abowd and Kramarz (1999) and Vilhuber (forthcoming) provides an overview over several approaches to correcting the weaknesses of administrative data sets. Abowd and Vilhuber (2005) discuss one particular weakness, a corrective measure, and the impact it has on aggregate statistics, including on measures similar to the mass layoﬀs of interest in our chapter. 3. Other authors working with administrative data have also addressed this problem in similar ways (see Anderson and Meyer 1994 and Jacobson, LaLonde, and Sullivan 1993a).

The Link between Human Capital, Mass Layoﬀs, and Firm Deaths

453

A second, not uncommon, event observed in administrative data is the failure of a firm to file a timely report. In general, such an absence will trigger a follow-up by the state administration, since tax payments are linked to the report. However, for multiple reasons, the corrected or late filing by the firm may not get entered into the database transmitted to LEHD. The result is a hole in the firm’s activity. Holes, however, are precisely what would also be observed if a firm laid oﬀ its entire workforce for more than a quarter—a mass layoﬀ—and hired them or other workers back later. The approach we have taken to distinguish data-related holes from true layoﬀs is the following. Consider the diﬀerent employment path of an individual i at some firm j. Define time to be the elapsed fraction of a quarter t, ∈ [0, 1]. If the individual left the firm at some point 0 1, then observed earnings Eit will be Eit eit , where eit is the quarter t wage rate for individual i. If the individual worked for the entire quarter, then observed earnings will be Eit eit. We do not observe , but assume that the quarterly wage rate is constant (eit ei0), where 0 is some baseline period, typically a prior quarter within the same job history. Compute Ei Eit /Ei0. Then Ei 1 implies 1. On the other hand, if Ei 1, then the worker left at some time 1. Now compute the average, Ejt for all workers within a firm j at date t; that is,

E jt

∑ E ∑1[ j J(i,t)] iT

i∈J(i,t)

i

where J(i,t) is a function giving the identity, j, of the firm employing i at date t (Abowd, Kramarz, and Margolis 1999). Now consider if E jt 1 (i.e., the average ratio of earnings is equal to unity). It is unlikely that all workers leave the firm at the same time, except, of course, if a mass layoﬀ occurs. It is, however, even more unlikely, though not impossible, that that mass layoﬀ occurs on the last day of the quarter, which is what Ejt 1 implies. Suppose further that no employment is observed at firm j in period t 1, but positive and large employment is observed in period t 2, with

E jt2 1 as well. This hole is very unlikely to occur under normal circumstances—it implies that all workers left the firm at the end of quarter t, and all workers started working again on the first day of quarter t 2. It is, however, the data pattern that is expected when a firm neglects to file all worker records for quarter t 1. We consider mass layoﬀ events that are synchronous with such holes to be data artifacts, not true layoﬀ events, and we filter them out. 12.3 Data To estimate the impact of displacement, we use data from three states, California, Illinois, and Maryland, covering the time period 1990 to 2003. The data used to estimate the human capital model and to identify firm

454

John M. Abowd, Kevin L. McKinney, and Lars Vilhuber

displacement events are derived from the LEHD Infrastructure Files. The LEHD Infrastructure Files provide a worker’s quarterly earnings history, basic demographic information, and, most importantly, identify a person’s employer. The fact that we know the history of the firm and the employees at that firm over time allows us to estimate displacement events as well as provide a richer characterization of the employees at the firm. The human capital estimates (equation 2) are calculated using data from the LEHD infrastructure files for the twenty-two states available as of November 30, 2004. Once the estimates have been produced, we select workers employed in California, Maryland, and Illinois during 1992 and/or 1997. In order to get a snapshot or point-in-time measure of the human capital at the firm, we further restrict our analysis to workers employed at the end of quarter 1 (a date that roughly coincides with the collection of Economic Census data). Finally, we only keep workers between the ages of 18 and 70, with earnings during the quarter of greater than $250.00. Additional information, such as a firm’s sales and capital stock, is gathered from the 1992 and 1997 Economic Censuses. Capital stock is only available for the manufacturing sector. Sales variables are available only for a subset of firms. At the time we were preparing this version, the 2002 Economic Census was not available, although we expect to use these data in the future. Because of our desire to incorporate information from the Economic Censuses, we will only directly analyze individuals and firms during the period 1992 to 1997, even though the human capital estimates are calculated using data for the full time period. Due to the dynamic nature of the U.S. economy it is diﬃcult to diﬀerentiate normal flows of employment from displacements for smaller firms. For example, a firm with ten employees that has three workers leave during the quarter would be classified as having a displacement under our standard definition, even though three workers leaving a firm in the same quarter is not a particularly unusual event. In order to focus our analysis on large displacement events, we limit our sample to firms that average at least fifty workers across the entire time period. The firm-level displacement database contains indicators for all displacement events that occur during three time periods: 1992, 1993 to 1996, and 1997. We select a sample of firms that were active during 1992. Firms still present in 1997 are survivors, otherwise firms are called exiters. This sample diﬀers from Abowd, Lengermann, and McKinney (2002), where firms entering the sample between 1992 and 1997 were also included. We do not impose any restrictions on the incidence of displacement events during either 1992 or 1997, although the eﬀect of this decision is worthy of further exploration. Firms are classified as to whether they experienced zero, one, or multiple displacement events in the years 1993 to 1996. Crossclassifying this grouping with exit status yields the six diﬀerent types of firms we focus on in this chapter.

The Link between Human Capital, Mass Layoﬀs, and Firm Deaths

455

12.4 Analysis The incidence and magnitude of displacement events in our data are shown in table 12.1. About 70 percent of the nearly 50 thousand firms in our sample never experience a displacement event between 1993 and 1996. On the other hand, 15 percent experience multiple displacement events over the same period, with the remaining firms experiencing a single event over the period. These groups turn out to be analytically distinct. For most of our analysis we will concentrate on the comparison between firms experiencing zero and one displacement events. However, multiple-displacement firms are a group that remains of interest for this research. A more detailed characterization of the excluded multiple-displacement group can be found in the appendix. About 18 percent of firms die between 1992 and 1997 and, in line with our intuition, firms experiencing a single displacement are much more likely to die, as shown by row two of table 12.2. In this respect, firms with multiple displacements are very similar to firms that experience no displacement at all, suggesting that most firms with multiple displacements structure the firm with the expectation that a sequence of large demand shocks will occur. In general, firms with no displacements are bigger both in terms of their workforce as well as in terms of total sales and sales per worker (when available). In terms of their per-worker capital stock (available for manufacturing firms) firms with no displacement events and firms with multiple displacements are very similar. Single displacement firms have a lower capital intensity than either of the other two groups. Turning to the distribution of human capital, firms with displacement events have an above-average fraction of their workforce in the lower skill () and human capital (h) distribution (1), at the expense of the upper tail of the distribution (4). In general, the human capital seems to be shifted further to the left the more displacement events a firm has. Table 12.1

Displacement and survival among 1992 firms Displacement events

Firm type

0

1

Multiple

Total

Exiters

4,575 50.63 13.61

3,364 37.22 45.18

1,098 12.15 14.86

9,037 18.65

Survivors

29,045 73.68 86.39

4,082 10.36 54.82

6,291 15.96 85.14

39,418 81.35

Total

33,620 69.38

7,446 15.37

7,389 15.25

48,455 100.00

Table 12.2

Characteristics of sample firms in 1992

Displacement events Variable

0

1

Multiple

33,620

7,446

7,389

Exiters

0.1360 (0.3428)

0.4517 (0.4977)

0.1485 (0.3557)

Month 3 employment

277.92 (1,371.33) 432.80 (1,719.46)

a

N

Max lifetime beginning-of-period employment Sales/workers (1000$) Cap stock/worker (1000$) Total receipts (1000$) (1) for hij (2) for hij (3) (4) for hij (4) for hij (1) for experience (2) for experience (3) (4) for experience (4) for experience (1) for (2) for (3) (4) for (4) for Agriculture, forestry, fisheries, mining Construction industries Manufacturing Transportation, communications, and utilities

232.7 (1,243.6) 75.6 (142.5) 47,671.5 (283,177.7)

210.83 (757.27) 314.75 (961.78)

188.66 (519.69) 358.21 (925.10)

166.2 (553.2) 68.5 (122.3) 33,927.8 (209,670.7)

91.1 (178.1) 72.1 (191.2) 15,968.6 (68,518.7)

0.2454 (0.1525) 0.2422 (0.0831) 0.5122 (0.1938) 0.2689 (0.1590) 0.2551 (0.1308) 0.3009 (0.0570) 0.4439 (0.1332) 0.2861 (0.1016) 0.2499 (0.1521) 0.2392 (0.0809) 0.5108 (0.1852) 0.2688 (0.1575)

0.2973 (0.1719) 0.2484 (0.0808) 0.4541 (0.2085) 0.2288 (0.1607) 0.2819 (0.1491) 0.2951 (0.0600) 0.4228 (0.1399) 0.2755 (0.1039) 0.2945 (0.1683) 0.2407 (0.0751) 0.4646 (0.1944) 0.2352 (0.1542)

0.3685 (0.1730) 0.2527 (0.0787) 0.3786 (0.2063) 0.1859 (0.1573) 0.3460 (0.1833) 0.2775 (0.0659) 0.3764 (0.1527) 0.2494 (0.1102) 0.3381 (0.1621) 0.2453 (0.0715) 0.4165 (0.1889) 0.2060 (0.1474)

0.0199 (0.1403) 0.0431 (0.2032) 0.2315 (0.4218) 0.0516 (0.2212)

0.0396 (0.1960) 0.0829 (0.2758) 0.1720 (0.3774) 0.0449 (0.2072)

0.1050 (0.3070) 0.0814 (0.2735) 0.0581 (0.2341) 0.0249 (0.1558)

The Link between Human Capital, Mass Layoﬀs, and Firm Deaths Table 12.2

457

(continued)

Displacement events Variable Wholesale trade Retail trade Finance, insurance, and real estate Service industries Public administration

0

1

Multiple

0.1027 (0.3036) 0.1160 (0.3202) 0.0839 (0.2773) 0.3257 (0.4686) 0.0251 (0.1566)

0.0710 (0.2569) 0.1817 (0.3856) 0.0564 (0.2307) 0.3349 (0.4720) 0.0161 (0.1259)

0.0274 (0.1634) 0.2722 (0.4451) 0.0179 (0.1329) 0.3920 (0.4882) 0.0205 (0.1419)

a

All cells correspond to the number of observations in line 1, except for “Sales per worker” (26,273/5,761/5,473), “Capital stock per worker” (6,720/1,045/346), and “Total receipts” (26,353/5,790/5,493).

While firms with multiple displacements diﬀer in their observable characteristics from firms in other categories, they do not seem to die as often as firms with singe displacements (row two of table 12.2). In other words, they seem to be stable firms with highly volatile, possibly seasonal, workforce fluctuations (see also the discussion in the appendix). Although it is worthwhile to try to further disentangle the correlates of the highly volatile employment patterns, in the remainder of this chapter we will concentrate on firms with zero or one displacement event. 12.4.1 The Probability of Displacement The summary data in table 12.2 suggest that firms with a single displacement event are more likely to die. They also have distinct observable characteristics prior to the displacement event. To disaggregate some of the possible causes of displacement events, we specify a univariate probit model: (8)

Pr (DWj 1|Yj) (Yj)

where DWj is equal to one when firm j experienced a displacement event in the eligible period (in this chapter, between 1993 and 1997). We include in Yj measures of the firm’s human and physical capital or capital intensity (if available), sales per worker (worker productivity), and indicators of firm structure (multi- or single-unit). Also, Yj includes industry level variables for firm j ’s primary SIC: the concentration index or unrestricted industry eﬀects. We estimate equation (8) for all firms active in 1992. Table 12.3 presents the results for the firm performance and human capital variables. Columns (1) through (6) report results for the sample of firms for which complete

Manufacturing Yes No Yes Yes Manufacturing Yes No No Yes

5,655

0.0227 (0.0214) 0.0456 (0.0356) 1.3107 ∗∗∗ (0.2449) –0.1371 (0.4425) –0.2215 (0.3847)

(3)

Manufacturing Yes No Yes Yes

5,561

0.0066 (0.0235) 0.0517 (0.0369) 1.9575 ∗∗∗ (0.3353) 0.1073 (0.463) 0.0838 (0.3991)

(4)

Manufacturing Yes No Yes Yes

5,561

— — 0.0556 (0.0342) 1.9508 ∗∗∗ (0.3344) 0.0998 (0.4621) 0.0809 (0.3989)

(5)

Manufacturing Yes No Yes Yes

5,561

— — — — 1.83 ∗∗∗ (0.3257) 0.0721 (0.4616) 0.0817 (0.3986)

(6)

Notes: Standard errors in parentheses. Sample excludes firms experiencing multiple displacement events. Industry eﬀects are at the SIC division level. All regressions include an indicator for multi-state firms. Workforce characteristics are firm-average race (percentage white), sex (percentage male), and age. ∗∗∗ Significant at the 1 percent level. ∗∗Significant at the 5 percent level. ∗ Significant at the 10 percent level.

Manufacturing Yes No No No

5,655

–0.0077 (0.0231) –0.0317 (0.0354) — — — — — —

–0.0408 ∗∗∗ (0.0211) –0.0598 ∗∗∗ (0.0346) — — — — — —

5,655

(2)

(1)

Probability of displacement, manufacturing

Industry State geography Industry state Workforce characteristics Firm size

Observations

(4) for hij

(2) for hij

(1) for hij

Sales/workers

Cap stock/worker

Variable

Table 12.3

The Link between Human Capital, Mass Layoﬀs, and Firm Deaths

459

sales and assets information is available. This specification is similar to others in the literature (Bernard and Jensen 2002; Carneiro and Portugal 2003; Dunne and Roberts 1990) but is not representative of all firms in our sample. Column (1) is a simple specification that correlates sales and assets per worker to the likelihood of a displacement, conditional on unrestricted industry and geography eﬀects. Both firm performance variables are significant. However, in column (2), the addition of worker demographics reduces the eﬀect of the firm performance variables, and neither is significant at conventional significance levels. Although not reported in the table, the estimated coeﬃcients on worker characteristics suggest that firms with a predominantly male workforce, with more part-time workers, and with an older workforce are all more likely to have a displacement event. In column (3), we replace the observable workforce characteristics with (1), (2), and (4). Note that the only significant variable is (1), a measure of how much of the workforce is in the lower tail of the skill distribution. A larger fraction of a firm’s workforce being in the lower tail of the skill distribution in 1992 increases the likelihood of a displacement event in the next five years. This result remains robust to the introduction of observable workforce characteristics in column (4). Thus, it would seem that the distribution of human capital captures a significant amount of the productivity factors that aﬀect the incidence of mass layoﬀs. As noted previously, the sample of firms with complete sales and asset information is not representative of the full sample. Columns (5) and (6) test specifications that successively eliminate the sales and assets variables for the same sample as in columns (1) and (4). The impact on the coeﬃcient on (1) is negligible. In table 12.4, we widen the sample selection criteria in two ways. First, we expand the sample to include all firms that have available sales data but do not necessarily have available capital intensity data. The sample increases to 25,236 firms. The specification in columns (1) and (2) in table 12.4 corresponds to the one in columns (5) and (6) in table 12.3 estimated for the new sample. In the wider sample, sales per worker still has a significant impact on the incidence of displacement events. The eﬀect of the lower tail of the human capital distribution is reduced, although still highly significant. Whether this result is due to greater homogeneity in the table 12.3 sample or some other factor remains to be explored. Dropping the sales variable in column (2) of table 12.4 increases the absolute value of the (4) coeﬃcient, suggesting that in this sample, there is some correlation between sales per worker and the human capital intensity in the firm. In order to control for local market eﬀects, we fully interacted geography and industry eﬀects, allowing industries in each state to have diﬀerent eﬀects. As the results in columns (3) and (4) illustrate, this has no meaningful impact on the results.

460

John M. Abowd, Kevin L. McKinney, and Lars Vilhuber

Table 12.4

Probability of displacement, other industries

Variable Cap stock/worker Sales/workers (1) for hij (2) for hij (4) for hij Observations Industry State geography Industry state Workforce characteristics Firm size

(1)

(2)

(3)

(4)

(5)

(6)

— –0.065 ∗∗∗ (0.013) 1.1265 ∗∗∗ (0.1355) –0.167 (0.2344) –0.2303 (0.1729)

— — — 1.2492 ∗∗∗ (0.1332) –0.1716 (0.2342) –0.35 ∗∗ (0.1713)

— –0.0609 ∗∗∗ (0.013) 1.149 ∗∗∗ (0.1361) –0.1346 (0.2351) –0.2627 (0.1739)

— — — 1.2665 ∗∗∗ (0.1338) –0.1361 (0.235) –0.3738 ∗∗ (0.1722)

— — — 1.3658 ∗∗∗ (0.0998) –0.2192 (0.1746) –0.1302 (0.1252)

— — — 1.376 ∗∗∗ (0.1002) –0.1994 (0.1754) –0.1377 (0.1258)

25,236

25,236

25,236

25,236

41,066

41,066

Yes Yes No Yes Yes

Yes Yes No Yes Yes

No No Yes Yes Yes

No No Yes Yes Yes

Yes Yes No Yes Yes

No No Yes Yes Yes

Notes: Standard errors in parentheses. Sample excludes firms experiencing multiple displacement events. Industry eﬀects are at the SIC division level. All regressions include an indicator for multi-state firms. Workforce characteristics are firm-average race (percentage white), sex (percentage male), and age. ∗∗∗ Significant at the 1 percent level. ∗∗ Significant at the 5 percent level. ∗ Significant at the 10 percent level.

For a substantial number of establishments, neither sales nor capital were available in the Business Register. Columns (5) and (6) of table 12.4 report results for this most-inclusive sample using specifications similar to columns (2) and (4), respectively. The results are quite congruent with those of the previous columns. We note, however, that in the wider sample, the apparent correlation between sales and (4) found earlier is not statistically significant. 12.4.2 The Probability of Firm Closure While equation (8) focuses on the likelihood of displacement, much of the interest in the literature has been on the likelihood of firm or plant closure. We model this outcome using the probit equation: (9)

Pr(Dj 1|Yj, DWj) (yYj DW DWj)

where Dj is equal to one if firm j exited the market (economy) between 1993 and 1997, and the other vectors are defined as above. Table 12.5 reports results comparable to the previous literature, with DW 0, while table 12.7 introduces the displacement indicator into the equation. The specifications in columns (1) through (6) mirror those in table 12.3, except for the change in the dependent variable. Neither the capital stock/

Manufacturing Yes No Yes Yes

5,561

–0.0082 (0.0225) –0.049 (0.0342) — — — — — —

(2)

Manufacturing Yes No No Yes

5,561

0.0162 (0.0225) 0.0056 (0.0354) 0.2273 (0.2261) –0.8138∗∗ (0.4068) –1.3155∗∗∗ (0.3513)

(3)

Manufacturing Yes No Yes Yes

5,561

0.0045 (0.0228) 0.0176 (0.0356) 0.87∗∗∗ (0.315) –0.32 (0.4257) –0.9963∗∗∗ (0.3671)

(4)

Manufacturing Yes No Yes Yes

5,561

— — 0.0203 (0.0328) 0.8656∗∗∗ (0.3142) –0.3248 (0.425) –0.9981∗∗∗ (0.3669)

(5)

Manufacturing Yes No Yes Yes

5,561

— — — — 0.8209∗∗∗ (0.3056) –0.335 (0.4246) –0.9993∗∗∗ (0.3668)

(6)

Notes: Standard errors in parentheses. Sample excludes firms experiencing multiple displacement events. All regressions include an indicator for multi-state firms. Workforce characteristics are firm-average race (percentage white), sex (percentage male), and age. ∗∗∗ Significant at the 1 percent level. ∗∗ Significant at the 5 percent level. ∗ Significant at the 10 percent level.

Manufacturing Yes No No No

5,561

–0.0083 (0.0206) –0.0492 (0.0333) — — — — — —

(1)

Probability of firm closure, manufacturing

Industry State geography Industry state Workforce characteristics Firm size

Observations

(4) for hij

(2) for hij

(1) for hij

Sales/workers

Cap stock/worker

Variable

Table 12.5

462

John M. Abowd, Kevin L. McKinney, and Lars Vilhuber

worker nor sales/worker has a statistically significant impact on the probability of closure; however, the magnitude and sign of both eﬀects is comparable to that found in table 12.3, except for capital stock/worker in column (1). Introducing observable workforce characteristics has no eﬀect, by itself, on the estimated impact of sales and assets (compare column [2] to column [1]). The introduction of the human capital distribution into the equation shows that having a workforce heavily concentrated in the first quartile (lower tail) of the distribution is associated with a very strong increase in the probability of firm closure. This eﬀect is magnified, not attenuated, when the workforce characteristics are introduced (compare column [4] to column [3]). In contrast with the displacement equation, not only does the lower tail of the distribution matter for firm closure outcomes, but so does the upper tail—having a workforce heavily concentrated in the fourth quartile of the human capital distribution is associated with a strong decrease in the probability of closure. The magnitude of this eﬀect is slightly attenuated by the introduction of workforce characteristics. A firm with an above-average fraction of its workforce in the upper skill distribution, or a below-average fraction of its workforce in the lower tail, has a significantly lower likelihood of exiting the market in the next five years. Entering the human capital distribution into the equation changes the estimated impact of the capital stock/worker ratio from positive to negative, although the estimated eﬀect is still statistically insignificant. These results are robust to the elimination of the assets variable (see column [5]) and both the asset and sales variables together (see column [6])—there is no significant change in the estimated parameters. Table 12.6 shows that when estimating the specification in columns (5) and (6) from table 12.5 on the larger sample, the estimated parameters on (1) and now (4), while still highly significant, are substantially reduced in magnitude. The sales/ worker variable now has a substantial and significant impact on the likelihood of firm death. Introducing the displacement event indicator, DWj, into the firm closure probit equation has no substantive eﬀect on our interpretation of the results. The results are presented in table 12.7 and table 12.8. The displacement indicator is strongly statistically significant in all specifications and samples—not surprising given the raw statistics in table 12.2. The positive regression eﬀect of the displacement event on the probability of firm closure is robust to the inclusion of detailed observable characteristics of the workforce, a full set of industry, geography, and firm structure indicators, and the three skill distribution variables. For the manufacturing firms shown in table 12.7, the eﬀect of having a workforce in the lowest quartile of the human capital distribution is strongly attenuated in comparison to the results not conditioned on the displacement event (compare the same columns in table 12.7 and table 12.5), suggesting that most of the eﬀect of lower skill workers on firm closure operates through increasing the

Yes Yes No Yes Yes

25,236 Yes Yes No Yes Yes

25,236

— — — 0.7406∗∗∗ (0.1339) 0.1209 (0.2352) –0.5943∗∗∗ (0.1738)

(2)

No No Yes Yes Yes

25,236

— –0.0879∗∗∗ (0.0131) 0.5536∗∗∗ (0.1369) 0.1189 (0.236) –0.4664∗∗∗ (0.1762)

(3)

No No Yes Yes Yes

25,236

— — — 0.7229∗∗∗ (0.1345) 0.1023 (0.2357) –0.6255∗∗∗ (0.1744)

(4)

Yes Yes No Yes Yes

41,066

— — — 0.8337∗∗∗ (0.0976) 0.4073 ∗∗ (0.1694) –0.5014∗∗∗ (0.1233)

(5)

No No Yes Yes Yes

41,066

— — — 0.8022∗∗∗ (0.0982) 0.3858 ∗∗ (0.1704) –0.5519∗∗∗ (0.124)

(6)

Notes: Standard errors in parentheses. Sample excludes firms experiencing multiple displacement events. Industry eﬀects are at the SIC division level. All regressions include an indicator for multi-state firms. Workforce characteristics are firm-average race (percentage white), sex (percentage male), and age. ∗∗∗ Significant at the 1 percent level. ∗∗ Significant at the 5 percent level. ∗ Significant at the 10 percent level.

Industry State geography Industry state Workforce characteristics Firm size

Observations

(4) for hij

(2) for hij

(1) for hij

— –0.085∗∗∗ (0.013) 0.5802∗∗∗ (0.1362) 0.1415 (0.2354) –0.4383∗∗ (0.1755)

(1)

Variable

Cap stock/worker Sales/workers

Probability of firm closure, all industries

Table 12.6

Manufacturing Yes No No No

— Manufacturing Yes No Yes Yes

—

1.2336∗∗∗ (0.0541) –0.0075 (0.0237) –0.0407 (0.0356) — — — — — —

1.2332∗∗∗ (0.0538) 0.0043 (0.0218) –0.0301 (0.0346) — — — — — —

Manufacturing Yes No No Yes

5561

1.2133∗∗∗ (0.0545) 0.0087 (0.0236) –0.0079 (0.0368) –0.2006 (0.2373) –0.7743∗∗∗ (0.4256) –1.3289∗∗∗ (0.366)

(3)

Manufacturing Yes No Yes Yes

5561

1.2135∗∗∗ (0.0547) 0.0012 (0.024) 0.0003 (0.037) 0.193 (0.3309) –0.3727 (0.4458) –1.1355∗∗∗ (0.3833)

(4)

Manufacturing Yes No Yes Yes

5561

1.2135∗∗∗ (0.0547) — — 0.001 (0.0341) 0.1919 (0.3301) –0.3738 (0.4452) –1.1359∗∗∗ (0.3832)

(5)

Manufacturing Yes No Yes Yes

5561

1.2136∗∗∗ (0.0547) — — — — 0.1897 (0.3208) –0.3743 (0.4449) –1.1359∗∗∗ (0.3832)

(6)

Notes: Standard errors in parentheses. Sample excludes firms experiencing multiple displacement events. All regressions include an indicator for multi-state firms. Workforce characteristics are firm-average race (percentage white), sex (percentage male), and age. ∗∗∗ Significant at the 1 percent level. ∗∗ Significant at the 5 percent level. ∗ Significant at the 10 percent level.

Industry State geography Industry state Workforce characteristics Firm size

Observations

(4) for hij

(2) for hij

(1) for hij

Sales/workers

Cap stock/worker

(2)

(1)

Probability of firm closure conditional on displacement, manufacturing

Displacement occurred

Variable

Table 12.7

Yes Yes No Yes Yes

25,236 Yes Yes No Yes Yes

25,236

1.0056∗∗∗ (0.0234) — — — 0.4208∗∗∗ (0.1393) 0.1948 (0.2446) –0.5236∗∗∗ (0.1804)

1.0029∗∗∗ (0.0234) — –0.0739∗∗∗ (0.0135) 0.2805∗∗ (0.1418) 0.2114 (0.2448) –0.3902 ∗∗ (0.1821)

No No Yes Yes Yes

25,236

1.0081∗∗∗ (0.0234) — –0.0788∗∗∗ (0.0136) 0.2368∗∗∗ (0.1425) 0.176 (0.2456) –0.4158∗∗ (0.1829)

(3)

No No Yes Yes Yes

25,236

1.0105∗∗∗ (0.0234) — — — 0.3889 ∗∗∗ (0.14) 0.1616 (0.2454) –0.5567∗∗∗ (0.1812)

(4)

Yes Yes No Yes Yes

41,066

0.9519∗∗∗ (0.0174) — — — 0.4948∗∗∗ (0.101) 0.5188 ∗∗∗ (0.1751) –0.497∗∗∗ (0.1275)

(5)

No No Yes Yes Yes

41,066

0.9571∗∗∗ (0.0175) — — — 0.4519∗∗∗ (0.1017) 0.4847 ∗∗∗ (0.1763) –0.5525∗∗∗ (0.1283)

(6)

Notes: Standard errors in parentheses. Sample excludes firms experiencing multiple displacement events. Industry eﬀects are at the SIC division level. All regressions include an indicator for multi-state firms. Workforce characteristics are firm-average race (percentage white), sex (percentage male), and age. ∗∗∗ Significant at the 1 percent level. ∗∗ Significant at the 5 percent level. ∗ Significant at the 10 percent level.

Industry State geography Industry state Workforce characteristics Firm size

Observations

(4) for hij

(2) for hij

(1) for hij

Cap stock/worker Sales/workers

(2)

(1)

Probability of firm closure conditional on displacement, all industries

Displacement occurred

Variable

Table 12.8

466

John M. Abowd, Kevin L. McKinney, and Lars Vilhuber

probability of a displacement event. For both manufacturing and all industries, on the other hand, the reduction in the probability of firm closure associated with having a workforce concentrated in the fourth quartile of the human capital distribution is statistically significant and of the same magnitude whether or not the equation controls for the displacement event. 12.4.3 Changes in the Distribution of Human Capital of Survivors In the previous section, we compared employers with and without displacement events, and employers who exited with continuers. In this section, we consider only the continuing employers—those who were active in 1992 and 1997. We compare continuing firms with no displacement events to those with a single displacement event and those with multiple displacement events. We compare how the human capital distribution has changed within continuing employers—contrasting those with one displacement event and those with multiple displacement events to those with no displacement events. The dependent variable is specified as (k) 1997(k) – 1992(k). The regression controls include industry, workforce demographics, state eﬀects, and multi-unit eﬀects. We focus on the results for the first and fourth quartiles (i.e., k 1, 4, the lowest and highest tails of the distribution). Table 12.9 reports results from reduced-form regressions of (k), k 1, 4 on a set of indicators representing single and multiple displacement events. Continuing firms reduced their employment of workers in the lowest quartile of the human capital distribution, as indicated by the negative intercept in the (1) equation. Continuing firms increase their employment of workers in the highest quartile of the human capital distribution, as indicated by the positive intercept in the (4) equation. Both of these results are robust to the inclusion of the full set of controls in the equations. These results are also consistent with other studies that have used these human capital components to study how the the distribution of skill within employer and between employers has changed Abowd, Lengermann, and McKinney (2002). Continuing firms that experienced multiple displacement events are not statistically diﬀerent from continuers that experienced no displacement events once all controls have been entered into the equations. Continuers with a single displacement event are also not statistically diﬀerent from other continuers with regard to the change in their employment of the lowest quartile in the skill distribution once all controls are entered. However, continuer firms that experienced a single displacement event increased their employment of the highest quartile of human capital by substantially less than other continuers. This result is robust to the set of controls entered into the equation. A conclusion consistent with this analysis is that a single displacement event is associated with a significant reduction in the upskilling of the continuing firm’s workforce.

The Link between Human Capital, Mass Layoﬀs, and Firm Deaths Table 12.9 Dependent variable

(1)

(4)

467

Changes in the distribution of human capital within continuing firms: 1992–1997

Variable Intercept Single displacement Multi displacement Controls for SIC Controls for demo Other controls Intercept Single displacement Multi displacement Controls for SIC Controls for demo Other controls Intercept Single displacement Multi displacement Controls for SIC Controls for demo Other controls Intercept Single displacement Multi displacement Controls for SIC Controls for demo Other controls Intercept Single displacement Multi displacement Controls for SIC Controls for demo Other controls Intercept Single displacement Multi displacement Controls for SIC Controls for demo Other controls

Parameter estimate –0.02664 –0.00171 –0.00311

–0.03244 –0.00367 –0.00642

–0.07421 –0.00132 –0.00136

0.00963 –0.02014 –0.00712

0.02379 –0.01674 0.00086058

0.03489 –0.01420 0.00118

Standard error 0.00044287 0.00144 0.00129 No No No 0.00077698 0.00144 0.00133 Yes No No 0.00221 0.00143 0.00133 Yes Yes Yes 0.00062011 0.00202 0.00180

0.00109 0.00201 0.00187 Yes No No 0.00311 0.00201 0.00188 Yes Yes Yes

t value

Pr |t|

–60.15 –1.19 –2.41

.0001 0.2360 0.0158

–41.75 –2.55 –4.82

.0001 0.0107 .0001

–33.65 –0.92 –1.02

.0001 0.3555 0.3055

15.53 –9.98 –3.95 No No No 21.87 –8.32 0.46

.0001 .0001 .0001

11.22 –7.06 0.63

.0001 .0001 0.53029

.0001 .0001 0.6448

Notes: All regressions weighted by the number of employees with 6 h 14. Demographic controls are the fraction of white workers and fraction of male workers. “Other controls” include state eﬀects, indicators for multi-state and multi-unit firms and for firm size. N 39418.

12.5 Conclusion Our analysis can be summarized as follows. Single displacement events occur substantially more often in firms that disproportionately employ workers in the lowest quartile of the human capital distribution. This relation holds in manufacturing and in all industries, whether or not we control for capital/worker or sales/worker, and regardless of the other controls

468

John M. Abowd, Kevin L. McKinney, and Lars Vilhuber

in the equation. Firm closures occur substantially more often in firms that disproportionately employ workers in the lowest quartile of the human capital distribution and disproportionately less often in firms that disproportionately employ workers in the highest quartile of this distribution. These relations hold in manufacturing and in all industries, whether or not we control for capital/worker or sales/worker, and are robust to most changes in the control variables. Once we condition the firm closure analysis on the displacement event, our analysis suggests that firms that disproportionately employ workers in the highest quartile of the skill distribution are less likely to close, even given a displacement event, than are other firms that experience displacement events. In this analysis, however, there does not seem to be an eﬀect of disproportionately employing those in the lowest quartile—suggesting that the eﬀect of employing workers who come disproportionately from the lowest part of the human capital distribution on firm closure works through increasing the probability of a displacement event. Firm or plant survival obviously depends on more variables than the ones included in the analyses presented in this chapter. The age of the firm (Dunne, Klimek, and Roberts 2003; Dunne and Roberts 1990; Freeman and Kleiner 1999), market share and measures of cost (Dunne and Roberts 1990), industry characteristics such as Herfindahl indices or import penetration rates (Freeman and Kleiner 1999), and unionization rates (Abowd 1989; Carneiro and Portugal 2003; Freeman and Kleiner 1999) all appear to matter and have not been included in our analysis. The eﬀect of previous displacement events on the probability of another displacement event also should be studied further. Some, but not all of these issues, can be addressed by incorporating data from the 2002 Economic Census.

Appendix Characterization of Firms with Multiple Displacement Events By our definition, a multiple displacement firm has more than one displacement event during the 16 quarter period beginning in 1993 quarter 1 and ending in 1996 quarter 4. In addition to the multiple displacement events, these firms diﬀer along other dimensions from single displacers; they have a low exit rate (similar to no displacement firms), use relatively low skill workers, and are disproportionately represented in agriculture, retail, and the service sector. This evidence suggests these firms may be responding (not unexpectedly perhaps, given the low exit rate) to cyclical demand factors, an idea we explore further by looking for a high incidence of annual cyclical displacements. Of course other patterns are present in the

The Link between Human Capital, Mass Layoﬀs, and Firm Deaths

469

data, but this approach captures important economic events such as the December holiday season and the seasonal harvesting of agriculture products. More formally, let t 1, . . . , T represent the 16 quarterly time periods in our data. The event space of firm displacement patterns is represented by where at 0 when no displacement has occurred and at 1 when a displacement has taken place. = {(a1, . . . , aT) : at 0, 1}

(A1)

Since the firms have multiple displacement events, not all patterns are present in the data. We must also define S Σat, the sum of all displacement events during the period, which we condition upon in the multiple displacement event space below. {(a1, . . . , aT) : at 0, 1|S ∑at 1}

(A2)

Our approach involves counting the number of annual cyclical patterns 16

(A3) C ∑I([{at 4, . . . , at} 10001] [{at 4, . . . , at 1} 1001]) t5

within each firm’s T-tuple pattern of displacements wj. Given the short time period over which we observe the firms, C ranges from a minimum of zero to a maximum of four. In tables 12A.1 and 12A.2, we compare the observed C for the 7,389 multiple displacement firms with simulated data, where the simulation assumes random generation of displacements without any seasonality. Conceptually the simulation is simple; a string of zeros and ones is created for each simulated firm by randomly drawing 16 values from a uniform [0, 1] distribution with the values equal to or below cons .262 becoming a one (displacement) and those above a zero. The value cons .262

Table 12A.1

Observed seasonal patterns of multiple displacement firms Frequency of seasonal patterns

Number of displacements 2 3 4 5 and more Total

0

1

2

3

4

Total

1,382 65.25 498 34.80 224 18.56 453 17.20 2,557 34.61

736 34.75 592 41.37 332 27.51 840 31.90 2,500 33.83

0 0.00 341 23.83 257 21.29 827 31.41 1,425 19.29

0 0.00 0 0.00 394 32.64 507 19.26 901 12.19

0 0.00 0 0.00 0 0.00 6 0.23 6 0.08

2,118 1,431 1,207 2,633 7,389 100.00

470

John M. Abowd, Kevin L. McKinney, and Lars Vilhuber

Table 12A.2

Simulated seasonal patterns of multiple displacement firms Frequency of seasonal patterns

Number of displacements 2 3 4 5 more Total

0

1

2

3

4

Total

79.97 56.75 39.93 26.10 44.07

20.03 37.15 42.27 40.72 37.31

0.00 6.10 15.59 24.95 14.88

0.00 0.00 2.22 7.60 3.50

0.00 0.00 0.00 0.63 0.25

100.00

is determined from the real data by E [a 1|S Σat 1], where the expectation is taken over the entire 16 quarter period. This method insures that the overall proportion of displacement events is the same in both samples, but depending on the data generating process underlying the real data, the distribution of the T-tuple patterns may diﬀer substantially. Interestingly, the real multiple displacers look quite a bit diﬀerent than the simulated multiple displacers, both in terms of the pattern of total displacements and the frequency of seasonal displacement events. For example, although there is a higher proportion of firms in the real data with two displacement events, there are also more firms with large numbers of displacements (7), thus ensuring that the average number of displacements per firm is the same as the simulated data (by design). The strong annual cyclical nature of the actual data relative to the simulation can be clearly seen by comparing the frequency of seasonal patterns within each row across tables 12A.1 and 12A.2. For example, if we look at firms with three displacement events, almost 24 percent of the firms in the real data have two seasonal patterns, while only a little over six percent have the same pattern in the simulated data. This evidence suggests that seasonal economic factors play an important role in determining the entry to and the likely structure of the multiple displacement category. We also explored whether the frequency of seasonal patterns varied substantially across industry. In general, except for agriculture where the frequency of seasonal patterns was exceptionally strong, the pattern of seasonality looked surprisingly similar across industries. Multiple displacers in manufacturing, for example, tend to have a similar frequency of annual seasonal patterns as do multiple displacers in retail, services, and so forth.

References Abowd, J. M. 1989. The eﬀect of wage bargains on the stock market value of the firm. American Economic Review 79:774–800.

The Link between Human Capital, Mass Layoﬀs, and Firm Deaths

471

Abowd, J. M., P. Corbel, and F. Kramarz. 1999. The entry and exit of workers and the growth of employment: An analysis of French establishments. Review of Economics and Statistics 81 (2): 170–87. Abowd, J. M., R. H. Creecy, and F. Kramarz. 2002. Computing person and firm eﬀects using linked longitudinal employer-employee data. Technical Paper TP2002-06, Longitudinal Employer-Household Dynamics (LEHD), U.S. Census Bureau. Abowd, J. M., and F. Kramarz. 1999. Econometric analysis of linked employeremployee data. Labour Economics 6 (1): 53–74. Abowd, J. M., F. Kramarz, and D. Margolis. 1999. High wage workers and high wage firms. Econometrica 67 (2): 251–333. Abowd, J. M., P. A. Lengermann, and K. L. McKinney. 2002. The measurement of human capital in the U.S. economy. Technical Paper TP-2002-09, Longitudinal Employer-Household Dynamics (LEHD), U.S. Census Bureau. Abowd, J. M., and L. Vilhubert. 2005. The sensitivity of economic statistics to coding errors in personal identifiers. Journal of Business and Economic Statistics 23 (2): 133–52. Anderson, P. and B. Meyer. 1994. The extent and consequences of job turnover. Brookings Papers on Economic Activity, Microeconomics: 177–248. Audretsch, D. B. 1994. Business survival and the decision to exit. Journal of the Economics of Business 1 (1): 125–37. Benedetto, G., J. Haltiwanger, J. Lane, and K. McKinney. 2007. Using worker flows in the analysis of the firm. Journal of Business and Economic Statistics 25 (3): 299–313. Bernard, A. B., and J. B. Jensen. 2002. The deaths of manufacturing plants. NBER Working Paper no. 9026. Cambridge, MA: National Bureau of Economic Research, May. Bowlus, A., and L. Vilhuber. 2002. Displaced workers, early leavers, and reemployment wages. Technical Paper TP-2002-18, Longitudinal EmployerHousehold Dynamics (LEHD), U.S. Census Bureau. Carneiro, A., and P. Portugal. 2003. Wages and the risk of displacement. Working Paper WP 10-03, Banco de Portugal. Davis, J. A., G. Savage, and R. T. Stewart. 2003. Organizational downsizing: A review of literature for planning and research. Journal of Healthcare Management 48 (3): 181–99. Davis, S. J., J. C. Haltiwanger, and S. Schuh. 1996. Job creation and destruction. Cambridge, MA: MIT Press. Dunne, T., S. D. Klimek, and M. J. Roberts. 2003. Entrant experience and plant exit. NBER Working Paper no. 10133. Cambridge, MA: National Bureau of Economic Research, November. Dunne, T., and M. Roberts. 1990. Wages and the risk of plant closings. Discussion Paper 90-6, CES, U.S. Bureau of the Census, Center for Economic Studies, Washington, D. C. Dunne, T., M. J. Roberts, and L. Samuelson. 1988. Patterns of firm entry and exit in U.S. Manufacturing Industries. The RAND Journal of Economics 19 (4): 495–515. Fallick, B. C. 1996. A review of the recent empirical literature on displaced workers. Industrial and Labor Relations Review 50 (1): 5–16. Foster, L., J. Haltiwanger, and C. J. Krizan. 1998. Aggregate productivity growth: Lessons from microeconomic evidence. NBER Working Paper no. 6803. Cambridge, MA: National Bureau of Economic Research, October. ———. 2002. The link between aggregate and micro productivity growth: Evidence from retail trade. NBER Working Paper no. 9120. National Bureau of Economic Research, July.

472

John M. Abowd, Kevin L. McKinney, and Lars Vilhuber

Freeman, R. B., and M. M. Kleiner. 1999. Do unions make enterprises insolvent? Industrial and Labor Relations Review 52 (4): 510–27. Gibbons, R., L. F. Katz, T. Lemieux, and D. Parent. 2005. Comparative advantage, learning, and sectoral wage determination. Journal of Labor Economics 23 (4): 681–724. Grossman, N. 2002. Shrinking the workforce in an economic slowdown. Compensation and Benefits Management 18 (2): 12–23. Haltiwanger, J. C., J. I. Lane, and J. R. Spletzer. 2007. Wages, productivity, and the dynamic interaction of businesses and workers. Labour Economics 14 (3): 575–602. Jacobson, L., R. LaLonde, and D. Sullivan. 1993a. The costs of worker dislocation. Kalamazoo, MI: W. E. Upjohn Institute. ———. 1993b. Earnings losses of displaced workers. American Economic Review 83 (4): 685–709. Kletzer, L. G. 1998. Job displacement. Journal of Economic Perspectives 12 (1): 115– 36. Kuhn, P., and A. Sweetman. 1998. Unemployment insurance and quits in Canada. Canadian Journal of Economics 31 (3): 549–72. Lengermann, P. A., and L. Vilhuber. 2002. Abandoning the sinking ship: The composition of worker flows prior to displacement. Technical Paper TP-2002-11, Longitudinal Employer-Household Dynamics (LEHD), U.S. Census Bureau. Lluis, S. 2005. The role of comparative advantage and learning in wage dynamics and intrafirm mobility: Evidence from Germany. Journal of Labor Economics 23 (4): 725–69. McGuckin, R. H., and S. V. Nguyen. 1995. On productivity and plant ownership change: New evidence from the LRD. The RAND Journal of Economics 26 (2): 257–76. ———. 2001. The impact of ownership changes: a view from labor markets. International Journal of Industrial Organization 19 (5): 739–62. Ruhm, C. J. 1994. Advance notice, job search, and postdisplacement earnings. Journal of Labor Economics 12 (1): 1–28. Schoeni, R. F., and M. Dardia. 1996. Wage losses of displaced workers in the 1990s. Labor and Population Program Working Paper 96-14, RAND Corporation. Stephens, Jr., M. 2002. Worker displacement and the added worker eﬀect. Journal of Labor Economics 20 (3): 504–37. Vilhuber, L. Forthcoming. Adjusting imperfect data: Overview and case studies. In Wage structure, raises, and mobility: International comparisons of the structure of wages within and across firms, NBER, ed. E. Lazear and K. Shaw. Chicago: University of Chicago Press.

13 The Role of Fringe Benefits in Employer and Workforce Dynamics Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

13.1 Introduction A growing literature in economics examines the role of human resource practices in determining firm performance. One strand of this literature has focused on how compensation design can aﬀect employee incentives and through them, a firm’s success. Fringe benefits are an important share of compensation, and economists have long studied the role of pensions and health insurance in labor markets, but have devoted relatively little attention to how decisions about benefits might interact with firm performance. Our aim here is to pull together these issues by examining how providing part of compensation to employees in the form of benefits interacts Anja Decressin is an economist in the U.S. Department of Labor. Tomeka Hill is a senior research associate at Watson Wyatt Worldwide. Kristin McCue is an economist at the U.S. Census Bureau. Martha Stinson is an economist at the U.S. Census Bureau. The authors wish to acknowledge the substantial contributions of LEHD staﬀ, in particular Paul Lengermann, Kevin McKinney, and Kristin Sandusky. We wish to thank John Abowd, Dan Beller, Keenan Dworak-Fisher, Julia Lane, Joe Piacentini, participants at the pre-conference, and especially our discussant, Dan Black, for their comments. This document reports the results of research and analysis undertaken by the U.S. Census Bureau staﬀ. It is released to inform interested parties of ongoing research and to encourage discussion. This research is a part of the Census Bureau’s Longitudinal Employer-Household Dynamics Program (LEHD), which is partially supported by the National Science Foundation Grants SES-9978093 and SES-0427889 to Cornell University (Cornell Institute for Social and Economic Research), the National Institute on Aging Grant R01 AG018854, and the Alfred P. Sloan Foundation. The views expressed herein are those of the authors and not necessarily those of the Department of Labor, or of the Census Bureau, its program sponsors, or data providers. Some of the data used in this chapter are confidential data from the LEHD Program. The Census Bureau supports external researchers’ use of these data through the Research Data Centers (see http://www.ces.census.gov/). For other questions regarding the data, please contact Jeremy S. Wu, Program Manager, Assistant Division Chief, U.S. Census Bureau. ([email protected], http://lehd.did.census.gov/).

473

474

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

with a firm’s human capital stock, and further, how this choice is related to firm productivity, growth, and survival. Clearly, whether and what sort of benefits to oﬀer is not given exogenously, but is a decision that will depend in part on firm performance. Our goal is not to establish causality, but rather to map out correlations among these variables as a building block for future work in this area. We approach these issues using a very rich data set that combines administrative data on benefit plans from IRS Form 5500 with the Longitudinal Employer-Household Dynamics (LEHD) integrated employeremployee data. These data have many unique characteristics that allow us to explore how firm productivity, growth, and survival are related to benefit oﬀering in a way that has not been possible before. Each data source provides a piece of the overall picture. The data from IRS Form 5500 provide us with the means to identify which employers oﬀer benefit packages, as well as some information about the type of benefits oﬀered. These public filings cover most types of tax-preferred benefits, enabling us to characterize the benefit oﬀerings of a much broader group of employers than would be possible with survey data. The firm-level data, drawn from the Census Business Register and State ES-202 data, report other firm characteristics such as firm size, industry, number of establishments, and a longitudinal measure of firm life span. The worker-level data, drawn from State Unemployment Insurance (UI) wage records, inform us about which workers are employed by which firms in a quarter and give us some basic demographic characteristics of those workers. These links between workers and firms allow us to follow the movement of workers between firms and in and out of the labor market over time. These longitudinal links provide us with measures of human capital and firm compensation practices that are of key interest to questions about benefit oﬀering. They also give us measures of the number of workers joining and leaving each employer’s workforce, providing us with more detailed data on the amount of employee turnover than is available from either firm- or worker-level data alone. Employers’ compensation choices are driven both by the characteristics of their businesses and the workforce that they choose to employ. By integrating these three diﬀerent types of data—benefit oﬀerings, firm characteristics, and worker flows—we are able to examine the whole picture. Because the data sources we draw on cover populations rather than samples, we can use a very rich combination of measures while retaining a reasonably large and representative sample—something that is very diﬃcult to obtain when combining data from several surveys. We use these data to measure how firm choices about compensation packages are correlated with a firm’s workforce turnover and growth and how both of these are in turn correlated with firm outcomes.

The Role of Fringe Benefits in Employer and Workforce Dynamics

475

The chapter proceeds as follows. In section 13.2 we briefly discuss the literature that we build on and present a framework for thinking about our estimation. We discuss data issues in some detail in section 13.3, and then present results in section 13.4. Section 13.5 summarizes our results. 13.2 Background As background for our empirical work, we briefly review related work on human resource practices and firm performance to give a context for our approach. Following that we provide a framework for our estimation. 13.2.1 How do Firms Use Benefits to Shape Their Human Capital Stock? In hedonic wage models, workers face a continuum of diﬀerent compensation packages given by the envelope of firms’ varying wage/benefit isoprofit lines.1 In response to this set of choices they sort themselves into diﬀerent types of firms. Variation in workers’ willingness to trade oﬀ wages for benefits leads to sorting of workers into firms on the basis of fringe benefit oﬀerings. In this model, sorting matches workers with their preferred compensation package and minimizes employers’ costs of employing labor. Workers vary in their willingness to trade oﬀ wages for benefits because of diﬀerences in factors such as marginal tax rates, age, and rates of time preference. Firms that need skills held mainly by young workers, for example, might make themselves more attractive to the employees they wish to hire and keep by oﬀering higher wages and no benefits, if young workers have a lower willingness to trade oﬀ cash wages for fringes. The relative cost of fringe benefits may vary across firms because of economies of scale in providing benefits or varying access to particular types of fringe plans. For example, large firms may have lower costs of providing benefits because of the cost advantages of pooling across a large group of employees. As Montgomery and Shaw (1992) point out, any direct productivityenhancing eﬀects of benefits will also alter the firm side of this trade oﬀ, increasing the cost eﬀectiveness of fringes relative to wages. Alternatively, dual labor market theory posits that there are two sectors: one with rationed good jobs that pay well and have good fringe benefits, and a second with bad jobs having low pay and few benefits (Bulow and Summers 1986; Dickens and Lang 1985). Sorting in this model need not reflect diﬀerences in the costs of providing benefits, nor in willingness to trade oﬀ wages for fringes. Similarly, eﬃciency wage models also generate an equilibrium in which workers with the same productive characteristics would have jobs with diﬀerent levels of compensation that might include diﬀerent benefits. 1. For example, as in the standard textbook by Ehrenberg and Smith (1996, p. 247).

476

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

A long literature has documented that both pensions and health insurance are associated with a workforce with lower turnover. Economists have found a negative relationship between pensions and quit rates for both defined benefit (DB) and defined contribution (DC) plans.2 In the case of DB plans, implicit contract theory has been the primary framework used to interpret this pattern: a loss of pension wealth penalizes workers who break their implicit contract by leaving prior to retirement. This compensation structure leads to self-selection so that firms oﬀering pensions end up with a workforce made up of stayers, which is what motivates oﬀering the DB plan. Ippolito (2002) oﬀers an alternative explanation. One problem for implicit contract theory has been the finding that quit rates are low for firms oﬀering DC plans as well as those oﬀering DB plans, despite the fact that DC plans impose much smaller quitting costs (Gustman and Steinmeier 1993 and 1995; Even and Macpherson 1996; Ippolito 2002). Ippolito argues that quit rates are low because pensions in general attract savers, and that those who save at a higher rate also have lower quit propensities. His 2002 paper expands on earlier work (Ippolito 1997) that argues that having a low discount rate makes both saving and staying more attractive, so pensions are one method of attracting those with low discount rates who may also have higher productivity. He presents evidence that those with characteristics that might be correlated with a low discount rate are more likely to have a pension, and are also more likely to have high performance ratings. An alternative explanation for low quit rates under pension plans is that firms with pensions have higher total compensation than firms without, and that the diﬀerence in compensation accounts for lower turnover (Gustman and Steinmeier 1995; Even and Macpherson 2001). Gustman and Steinmeier (1995) point out that compensation diﬀerences would help explain some other puzzles as well—why the reduction in turnover associated with pensions is largest for young workers, for whom the associated pension losses are small, and why the reduction appears to occur primarily through fewer layoﬀs rather than through fewer quits. If benefits play an important role in firms’ compensation strategies, there ought to be measurable eﬀects on firm outcomes such as productivity, firm growth, and survival. Such eﬀects may come about indirectly through changes in the recruitment and retention of labor. But benefits may also aﬀect productivity more directly by altering employees’ incentives to invest in firm-specific knowledge or by reducing turnover and training costs (Even and Macpherson 2001). There is little existing empirical evidence on the relationship between benefits oﬀering and productivity. One exception is work by Dorsey, Cornwell, and Macpherson (1998) using Compustat data to estimate eﬀects of DB plans on productivity us2. See, for example, the review in Gustman, Mitchell, and Steinmeier (1994).

The Role of Fringe Benefits in Employer and Workforce Dynamics

477

ing a production function framework. They find evidence of higher labor productivity in firms with DB plans, but the evidence on overall productivity eﬀects is mixed. 13.2.2 Schematic Framework for Our Empirical Work Our starting point is to posit that productivity is a function of a firm’s human capital stock (HC ) and other inputs (Z ). (1)

lnYit f(HCit, Zit) εit.

We assume that the composition of compensation aﬀects productivity through its eﬀects on the human capital stock. A traditional representation of the evolution of a capital stock is useful in considering the mechanisms through which this might happen: (2)

HCit (1 it)HCi,t1 Iit

where HCit represents the human capital stock of employer i at time t, and Iit represents investment. Investment in human capital happens through accessions (new hires) and on-the-job acquisition of skill by current employees. Depreciation occurs both through employee separations and through deterioration or obsolescence of employees’ skills. Because we cannot measure on-the-job investment or the deterioration of skills, we focus on how compensation practices aﬀect accessions and separations. We do so by running a series of regressions to examine diﬀerences in churning (turnover in excess of that needed to grow or shrink a firm) and net employment growth for benefit- and non-benefit-providing employers. If benefit provision is correlated with lower churning and net employment growth or stability, this will support the findings of previous studies that benefits are associated with lower turnover. We then come back to equation (1), and estimate the correlation between output and changes in HC by running a series of productivity regressions. Our ability to measure the other inputs, Z, is quite limited. Outside of manufacturing, we have no direct measures of other inputs, so we rely on controls for industry, location, and size to capture some of these eﬀects. We are in somewhat better shape in manufacturing where we can at least measure capital-labor ratios and the use of materials, so we present some of our productivity results for manufacturing alone. A positive correlation between benefits and productivity will support the hypothesis that firms are choosing benefit packages to attract and retain high quality workers, or at least workers whose skills closely match the firm’s needs. Finally, using the data we have developed on a firm’s human capital stock, productivity, and benefits, we look at what relationship fringe benefit compensation has with a longer run firm outcome—the likelihood of firm failure. We estimate hazards of firm death to examine how the human capital stock, productivity, and fringe benefit compensation are correlated with

478

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

firm survival. A positive correlation between longer firm life and benefit provision will be consistent with the hypothesis that firms gain real advantage from choosing a compensation package that involves pensions and health insurance, though it could also be that expected firm success makes employers more willing to oﬀer benefits. 13.3 Data Our estimates are based on a very rich database that we created by combining data from several sources. It consists of microdata on whether a business provided benefits, detail on the types of benefits provided, workforce composition, turnover, the distribution of worker earnings, and labor productivity. Much of this data is also longitudinal, allowing us to measure changes in benefit oﬀerings, and firm survival and growth. 13.3.1 Dataset Construction The database combines information from five sources:

• Firm reports on benefit plans oﬀered to employees (the Internal Rev• • • •

enue Service/Department of Labor Form 5500 file) The Census Bureau’s Business Register (BR) Unemployment Insurance (UI) wage record data from seven states The Census NUMIDENT file The Economic Census

Benefit information comes from Form 5500 annual reports on employee benefit plans filed by the plan sponsor (usually the employer). These public filings are required under ERISA for most types of tax-preferred benefits, with some exemptions for small health plans. Here we use data on plans that end in 1997 and 2001, drawing from the 1996/1997 and 2000/2001 data files. The Form 5500 collects information about employer-provided pensions (defined benefit and various types of defined contribution plans), welfare plans (health, life, supplemental unemployment, and disability insurance plans) and fringe benefit plans (cafeteria or flexible benefit plans and educational assistance plans).3 In addition to variables describing plan features, the data also include name, address, and a federal Employer Identification Number (EIN) for the plan sponsor. Employer Identification Numbers (EINs) are also used in a wide variety of other employer tax filings, including those underlying the Census Bureau’s business list and the UI wage record data. Figure 13.1 describes how the data set is constructed. The 5500 file is first integrated with Census’s Business Register (BR) using sponsor EINs. The BR is a list of all private establishments with paid employment that is con3. See appendix table 13A.1 for a listing of the benefit plan types and the associated frequencies among 1997 plans in the 5500 files.

The Role of Fringe Benefits in Employer and Workforce Dynamics

Fig. 13.1

479

Schemata for construction of database

structed from a variety of administrative and survey sources, but its backbone is quarterly employment tax filings that include EINs.4 Census uses the quinquennial economic censuses and the annual Company Organization Survey in constructing the BR to break out diﬀerent business locations that may be filed under a single EIN. Many large firms use more than one EIN, so these survey sources, in combination with administrative data, also identify EINs that are aﬃliated through parent-subsidiary relationships. We do an initial match of the list of 5500 EINs to the BR. If a 5500 EIN matches to part of a multi-location firm, we use information on company structure from the BR to identify any other EINs (and aﬃliated establishments) that belong to the same company. One diﬃcult question is whether a particular benefit is in fact oﬀered at all establishments belonging to a company. Here, we treat all parts of a company as oﬀering benefits if at least one EIN belonging to that company matches to the 5500 file. Our next step is to bring in the UI data. These data are described extensively elsewhere (Burgess, Lane and Stevens 2000), but we note several salient characteristics here. First, they include longitudinal data on both firms and workers from the mid-1990s to 2003, which permits an analysis of the dynamics of employment flows, workforce change, and firm entry and exit over this period. Second, because earnings data are available for individual workers at each of their employing firms, it is possible to analyze both earnings and employment outcomes for workers in each business. Finally, the data are almost universal in nature, capturing some 98 percent of employment in each state for which the data are collected. The results presented here include data for seven states. Although the UI wage record data are very rich in terms of sample size 4. The BR was historically known as the Standard Statistical Establishment List (SSEL). An establishment is defined as a single physical location where business is conducted or where services or industrial operations are performed.

480

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

and coverage, they lack demographic information on workers. Limited demographic information is obtained by matching the UI records with internal administrative records (the Census NUMIDENT file) that have information on date of birth, place of birth, race and sex for all workers. About 96 percent of the records in each state’s UI wage data can be matched to this source.5 In addition, we make use of human capital measures constructed by other LEHD researchers. The LEHD staﬀ (as described in Abowd, Lengermann, and McKinney [2002], henceforth ALM) have estimated fixed eﬀects for individual firms and workers based on the following wage equation: (3)

ln (wijt ) xit i ψj (i,t) εijt

where i is a fixed worker eﬀect, ψj(i,t) is a fixed firm eﬀect, and j(i,t) indexes the firm j for which worker i works at time t. Ln (wijt ) represents the log of full-time earnings, so the fixed eﬀects i and ψj (i,t) are in terms of log earnings diﬀerences. This decomposition provides a measure of the fixed, portable component of worker i ’s skills (i), and a measure of the fixed premium (or discount) that firm j pays after accounting for worker skills (ψj (i,t)). In our empirical results we use the following as a measure of general human capital: (4)

hit xit i

where xit consists of quarters of work experience interacted with race and gender. As described in ALM, a seven-state distribution of hit was created and individual workers were classified according to their location in this distribution. Summary-level statistics for firms were created by calculating the percentage of workers at each firm that belonged to each quartile of the overall human-capital distribution. In this work, we have these measures for 1997 only. Finally, we bring in measures of labor productivity based on data from the 1997 Economic Census (EC). We measure labor productivity as the logarithm of sales per employee deviated from the two-digit industry mean. For multi-unit firms, we aggregate establishment-level data from the EC to the state-EIN-two-digit SIC level before matching to the 5500/UI data. For each multi-unit firm we define a primary SIC by aggregating payroll across establishments within a state that have the same two-digit SIC code, and then taking the SIC code associated with the largest aggregated payroll.6 5. See Staﬀ of the LEHD Program (2002) for further discussion. 6. Because at the time of writing data were not yet available from the 2002 Economic Census, we do not have longitudinal data on productivity.

The Role of Fringe Benefits in Employer and Workforce Dynamics

481

13.3.2 Data Coverage Issues—5500/Business Register Match In the results that follow, we use the presence of a matching record in the 5500 file as an indicator that a firm oﬀers benefits and then use additional information from the file to determine what sorts of benefits are oﬀered. Whether these are reasonably accurate measures depends first on the filing requirements for the Form 5500 (e.g., do all plans in fact appear in this data set?) and secondly on our success in matching an employer to the 5500 data set when it in fact contains information on the employer’s benefit plan(s). We briefly describe these issues here; they are more fully documented in Decressin et al. (2005). Filing requirements diﬀer somewhat for pensions and other types of plans (welfare or fringe benefit plans). For pensions, only a few specialized types of plans are exempt from the requirement to file.7 However, welfare and fringe plans with fewer than 100 participants are exempt if they are either unfunded (i.e., the employer pays the costs out of general funds) and/ or fully insured through an insurance provider (e.g., a Blue Cross/Blue Shield company). Thus, most small health plans are probably not included in the 5500 file. Where we focus on benefits in general, the firms for whom we mismeasure benefit oﬀering are those who oﬀer a small health plan but no pension plan. For most plans, employer and sponsor are one and the same, and integration of 5500 data with the BR is straightforward. However, for plans that involve multiple employers the sponsor EIN generally belongs to an entity other than one of the participating employers. For example, a trade union might sponsor a Taft-Hartley pension plan for unionized electricians working for many diﬀerent employers. Given that we cannot identify which employers are involved with our current resources, we drop those plans in what follows. Some employers oﬀer more than one benefit plan, so we summarize plan-level information associated with the same EIN before matching to the BR.8 Ninety-seven percent of the 731,609 EINs in our 1997 Form 5500 extract can be found on the 1997 BR. Limiting our analysis to records in the BR that meet our criteria for being active and in-scope gives us a match rate of 90 percent for the 5500 EINs in 1997 and 89 percent for the EINs in 2001.9 The fraction of firms on the BR that match to a plan in the 5500 file 7. Simplified Employee Pension (SEP) plans are exempt, as are Savings Incentive Match Plans for Employees (SIMPLE) if they take the form of an IRA (but not SIMPLE 401(k) plans). Both plans can be used only by employers with at most 100 eligible employees. SEP plans do not allow for employee contributions, and employer contributions must be a fixed percentage of pay up to a maximum. Church plans are also exempt. 8. Thirty percent of EINs in 1997 are associated with two or more plan filings. 9. The BR records that we exclude from matching either do not report any payroll for the current year or have codes that indicate that they should be outside the scope of our investigation (e.g., they are government-owned entities, which are not required to file Form 5500, or

482

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

Table 13.1

Business register match rates to 1997, 5500 data Single-unit firms

Multi-unit firms

All firms

Number of employees

Number

Match rate (%)

Number

Match rate (%)

Number

Match rate (%)

Missing or 0 1–4 5–99 100–999 1,000

955,116 2,624,082 1,918,184 44,708 1,130

1.5 4.1 19.5 62.3 74.3

2,880 7,728 139,724 44,166 7,074

16.6 15.4 47.2 78.5 92.6

957,996 2,631,810 2,057,908 88,874 8,204

1.5 4.1 21.3 70.4 90.1

Total

5,543,220

9.5

201,572

54.0

5,744,792

11.0

Notes: Employment data on the BR come primarily from filings of IRS Form 941, which is used to report quarterly withholding of payroll and income taxes. Businesses are also asked to report employment on these forms, but the employment data are not as complete as the payroll data. We include only firms with positive payroll in the table (and the match), but some of these firms do not report any employment. This could be because the employment question asks about a particular week in the quarter and the firm had no employees on the payroll that week, or it may be that firms neglected to report employment, which is not directly tied to the tax liability. Employment data used in the estimation section is taken from the UI data.

is much lower. As tables 13.1 and 13.2 illustrates in 1997, only 11 percent (9.4 percent in 2001) of the 5.7 million businesses in the 1997 Census Business Register have a match to a 5500 form, but the vast majority of companies that do not match to the Form 5500 data are in fact very small. The match rates in 2001 are similar to the ones in 1997, but usually a little lower.10 Because we have a wider variety of data available in 1997 than in 2001 and we would like to be as consistent as possible across tables, in the following most statistics are only presented for 1997. The low overall match rate simply reflects the predominance of firms with few employees in the overall count of firms. Of the nonmatches in 1997, 54 percent have five or fewer employees, and an additional 23 percent have between six and twenty-five employees. Large firms (≥ 100 employees) account for only 0.5 percent of all nonmatches compared to 13 percent of all matches. Larger firms are more likely to oﬀer benefits and are also more likely to be required to file Form 5500, given that they oﬀer plans. Thus, it is encouraging that the majority of large firms in the Business Register can be matched to a Form 5500 filing. Because of filing exemptions and diﬃculties represent a trust rather than an employer). The 11 percent of EINs in 1997 that match to these sorts of BR records might match to adjacent years of the BR, or may provide information on what sorts of plans we do not accurately match—both possibilities we plan to investigate in future work. Extensive documentation of the matching exercise is provided in Decressin, McCue, and Stinson (2003). 10. The data for the 1997 match did receive a much bigger data cleaning eﬀort than the 2001 data and this might have caused better match numbers for 1997.

The Role of Fringe Benefits in Employer and Workforce Dynamics

483

in matching, we expect coverage to be incomplete for small firms and in industries with large numbers of Taft-Hartley plans. In some of what follows we present results for manufacturing alone (which has little Taft-Hartley coverage) or for the sample of firms with at least 100 employees. If we weight these match rates by employment (see table 13A.2), firms in 1997 that match to the 5500 file employ 66 percent of all workers (70 million out of the 106 million in the 1997 workforce). The corresponding number for 2001 is 64 percent. Among multi-units the match rate is even higher: 90 percent of workers in 1997 are employed by matching firms (89 percent in 2001), compared with 36 percent for single-unit firms (32 percent in 2001). Table 13.3 gives the distribution of type of plans among firms that match to at least one plan in 1997. A very large share of firms with some sort of matched benefit oﬀer a pension plan, regardless of size. This probably reflects the fact that coverage of benefits by Form 5500 filings is most complete for pensions. The most notable size eﬀect is for health plans, for which the percent oﬀering a health plan is substantially larger for firms with greater than 100 employees. This appears to reflect the exemption from filing for most health plans with fewer than 100 enrollees. Comparing benefit coverage rates implied by our matched data to national survey estimates suggests that we do quite well in matching pension Table 13.2

Business Register Match Rates to 2001, 5500 Data Single-unit firms

Multi-unit firms

All firms

Number of employees

Number

Match rate (%)

Number

Match rate (%)

Number

Match rate (%)

Missing or 0 1–4 5 – 99 100 – 999 1000

929,468 2,686,090 2,051,445 53,590 1,495

1.3% 3.3% 16.6% 54.9% 62.1%

3,117 6,052 101,799 40,807 7,312

18.7% 15.5% 43.0% 74.4% 90.1%

932,585 2,692,142 2,153,244 94,397 8,807

1.37% 3.36% 17.81% 63.32% 85.35%

Total

5,722,088

8.2%

159,087

51.7%

5,881,175

9.42%

Table 13.3

Types of plans found for matched employers, by firm size—1997

Number of employees

Pension (%)

Health (%)

Other fringe (%)

1–50 51–100 101–250 251–750 750

87 85 86 88 92

2 7 28 57 80

24 52 55 59 72

Total

87

7

30

484

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

coverage but that we understate health coverage by even more than we would have expected. In our data for 1997, 61 percent of employees work for businesses that oﬀer pension benefits, while 34 percent work for businesses that oﬀer health benefits. BLS survey estimates from 2003 indicate that 57 percent of employees have access to retirement benefits. This figure excludes employees who have not met minimum length of service requirements and so would be expected to be somewhat below our estimate, which implicitly includes them.11 In contrast, a 1997 survey estimated that 86 percent of employees work for establishments that oﬀer health benefits—more than twice our figure.12 We expect to understate health coverage somewhat given that certain small health plans are not required to file, but the diﬀerence seems too large for that to be the only problem. At the same time, we seem to find too much coverage under plans classified by sponsors as Fringe benefit plans on the Form 5500. This should include only Section 125 cafeteria plans (flexible benefit, reimbursement, and premium conversion plans) and non-jobrelated education benefit plans (under Section 127 of the tax code). We find that 36 percent of employees in 1997 work for firms oﬀering plans classified in this way, while 1999 survey estimates imply that 28 percent of employees have access to Section 125 plans and only 10 percent have access to nonwork related educational assistance (and presumably there is considerable overlap in those types of benefits). Because we think that some health insurance plans may appear in the 5500 files as Fringe benefit plans, we start with estimates based on the more general question of whether an employer oﬀers some form of fringe benefits, but we also examine how the estimates change when we look at a breakout of benefit types. 13.3.3 Sample Characteristics While we match the 5500 data to Census’s Business Register as a whole, most of our empirical work is based on the subset of those businesses for which we also have UI State data. In looking at employment growth we must also restrict our sample to firms continuing from 1997 to 2001. Before proceeding with our results, we briefly describe the diﬀerences in samples. While the 5500/BR data exists for all fifty-one states, the UI data restrict us to look at seven states. Thus, we have data for parts of firms that operate across multiple states, and no data for firms that operate only in other states, which leaves us as much as 1.8 million firms operating in 1997 and 11. It is diﬃcult to put together a comparable total coverage number for pension benefits for either 1997 or 2001 for a couple of reasons. Prior to 1999, the BLS survey that collected data for benefits estimates (the Employee Benefit Survey) surveyed small and medium/large employers in alternating years. Beginning in 1999, BLS publishes estimates for all private employers, but access numbers are not available until 2003. The 2003 estimates are available at http://www.bls.gov/ncs/ebs/sp/ebbl0020.pdf 12. See http://www.meps.ahrq.gov/mepsweb/data_stats/summ_tables/insr/national/series_1/ 1997/tib2.pdf

The Role of Fringe Benefits in Employer and Workforce Dynamics

485

1.2 million firms operating in both 1997 and 2001 (Continuers).13 As explained in ALM, human-capital summary statistics are only created for firms with at least five employees, due to the diﬃculty of applying kernel density estimation techniques for calculating distributions to firms of very small sizes. Our regressions all make use of these wage decompositions, so we further restrict our sample to firms having at least five employees. This reduces the sample for 1997 to almost 400,000 and for the 1997/2001 Continuers sample to a little more than 300,000. In addition, we do some of our productivity analyses on the subset of firms that also appear in the Annual Sample of Manufacturing (ASM) from which we derive measures of capital intensity. For 1997, this sample contained 10,263 firms. Table 13.4 presents 1997 sample statistics for the seven-state 5500/BR/ UI data and the sample for which we also have HC estimates.14 Given that more than 50 percent of firms in the United States have less than five employees, the reduction in sample size from column 1 to 3 is dramatic. As expected, the sample without the small firms has on average older firms and a higher percentage of benefit-oﬀering firms. The fraction of multi-unit firms is also higher. The statistics for 1997/2001 continuers are fairly similar to the more comprehensive samples. The average firm labor productivity and mean firm wage eﬀect for the sample of continuers (column [4]) is larger than for the sample that includes all the observations in 1997 (column [3]). The last row gives mean churning rates, which measure the rate of accessions and separations that occur at a firm over one quarter, above and beyond those needed to accomplish the firm’s net growth during that period.15 Net growth is defined as total employment at the end of a quarter minus total employment at the beginning of the quarter and can be positive or negative. For firms with human capital estimates (i.e., at least five employees), the churning rates are similar for the 1997 firms and the continuers. For the entire sample, churning rates are lower for the continuers. 13.4 Results 13.4.1 Evolution of the Human Capital Stock A firm that grows by one employee over a quarter may do so by simply hiring one more employee, or by hiring five new employees and letting four employees (new or old) go. The latter is likely to be more costly, but also 13. The unit of observation is generally a firm/state record; that is, a multi-unit firm that operates in several of the states for which we have data will have more than one record. 14. Table 13A.3 of the Appendix shows firm characteristics by benefit oﬀer for the sample. 15. The formula used is (|A S| – |E – B|) / [(B E ) / 2], where A accessions, S separations, B employment at the beginning of the quarter, and E employment at the end of the quarter.

486 Table 13.4

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson Characteristics of alternative samples Data required

5,500/BR/UI data

5,500/BR/UI data and HC estimates

Sample characteristics

All 1997

Continuers

All 1997

Continuers

Sample size

1,860,072

1,196,541

390,635

302,079

5.44 % 1.94 % 9.68 % — 3.31 %

4.29 % 2.83 % 13.76 % — 3.84 %

10.52 % 3.53 % 21.27 % –0.1049 7.74 %

7.76 % 4.44 % 26.61 % –0.0966 7.89 %

31.88 % 19.74 % 13.58 % 8.89 % 14.11 %

28.15 % 20.77 % 15.33 % 10.50 % 16.75 %

21.64 % 21.47 % 17.49 % 13.95 % 25.45 %

18.66 % 20.84 % 18.04 % 14.93 % 27.53 %

56.21 % 41.12 % 2.45 % 0.22 %

50.35 % 46.43 % 2.95 % 0.27 %

— 92.18 % 7.21 % 0.61 %

— 92.07 % 7.31 % 0.61 %

— —

— —

27.10 25.75

26.92 % 25.88 %

70.97 % 46.37 % 19.04 % 71.65 %

71.57 % 46.14 % 18.70 % 72.23 %

71.08 % 47.77 % 16.78 % 71.26 %

71.38 % 47.57 % 16.64 % 71.82 %

— 0.4063

— 0.3289

0.0461 0.2886

0.0809 0.2682

Oﬀer benefits in 1997 only Oﬀer benefits in 2001 only Oﬀer benefits in 1997 and 2001 Mean firm wage eﬀect ψ Multi-unit firm Firm age: 5 years old 5 to 10 years old 10 to 15 years old 15 to 20 years old 20 or more years old Firm size: 5 employees 5 to 99 employees 100 to 999 employees 1000 employees HC quartiles: lowest highest Workers: % prime age % female % foreign born % white, non-Hispanic Relative labor productivity: deviation of log (productivity) from 2-digit SIC mean Churning rate (1st quarter, 1997)

Notes: The observations are on an EIN/state level. All variables are measured in 1997, except where noted otherwise.

makes possible more dramatic changes in the total skills embodied in the firm’s workforce—for better or worse. While we do not do a formal decomposition, we do break our analysis into these two parts: churning and net growth. If a firm follows the first strategy of hiring and keeping one more person, then net growth is one person and churning is zero. If instead it hires five new employees and lets four go, churning is 8 ( 5 4 – 1)— the number of extra employees who arrived or left above those needed to increase employment by one. We are interested in how both net growth and churning relate to benefit oﬀering and how these two correlations might be diﬀerent from each other.

The Role of Fringe Benefits in Employer and Workforce Dynamics

487

Churning Regressions Existing work, based primarily on household data, has established that there is a strong negative association between rates of employee turnover and benefits. We present churning regressions here in part to confirm that this is true in our firm-based data as well, but also because our data allow us to address several unanswered questions. One advantage we have is that including the firm wage premium allows us to control for diﬀerences in other parts of firm compensation schemes. We also have richer measures of worker skill and more accurate measures of employer characteristics than are typically available from household survey data. Table 13.5 presents the results from our analysis of churning rates for the first quarter of 1997.16 We exclude observations with churning rates of greater than 2 as likely to be subject to measurement error. Column (1) includes a dummy for whether or not the firm oﬀered benefits in 1997 along with controls for the following firm characteristics: the wage premium or firm wage eﬀect (ψ), size, age, industry, state, and workforce demographic characteristics. In column (2) we add human capital characteristics, while columns (3) and (4) repeat those two specifications, but with more detailed benefit dummies. In all specifications, benefits in general are negatively related to churning rates. In column (1), firms that currently oﬀer benefits have about a 3 percentage point lower churning rate (in a sample with a mean churning rate of 28.9 percent). In column (2), adding human capital characteristics substantially reduces the size of the benefit coeﬃcient, though it remains negative and significant. Thus, a substantial part of the negative association between benefits and churning is because businesses that employ workers with higher human capital levels are more likely to oﬀer benefits and higher human capital levels are associated with reduced churning. Including dummies for specific benefit types in columns (3) and (4) in place of the general benefits dummy shows that the negative correlation with churning is common to all types of pensions plans and to health plans, but that DB and other (non-DC) pension plans have the most substantial and consistently negative associations with churning. A comparison of (3) and (4) shows that the negative association is reduced for all types of plans when the human capital measures are included, implying that an important component of the correlation found in (3) for all benefits is that more skilled (and higher paid) employees are more likely to be oﬀered benefits. We find somewhat odd results for firm size: churning rates are significantly higher for the 100 to 999 employee and 1000 employee groups of firms than for the smallest category, though the largest category has slightly lower rates than the middle category. Note that we are controlling 16. We have estimated analogous regressions using churning measures from the 1st quarter of 2001, and get very similar results so we have not presented them here.

488

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

Table 13.5

Churning regressions: 1st quarter of 1997

Independent variables

(1)

(2)

(3)

(4)

Oﬀered benefits in 1997

–.0279 (.0013) — — — — — — — — — — –.1029 (.0016) .0299 (.0022) –.0703 (.0016) –.0990 (.0017) –.1206 (.0018) –.1371 (.0016) .0913 (.0022) .0797 (.0068) — — — —

–.0045 (.0012) — — — — — — — — — — –.0842 (.0017) .0173 (.0021) –.0662 (.0015) –.0914 (.0016) –.1120 (.0018) –.1312 (.0016) .0724 (.0021) .0651 (.0066) .1367 (.0038) –.1598 (.0038)

— — –.0287 (.0033) –.0246 (.0014) –.0400 (.0061) –.0165 (.0028) –.0056 (.0018) –.1018 (.0016) .0313 (.0022) –.0705 (.0016) –.0993 (.0017) –.1204 (.0018) –.1356 (.0016) .0970 (.0023) .0947 (.0069) — — — —

— — –.0185 (.0032) –.0031 (.0014) –.0218 (.0059) –.0083 (.0027) .0026 (.0017) –.0838 (.0017) .0178 (.0021) –.0664 (.0015) –.0915 (.0016) –.1118 (.0018) –.1304 (.0016) .0752 (.0022) .0729 (.0068) .1363 (.0038) –.1594 (.0038)

.150

.196

0.151

0.196

Oﬀered DB pension plan in 1997 Oﬀered DC pension plan in 1997 Oﬀered other pension in 1997 Oﬀered health plan in 1997 Oﬀered other fringe plan in 1997 Firm wage eﬀect (ψ) Multi-unit firm Firm age 5 to 10 Firm age 10 to 15 Firm age 15 to 20 Firm age 20 or more Firm size 100 to 999 employees Firm size 1000 employees % in lowest HC quartile % in highest HC quartile R-squared

Notes: N 390,635. Dependent variable is the churning rate for the first quarter of 1997, as defined in footnote 15; mean .289. Sample is 1997 firms with churning rates of less than 2. Regressions also include controls for state, two-digit industry, and worker demographics. All controls measured in 1997.

for firm age here, and the age eﬀects are quite large, so it is not the case that large, well-established firms have higher churning rates than young, small firms. The fact that we are looking at churning over a very short period of time (one quarter) may contribute to this result, as small firms are much more likely to have a churning rate of zero in a quarter than is a firm with 1,000 employees. One interesting result is that the inclusion of the human capital variables reduces the coeﬃcient for DC plans much more than the coeﬃcient for DB

The Role of Fringe Benefits in Employer and Workforce Dynamics

489

plans. This suggests that selection on the human capital characteristics that are consistently valued across employers (i.e., those captured by the individual fixed eﬀects) is more important in explaining the lower turnover associated with DC plans than the turnover associated with DB plans. These results fit nicely with the pension literature’s emphasis on the incentive eﬀects of DB plans—high costs imposed on separations lead to reduced turnover—and with the observation that the similarly low turnover rates for DC plans probably have a diﬀerent cause, namely selection eﬀects. Firm Growth In table 13.6 we examine how firm growth is correlated with benefit provision. The sample consists of firms that were in existence in both 1997 and 2001. We regress the log diﬀerence in employment on benefits dummies, while controlling for characteristics of the firm and its workforce. In column (1) we simply control for whether a firm oﬀers benefits in 1997, while in (2) we distinguish whether or not benefits are oﬀered in 1997 only, in 2001 only, or in 1997 and 2001, leaving firms that oﬀered benefits in neither year as the omitted category. In columns (3) and (4) we repeat these specifications with the addition of a control for labor productivity. Overall, the continuers’ sample has a negative growth rate: employment shrinks on average by about 4 percentage points over the period. Our regression results in columns (1) and (3) show that firms that oﬀered benefits in 1997 grew more over the subsequent period than those that did not, conditional on surviving to 2001. Firms that oﬀered benefits in both years grew substantially faster than those that oﬀered them in neither: their growth rates were about 11 percentage points greater than the omitted category and that diﬀerence is significant. Changes in benefits are also correlated with firm growth, though it seems more likely that firms’ choices to change their benefits were motivated by their growth experience than the converse. Firms that dropped benefits between 1997 and 2001 had growth rates that were on average about 3 percentage points lower than those of firms that did not oﬀer benefits in either year in both columns (2) and (4). Firms that added benefits had growth rates that were about 25 percentage points higher than those for firms that oﬀered benefits in neither year. The other coeﬃcients make clear that older firms and larger firms, unsurprisingly, grow more slowly. Firms with high churning rates also tend to grow more slowly. One might expect that, other characteristics constant, firms that are growing quickly would have high churning rates, as increasing the number of employees requires increasing the proportion with low tenure, and low tenure employees tend to have higher turnover rates. However, high turnover might also hamper desired firm growth. The negative coeﬃcient suggests that the latter is likely the predominate eﬀect. The coeﬃcients on the human capital measures indicate that employing many less-skilled workers is associated with lower growth rates. The coeﬃ-

490

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

Table 13.6

Firm employment growth regressions

Independent variables

(1)

(2)

(3)

(4)

Oﬀered benefits in 1997

.0470 (.0026) — — — — — — .1303 (.0035) –.1917 (.0044) –.0724 (.0034) –.0989 (.0035) –.1181 (.0038) –.1381 (.0034) –.0030 (.0045) –.0286 (.0140) –.1550 (.0082) .0675 (.0079) –.0641 (.0036) — —

— — –.0259 (.0041) .2612 (.0052) .1098 (.0029) .1098 (.0035) –.1975 (.0044) –.0701 (.0034) –.0970 (.0035) –.1180 (.0037) –.1410 (.0034) –.0162 (.0045) –.0390 (.0139) –.1372 (.0082) .0461 (.0079) –0.0675 (.0035) — —

.0372 (.0026) — — — — — — .0722 (.0038) –.1928 (.0044) –.0741 (.0034) –.1006 (.0035) –.1190 (.0038) –.1392 (.0034) –.0042 (.0045) –.0325 (.0139) –.1134 (.0083) .0018 (.0081) –.0534 (.0036) .0698 (.0017)

— — –.0305 (.0041) .2538 (.0052) .0982 (.0029) .0566 (.0038) –.1984 (.0044) –.0717 (.0034) –.0987 (.0035) –.1188 (.0037) –.1419 (.0034) –.0168 (.0045) –.0421 (.0139) –.0990 (.0082) –.0140 (.0080) –.0575 (.0035) .0647 (.0017)

0.0377

0.0485

.0430

.0531

Oﬀered benefits in 1997 only Oﬀered benefits in 2001 only Oﬀered benefits in both years Firm wage eﬀect (ψ) Multi-unit firm Firm age 5 to 10 Firm age 10 to 15 Firm age 15 to 20 Firm age 20 or more Firm size 100 to 999 employees Firm size 1000 employees % in lowest HC quartile % in highest HC quartile Churning rate (1st quarter, 1997) Relative labor productivity (log) R-squared

Notes: N 302,079. Dependent variable is the log diﬀerence in employment: ln(Emp2001) – (ln(Emp1997); mean –.04. Sample is all firms continuing from 1997 to 2001 with churning rates 2. The regressions also include controls for state, two-digit industry, and worker demographic characteristics in 1997.

cient on the upper quartile measure is also consistent with a positive relationship between average human capital and firm growth, but the coeﬃcients are much smaller. When we add the productivity measure in (3) and (4) this coeﬃcient becomes insignificant and, in column (4), even negative so it appears that having high human capital employees is associated with greater firm growth only because it is associated with higher productivity. Once that is controlled for, the association is insignificant or negative, perhaps because of their cost. The firm wage eﬀect has a strong positive relationship with growth

The Role of Fringe Benefits in Employer and Workforce Dynamics

491

rates—firms that paid 10 percent more than the average had about 1.3 percentage point higher growth in column (1). This coeﬃcient is also sensitive to the inclusion of productivity in the regressions. It is substantially smaller in columns (3) and (4), but remains positive. So higher paying firms grew faster over this period, in part because high productivity firms were growing faster, and high productivity and the firm component of pay are positively correlated. For table 13.7 we substitute more detailed benefit dummies for the genTable 13.7

Firm employment growth regressions with type of plan controls

Independent variables Oﬀered DB pension plan in 1997 Oﬀered DC pension plan in 1997 Oﬀered other pension in 1997 Oﬀered health plan in 1997 Oﬀered other fringe plan in 1997 Firm wage eﬀect (ψ) Multi-unit firm Firm age 5 to 10 Firm age 10 to 15 Firm age 15 to 20 Firm age 20 or more Firm size 100 to 999 employees Firm size 1000 employees % in lowest HC quartile % in highest HC quartile Churning rate (1st quarter, 1997) Relative labor productivity (log) R-squared

(1)

(2)

–.0619 (.0065) .0388 (.0028) .0852 (.0119) .0162 (.0057) .0323 (.0035) .1311 (.0035) –.1920 (.0044) –.0723 (.0034) –.0987 (.0035) –.1173 (.0038) –.1368 (.0034) –.0077 (.0047) –.0314 (.0143) –.1542 (.0082) .0703 (.0079) –.0643 (.0036) — —

–.0758 (.0065) .0290 (.0028) .0787 (.0118) .0083 (.0057) .0319 (.0035) .0728 (.0038) –.1924 (.0044) –.0741 (.0034) –.1005 (.0035) –.1180 (.0038) –.1372 (.0034) –.0064 (.0047) –.0286 (.0142) –.1125 (.0083) .0044 (.0081) –.0537 (.0036) .0707 (.0017)

.038

.044

Notes: N 302,079. Dependent variable is the log diﬀerence in employment; mean –.04. Sample is all firms continuing from 1997 to 2001. The regressions include controls for state, two-digit SIC, and worker demographics. All control variables are measured as of 1997.

492

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

eral benefit variable in table 13.6, columns (1) and (3). The coeﬃcients on the control variables that are common to the two tables are very similar in the two tables. However, there are some striking diﬀerences in coeﬃcients across diﬀerent types of benefits. Firms that oﬀered DB pension plans had employment losses that were significantly larger than those for firms without benefits. DB pensions plans are most typically oﬀered by larger, older employers in more established parts of the economy (e.g., manufacturing), but we find this result even after controlling for age, size, and two-digit industry. Thus, oﬀering a DB plan may be correlated with being in shrinking parts of the economy even within industry and age class. The other types of benefit plans are all positively associated with employment growth, with the largest coeﬃcients for ‘other’ pension plans. Health plans have a positive but relatively small coeﬃcient in column (1), which becomes insignificantly diﬀerent from zero in column (2), where we control for productivity diﬀerences. 13.4.2 Relationship to Productivity At this point, we do not have changes in productivity to use to directly examine the relationship between post-1997 productivity growth and benefit oﬀering in 1997, so we instead look at the cross-sectional relationship. In table 13.8, the dependent variable is the log of labor productivity deviated from a two-digit industry-specific mean. The unit of observation is generally a firm/state record; that is, a multi-unit firm that operates in several of the states for which we have data will have more than one record. Some of the variables included are defined for the firm as a whole (whether or not benefits are oﬀered, firm size, multi-unit status, and firm age), while the others are generally measured within state.17 Our primary interest is in the coeﬃcient on the benefits indicator, which is positive and significant in each of our specifications. We recognize that several of our right-hand variables are likely endogenous, so the results should be interpreted simply as establishing correlations rather than causality. Because the dependent variable is a log diﬀerence, the coeﬃcient on this variable in column 1 indicates that the productivity of firms that oﬀer benefits is on average .32 log points (or roughly 38 percent) higher than that of nonbenefit-oﬀering firms with similar characteristics. Given that benefits 17. The level of aggregation is important only for multi-unit firms that have diverse operations within a state. Because of the computational resources needed to estimate the wage decomposition, our decisions about how to handle aggregation issues are in part driven by the availability of estimates originally generated for other projects. The labor productivity and capital measures are calculated at a state/EIN/two-digit SIC level, and then a single two-digit SIC record is selected (if more than one exists) by taking the record with the highest payroll. The demographic and churning variables were handled similarly, except that the original measures were calculated at a state/EIN level. The human capital and firm eﬀect variables are calculated at the state/EIN/two-digit SIC level. For these measures we use the observation with the highest employment because payroll is not available in these files.

The Role of Fringe Benefits in Employer and Workforce Dynamics Table 13.8

493

1997 Productivity diﬀerentials associated with benefit oﬀering

Independent variables

(1)

(2)

(3)

(4)

(5)

Oﬀers benefits in 1997

.3218 (.0028) — — .0111 (.0048) .0342 (.0035) .0517 (.0037) .0647 (.0040) .0867 (.0036) .0157 (.0048) .1102 (.0151) — — — — — — — —

.2572 (.0027) — — .0252 (.0046) .0269 (.0034) .0341 (.0036) .0374 (.0039) .0501 (.0035) .0400 (.0047) .1079 (.0146) –.5360 (.0084) .8372 (.0082) — — — —

.1488 (.0025) .8277 (.0034) .0214 (.0043) .0385 (.0031) .0415 (.0033) .0349 (.0036) .0345 (.0033) .0152 (.0044) .0612 (.0135) –.6111 (.0078) .9208 (.0077) –.1670 (.0033) — —

.0454 (.0144) 1.0185 (.0321) .0437 (.0187) .1916 (.0257) .1687 (.0241) .1492 (.0248) .1579 (.0197) .1191 (.0158) .1998 (.0345) –.7937 (.0594) .5223 (.0530) –.1286 (.0271) — —

.0481 (.0137) .8279 (.0310) .0269 (.0177) .1732 (.0244) .1250 (.0229) .1050 (.0236) .0973 (.0188) .0881 (.0150) .1242 (.0328) –.6171 (.0566) .4046 (.0504) –.1184 (.0257) .1899 (.0057)

All 0.131

All 0.187

All 0.302

ASM 0.306

ASM 0.374

Firm wage eﬀect (ψ) Multi-unit firm Firm age 5 to 10 Firm age 10 to 15 Firm age 15 to 20 Firm age 20 or more Firm size 100 to 999 employees Firm size 1000 employees % in lowest HC quartile % in highest HC quartile of Churning rate (1st quarter, 1997) Log(capital intensity) Sample R-squared

Notes: N 390,635 for overall sample; N 10,263 for ASM sample. The dependent variable is the deviation of the log of labor productivity from two-digit SIC mean; mean .046 for overall sample, .331 for ASM sample. All columns also include controls for two-digit SIC, state, and worker demographics.

are costly, it would be diﬃcult for benefit-oﬀering firms to stay competitive if they did not have higher labor productivity, so this is reassuring. In column (2), we add controls for the firm’s distribution of worker human capital. The first human capital variable gives the percent of the workforce with a worker fixed-eﬀect in the bottom quartile of the distribution; the second gives the percent in the top quartile. Both of these variables have large, significant coeﬃcients of the expected sign—productivity rises with the fraction of workers in the more skilled parts of the distribution. Including these controls reduces the benefits coeﬃcient by about .06 log points, indicating that about a fifth of the association between productivity and benefits found in the first column is explained by this fairly simple

494

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

characterization of worker human capital. In all columns we control for diﬀerences in productivity associated with industry, firm size, firm age, state, and worker demographics.18 The coeﬃcients on the firm size indicators are positive, significant, and generally increasing with size in the overall sample. Firms that are more than five years old have about 3 to 9 percent higher productivity in the overall sample, and about 10 to 19 percent higher productivity in manufacturing, but the diﬀerences between age groups among those older than five are generally not significant in manufacturing. In the third column, we add the firm wage eﬀect and a measure of workforce churning as additional controls. Across specifications, the churning measure has a negative coeﬃcient that is usually significant, but whether or not it is included has little eﬀect on the benefits coeﬃcient.19 Adding the firm wage eﬀect, however, reduces the benefits coeﬃcient substantially. While firms that oﬀer benefits have higher productivity, they also have compensation policies that pay what appear to be equivalent workers more than they receive in other jobs, and this component of pay has a very strong positive relationship to productivity, even when controlling for workforce composition. The fourth and fifth columns present results for the subset of our overall sample that is included in the 1997 Annual Survey of Manufactures. This subsample is of interest because we can construct measures of capital that are not available outside manufacturing.20 Column (4) includes the same controls as column 3—it is included to illustrate changes in coeﬃcients that are simply a result of the change in sample. The benefits coeﬃcient is smaller in manufacturing, but in general the results do not look radically diﬀerent. Adding our measure of capital—the log of capital per worker, based on the book value of capital divided by employment—has only a small eﬀect on the benefits coeﬃcient. Interestingly, it does reduce the size of the coeﬃcients on the human capital, firm wage eﬀect, and most demographic variables, so greater capital intensity does appear to account for some of the positive association found between worker skill and productivity. Table 13.9 reports the results of the productivity regressions with detailed benefit type dummies in place of the overall dummy. While for the 18. We do not report the worker demographic coeﬃcients in the tables to conserve space. 19. The order in which we introduce the human capital and firm eﬀect controls has little impact on the portion of the reduction in the benefits coeﬃcient that we attribute to the diﬀerent controls. 20. We can construct capital measures for a larger sample of manufacturing firms by also including those that are in the 1997 Economic Census of Manufacturing (CM) but not in the ASM, which adds a lot of smaller firms. However, the ASM sample is asked more detailed capital questions, and imputation is used for some components of capital in non-ASM cases. We have run the same sets of regressions for both manufacturing samples, and while the coeﬃcients are somewhat diﬀerent, the general conclusions we draw are not.

The Role of Fringe Benefits in Employer and Workforce Dynamics Table 13.9

1997 Productivity diﬀerentials associated with diﬀerent types of benefit oﬀering

Independent variables Oﬀered DB pension plan in 1997 Oﬀered DC pension plan in 1997 Oﬀered other pension in 1997 Oﬀered health plan in 1997 Oﬀered other fringe plan in 1997 Firm wage eﬀect (ψ) Multi-unit firm Firm age 5 to 10 Firm age 10 to 15 Firm age 15 to 20 Firm age 20 or more Firm size 100 to 999 employees Firm size 1000 employees % in lowest HC quartile of % in highest HC quartile Churning rate (1st quarter, 1997) Log(capital intensity) Sample R-squared

495

(1)

(2)

(3)

(4)

(5)

.2981 (.0072) .2976 (.0030) .2044 (.0135) .1348 (.0062) .0556 (.0039) — — –.0032 (.0048) .0371 (.0035) .0555 (.0037) .0637 (.0040) .0738 (.0036) –.0293 (.0050) –.0150 (.0153) — — — — — — — —

.2608 (.0070) .2367 (.0030) .1510 (.0131) .1372 (.0060) .0418 (.0037) — — .0116 (.0046) .0294 (.0033) .0372 (.0036) .0361 (.0039) .0380 (.0035) –.0042 (.0049) –.0088 (.0148) –.5314 (.0084) .8294 (.0082) — — — —

.2011 (.0065) .1434 (.0028) .0905 (.0121) .1070 (.0056) .0071 (.0035) .8207 (.0034) .0113 (.0043) .0399 (.0031) .0430 (.0033) .0327 (.0036) .0242 (.0033) –.0180 (.0045) –.0253 (.0137) –.6067 (.0078) .9125 (.0077) –.1654 (.0033) — —

.0976 (.0211) .0223 (.0153) .0529 (.0764) .0336 (.0187) –.0220 (.0157) 1.0009 (.0322) .0327 (.0188) .1998 (0.256) .1826 (.0241) .1630 (.0247) .1588 (.0196) .1059 (.0166) .1682 (.0355) –.7864 (.0594) .5185 (.0530) –.1258 (.0271) — —

.0699 (.0201) .0282 (.0146) .0324 (.0726) .0160 (.0178) –.0160 (.0149) .8187 (.0311) .0197 (.0179) .1802 (.0244) .1371 (.0229) .1175 (.0235) .1013 (.0187) .0826 (.0157) .1067 (.0338) –.6154 (.0567) .4042 (.0505) –.1171 (.0257) .1884 (.0057)

All 0.137

All 0.192

All 0.306

ASM 0.308

ASM 0.375

Notes: N 390,635 for overall sample; N 10,263 for ASM sample. The dependent variable is the deviation of the log of labor productivity from two-digit SIC mean; mean .046 for overall sample, .331 for ASM sample. All columns also include controls for two-digit SIC, state, and worker demographics.

employment growth regressions the coeﬃcients on the other control variables were little changed when we switched to detailed benefits controls, here we see one notable change between the two tables: the coeﬃcients on the size dummies in the overall sample go from positive and significant to negative, though only occasionally significant. This seems likely due to larger, more productive firms in the overall sample oﬀering more than one

496

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

type of benefit or being more likely to oﬀer the types of benefits most highly correlated with productivity. The pension coeﬃcients in table 13.9 are all positive and significant. Adding the human capital controls and then the churning rate and firm wage eﬀect each reduces the size of the pension coeﬃcients somewhat, but they all remain quite large in the general sample. Each type of pension is positively correlated with employee human capital and firm pay practices, and negatively correlated with turnover, which account for some but not all of their positive association with productivity. But it is interesting that the decreases are larger for the non-DB pension plans than for DB plans. As was the case with table 13.8, the estimated diﬀerences are smaller in manufacturing but they remain positive. The coeﬃcients for oﬀering health and fringe plans are positive and significant for the overall sample, but smaller than the pension coeﬃcients. In manufacturing, the health plan coeﬃcient loses its significance and the other fringe plan coeﬃcient becomes negative, though not significantly so. 13.4.3 Hazard Estimates of Firm Failure Do firms that currently oﬀer benefits have higher future survival rates, conditioning on other observable characteristics? Why might there be such a relationship? Including benefits as part of compensation is correlated with having a more stable workforce, and a more stable workforce may reduce the likelihood that a firm goes out of business. However, if benefits are of more value to employees when they expect their current employer to stay in business, or if an employer’s gain to oﬀering benefits accrues over a long period of time, then employers who are less likely to stay in business may also be less likely to oﬀer benefits. At this point, little is known about whether such a relationship even exists, so our goal here is simply to describe the empirical relationship rather than to establish causality. We do so using our complete sample of firms in existence in 1997 and exploiting their rates of failure over the 1997 to 2003 period. A firm is measured as having failed if it stops filing the UI records that underlie our human capital estimates. To examine this relationship, we use a Cox proportional hazard model to estimate the probability of a firm failing in the years after 1997, conditional on surviving until 1997, and include a dummy variable for having oﬀered benefits in 1997.21 In addition to benefits, in all specifications we also include controls for firm age, industry, state, multi-unit status, firm size, and workforce demographics. We measure all of these explanatory variables as of 1997. While alternative functional forms could be used here—for example, a probit—the hazard model takes advantage of 21. By using a conditional probability function, the Cox proportional hazard method controls for left truncation/delayed entry (firms are first observed in 1997, but were already in business before). Only observations from 1997 and onward are used in calculating the loglikelihood function.

The Role of Fringe Benefits in Employer and Workforce Dynamics

497

variation in the timing of failure as well as whether or not a failure occurred at some point in the six year period for which we have data. We present four sets of estimates: tables 13.10 and 13.11 give results for the overall sample, first using the general benefits dummy (13.10) and then using dummies for plan type (13.11). In tables 13.12 and 13.13, we restrict our sample to firms with at least 100 employees. Recall that for some types of benefits (primarily health plans) sponsors are not required to file a Form 5500 if the plan has fewer than 100 enrollees, and thus we measure benefits coverage less accurately for smaller firms, which are excluded in these tables. The specifications (2), (3), and (4) for each table diﬀer from (1) in that we progressively add the human capital quartile measures and the firm wage eﬀect (ψ), the churning measure, and then labor productivity. In all specifications in tables 13.10 and 13.11, we find a significant nega-

Table 13.10

Hazard estimates of the relationship between firm death and benefit oﬀering: All firms, 1997–2003

Independent variables

(1)

(2)

(3)

(4)

Oﬀers benefit in 1997

–.2315 (.0077) — — –.0227 (.0128) –.4595 (.0118) –.8975 (.0189) –1.4074 (.0266) –1.6950 (.0321) .1985 (.0125) .1926 (.0403) — — — — — — — —

–.2084 (.0079) .0448 (.0098 –.0288 (.0128) –.4579 (.0118) –.8942 (.0189) –1.4021 (.0266) –1.6912 (.0321) .1817 (.0126) .1839 (.0403) .0706 (.0213) –.4806 (.0235) — — — —

–.2068 (.0079) .0747 (.0100) –.0314 (.0127) –.4504 (.0118) –.8834 (.0189) –1.3883 (.0266) –1.6744 (.0321) .1559 (.0125) .1599 (.0404) .0326 (.0215) –.4314 (.0236) .2838 (.0083) — —

–.1853 (.0079) .2009 (.0110) –.0281 (.0127) –.4492 (.0118) –.8835 (.0189) –1.3910 (.0266) –1.6719 (.0321) .1550 (.0126) .1663 (.0405) –.0598 (.0218) –.2973 (.0241) .2591 (.0084) –.1482 (.0051)

Firm wage eﬀect (ψ) Multi-unit firm Firm age 5 to 10 Firm age 10 to 15 Firm age 15 to 20 Firm age 20 or more Firm size 100 to 999 employees Firm size 1000 employees % in lowest HC quartile % in highest HC quartile Churning rate (1st quarter, 1997) Relative labor productivity (log)

Notes: Number of EIN/State observations 389,051. Number of failures 107,652. Hazards also include controls for state, two-digit SIC, and workforce demographic characteristics. All controls are measured in 1997.

498

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

Table 13.11

Hazard estimates with plan type controls: All firms, 1997–2003

Independent variables Oﬀered DB pension plan in 1997 Oﬀered DC pension plan in 1997 Oﬀered other pension in 1997 Oﬀered health plan in 1997 Oﬀered other fringe plan in 1997 Firm wage eﬀect (ψ) Multi-unit firm Firm age 5 to 10 Firm age 10 to 15 Firm age 15 to 20 Firm age 20 or more Firm size 100 to 999 employees Firm size 1000 employees % in lowest HC quartile % in highest HC quartile Churning rate (1st quarter, 1997) Relative labor productivity (log)

(1)

(2)

(3)

(4)

–.0888 (.0215) –.2465 (.0086) –.4586 (.0490) .3268 (.0167) –.0471 (.0108) — — –.0352 (.0128) –.4615 (.0118) –.9011 (.0189) –1.4119 (.0266) –1.7193 (.0322) .1260 (.0133) .0842 (.0415) — — — — — — — —

–.0750 (.0216) –.2213 (.0088) –.4400 (.0490) .3232 (.0167) –.0431 (.0108) .0353 (.0098) –.0416 (.0128) –.4599 (.0118) –.8977 (.0189) –1.4065 (.0266) –1.7161 (.0321) .1105 (.0133) .0755 (.0415) .0838 (.0214) –.4807 (.0236) — — — —

–.0686 (.0216) –.2208 (.0088) –.4317 (.0489) .3283 (.0167) –.0432 (.0108) .0651 (.0100) –.0447 (.0128) –.4523 (.0118) –.8869 (.0189) –1.3928 (.0266) –1.7003 (.0321) .0844 (.0133) .0485 (.0414) .0458 (.0216) –.4314 (.0237) .2853 (.0083) — —

–.0383 (.0216) –.1989 (.0088) –.4220 (.0489) .3410 (.0167) –.0409 (.0108) .1931 (.0110) –.0425 (.0128) –.4511 (.0118) –.8871 (.0189) –1.396 (.0266) –1.7002 (.0322) .0797 (.0133) .0431 (.0416) –.0478 (.0241) –.2958 (.0241) .2603 (.0084) –.1514 (.0051)

Notes: Number of EIN/State observations 388,814. Number of failures 107,583. Hazards also include controls for state, two-digit SIC, and workforce demographics. All controls are measured in 1997.

tive relationship between the likelihood of post-1997 firm failure and the provision of benefits, with the notable exception of health insurance. The coeﬃcients on the overall benefit dummy in table 13.10 indicate that the hazard of firm death is on average roughly 20 percent lower for firms that provide benefits, with the size of the relationship somewhat smaller in specifications with additional controls. In 13.11 where we examine diﬀerent types of benefits we find similarly sized coeﬃcients for DC pension plans, and even larger ones for other (i.e., non-DC/non-DB) pensions. The DB

The Role of Fringe Benefits in Employer and Workforce Dynamics Table 13.12

499

Hazard estimates of firm death and benefit oﬀering: Firms with 100 or more employees, 1997–2003

Independent variables

(1)

(2)

(3)

(4)

Oﬀers benefit in 1997

–.2708 (.0234) — — –.0089 (.0247) –.3748 (.0617) –.7543 (.0863) –1.4061 (.1108) –1.6064 (.1189) –.0807 (.0425) — — — — — — — —

–.2887 (.0236) .4665 (.0631) –.0105 (.0247) –.3730 (.0617) –.7511 (.0863) –1.4056 (.1108) –1.6151 (.1190) –.0985 (.0429) .0607 (.1215) .1464 (.1051) — — — —

–.2820 (.0236) .5348 (.0647) –.0085 (.0247) –.3656 (.0618) –.7353 (.0863) –1.3852 (.1109) –1.5912 (.1191) –.1005 (.0429) –.0223 (.1228) .1554 (.1052) .2419 (.0394) — —

–.2808 (.0236) .6778 (.0704) –.0035 (.0248) –.3607 (.0617) –.7304 (.0863) –1.3838 (.1109) –1.5895 (.1190) –.0977 (.0429) –.1030 (.1233) .2299 (.1057) .2168 (.0397) –.1092 (.0177)

Firm wage eﬀect (ψ) Multi-unit firm Firm age 5 to 10 Firm age 10 to 15 Firm age 15 to 20 Firm age 20 or more Firm size 1000 employees % in lowest HC quartile % in highest HC quartile Churning rate (1st quarter, 1997) Relative labor productivity (log)

Notes: Number of EIN/State observations 30,191. Number of failures 8,498. Hazards also include controls for state, two-digit SIC, and workforce demographic characteristics. All controls are measured in 1997.

plans and other fringe plans have smaller but still negative and generally significant coeﬃcients. Health plans, however, have large positive coeﬃcients, indicating that on average firms that are matched to a health plan have about a 33 percent higher failure rate than similar firms that do not match to such a plan. In table 13.12, where we restrict our sample to firms with 100 or more employees, the overall benefit dummy coeﬃcients remain negative and significant, and increase slightly in magnitude. Similar comments apply to the coeﬃcients on detailed benefit types in table 13.13—there are no changes in sign and the coeﬃcients often (though not always) increase in magnitude, but again with the notable exception of health plans. The health plan coeﬃcients remain positive, but drop dramatically in size to imply only a five to six percent higher hazard rate for firms that oﬀer health plans. Thus, it is mostly among small firms that we find dramatically higher failure rates for firms that match to a Form 5500 health plan. Particularly with health

500

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

Table 13.13

Hazard estimates with plan type controls: Firms with 100 employees, 1997–2003

Independent variables Oﬀered DB pension plan in 1997 Oﬀered DC pension plan in 1997 Oﬀered other pension in 1997 Oﬀered health plan in 1997 Oﬀered other fringe plan in 1997 Firm wage eﬀect (ψ) Multi-unit firm Firm age 5 to 10 Firm age 10 to 15 Firm age 15 to 20 Firm age 20 or more Firm size 1000 employees % in lowest HC quartile % in highest HC quartile Churning rate (1st quarter, 1997) Relative labor productivity (log)

(1)

(2)

(3)

(4)

–.1113 (.0384) –.2037 (.0259) –.5753 (.0955) .0633 (.0273) –.1164 (.0251) — — –.0090 (.0248) –.3865 (.0618) –.7653 (.0864) –1.4137 (.1110) –1.6020 (.1192) –.0593 (.0431) — — — — — — — —

–.1311 (.0386) –.2128 (.0259) –.5774 (.0955) .0479 (.0274) –.1190 (.0251) .4715 (.0639) –.0088 (.0248) –.3848 (.0618) –.7625 (.0865) –1.4131 (.1110) –1.6080 (.1192) –.0729 (.0434) .0587 (.1217) .1806 (.1053) — — — —

–.1309 (.0386) –.2112 (.0260) –.5716 (.0956) .0539 (.0275) –.1179 (.0251) .5403 (.0654) –.0071 (.0248) –.3772 (.0619) –.7469 (.0865) –1.3927 (.1111) –1.5845 (.1193) –.0758 (.0434) –.0255 (.1229) .1896 (.1054) .2453 (.0395) — —

–.1199 (.0387) –.2117 (.0260) –.5705 (.0955) .0549 (.0275) –.1167 (.0251) .6764 (.0710) –.0029 (.0248) –.3724 (.0618) –.7417 (.0865) –1.3908 (.1111) –1.5836 (.1192) –.0745 (.0434) –.1014 (.1234) .2587 (.1059) .2215 (.0397) –.1047 (.0178)

Notes: Number of EIN/State observations 30,129. Number of failures 8,476. Hazards also include controls for state, two-digit SIC, and workforce demographics. All controls are measured in 1997.

plans, the diﬀerence in results between the two samples may be an artifact of filing rules. Recall that for small health plans, only plans that are at least partially self-insured must file. While the majority of self-insured plans use stop-loss coverage to limit their financial risk, it must still be a fairly risky strategy for small firms to forgo full insurance when providing health benefits. If firms that choose risky strategies in providing benefits are also more likely to take risks in other areas, it may not be surprising that they would be more likely to fail, particularly when the comparison group must include some firms that oﬀer health insurance but fully insure.

The Role of Fringe Benefits in Employer and Workforce Dynamics

501

Still, even among the large sample, we find a positive association between health insurance oﬀering and failure, conditioning on all of the other variables in the regressions. This indicates that it is not only diﬀerences in filing requirements that make health insurance appear so diﬀerent from the other types of benefits. We can only hypothesize why this might be so. One possibility is that firms have more flexibility in oﬀering pension plans (at least non-DB plans) when they are doing well and dropping them when business is poor because employees care less about the consistency of current pension coverage than of health coverage. A very generous plan this year and none the next may be equivalent to a meager plan both years to the average worker because they provide the same stream of future consumption. The same is not true for health plans, which may make it more diﬃcult for firms to cut health insurance in lean times, and might in turn lead to a much more negative coeﬃcient in hazard regressions for pension plans than for health plans. However, it is not clear that this could generate a positive coeﬃcient for health plans, which is in fact what we find. Unsurprisingly, firm age has a large and consistently negative association with the likelihood of failure, but our size eﬀects are puzzling. We actually find lower rates of failure for firms with fewer than 100 employees than for larger firms in the specifications in tables 13.10 and 13.11, though the diﬀerences between medium sized (100 to 999) and very large employers (1000) look more reasonable. The largest firms have either similar (in table 13.10) or lower rates of failure (tables 13.11, 13.12, and 13.13), depending on the specification. Note that the diﬀerences in hazard rates are much larger for age than for size. If we do not include age controls, the size coeﬃcients follow the expected pattern—they are negative and increase in magnitude with size. The human capital measures have less consistent eﬀects. In the overall sample, with either specification of benefits controls, firms with more skilled workers generally have higher survival rates, though adding productivity as a control reduces the size of the eﬀects. However, in the sample with only larger firms, the eﬀects are not significant and the coeﬃcients vary somewhat in sign. With churning and labor productivity controls in the hazard, the coeﬃcient on the upper quartile of the human capital distribution in 13.13 is positive and marginally significant, implying a higher rate of failure among firms that employ a higher proportion of skilled workers, perhaps because of their cost. The firm wage eﬀect, which captures firm pay diﬀerentials, has a positive eﬀect in all specifications, but for the overall sample the eﬀect is quite small unless we control for labor productivity. Interestingly, the diﬀerence in the size of the coeﬃcients on the firm wage eﬀect in columns (2) and (3) between tables 13.10, 13.11, 13.12, and 13.13 suggests that it is only among small firms that those with higher average pay have high enough labor productivity that they are less likely to fail. For both samples, controlling for

502

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

labor productivity results in a large positive coeﬃcient on the firm wage eﬀect: holding productivity, turnover, and workforce characteristics constant, higher average pay is associated with higher rates of failure. Reassuringly, labor productivity has a strong negative correlation with the probability of failure wherever we include it, while higher churning rates are positively related to failure rates, as we would expect. 13.5 Summary Using Form 5500 data combined with LEHD integrated UI data for 1997 to 2003, we find that firms that oﬀer benefits have lower turnover rates and grow faster than the average non-benefit-oﬀering firm. We examine diﬀerent types of benefits, and find the same general pattern for most types of plans, with the exception of defined benefit pension plans for which we find lower turnover but slower employment growth. Controlling for workforce human capital characteristics reduces the estimated eﬀects but does not eliminate them. Firms that add benefits over this period have particularly high rates of employment growth, suggesting that significant employment growth may be a factor in the decision to oﬀer benefits for firms that do not already do so. In our analysis of productivity diﬀerences across firms, we find that both benefits and the firm-specific component of pay are positively related to productivity. We also find that firms that oﬀer benefits are less likely to fail—even after controlling for other observable characteristics—than are firms that do not oﬀer benefits. Many interpretations could be put on this. One is that of endogeneity—firms that are more likely to die (either due to current financial problems, or perhaps because they are an inherently more risky business) are less likely to oﬀer benefits. This could either be as a way to cut down on current costs, or because workers value benefits less when the risk of future default is higher. In general, our findings verify that there is a correlation between a firm’s decision to oﬀer benefits and the mobility and productivity of its labor force as well as the firm’s length of life. While our results do not confirm a causal relationship between benefits and firm outcomes, they do highlight the importance of considering both benefits and wages when studying the labor market decisions made by firms and workers.

The Role of Fringe Benefits in Employer and Workforce Dynamics

503

Appendix Table 13A.1

Distribution of 5500 plan records by exclusive type

Exclusive benefit plan types

Number

Defined benefit plan Defined contribution plan Other pension plan Health plan Fringe benefit plan Welfare and fringe benefit plan Welfare and pension benefit plan No info on plan benefit type Total

Table 13A.2

Percent (%)

64,313 657,324 24,916 65,333 208,469 42,851 2,915 5,899

6.00 61.30 2.30 6.10 19.40 4.00 0.30 0.60

1,072,020

100.00

Business register match rates, weighted by employment—1997 and 2001 Single units

Multi-units

All Firms

Number of employees

Total employment

Match rate (%)

Total employment

Match rate (%)

Total employment

Match rate (%)

1 to 4 5 to 99 100 to 999 1000 Total

5,500,708 28,665,996 8,987,671 2,366,799 45,521,174

5.2 29.5 64.1 69.5 35.5

A. 1997 22,244 4,590,771 12,237,687 43,297,297 60,147,999

16.2 57.1 82.0 95.5 89.8

5,522,952 33,256,767 21,225,358 45,664,096 105,669,173

5.2 33.3 74.4 94.1 66.4

1 to 4 5 to 99 100 to 999 1000 Total

5,604,584 31,793,403 10,855,284 3,829,941 52,083,212

4.2 25.3 56.9 53.1 31.6

B. 2001 17,668 3,560,992 11,596,094 51,039,833 66,214,587

16.0 51.7 78.1 94.5 89.3

5,622,252 35,354,395 22,451,378 54,869,774 118,297,799

4.2 27.9 67.9 91.6 63.9

504

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

Table 13A.3

Firm characteristics by benefit oﬀer—1997 Benefit-providing firms

Non-benefit-providing firms

Sample size

124,188

266,447

Mean firm wage eﬀect ψ Multi-unit firm Firm age: 5 years old 5 to 10 years old 10 to 15 years old 15 to 20 years old 20 or more years old Firm size: 5 to 99 employees 100 to 999 employees 1000 employees HC quartiles: lowest Highest Workers: % prime age % female % foreign born % white, non-Hispanic Relative labor productivity: deviation of log (productivity) from 2-digit SIC mean Churning rate (1st quarter, 1997) Number of establishments Industry: Construction Manufacturing Transportation Wholesale trade Retail trade Finance Services

0.0656 13.11 %

–0.1844 5.24 %

9.79 % 14.73 % 16.78 % 18.02 % 40.67 %

27.17 % 24.60 % 17.82 % 12.06 % 18.35 %

83.88 % 14.80 % 1.33 %

96.05 % 3.67 % 0.27 %

20.56 % 32.02 %

31.14 % 21.68 %

75.80 % 47.44 % 13.36 % 76.33 %

68.88 % 47.93 % 18.38 % 68.90 %

0.2963 0.2066 2.28

–0.0704 0.3269 1.34

5.19 % 14.57 % 4.28 % 13.76 % 10.66 % 8.69 % 42.42 %

4.22 % 9.77 % 5.07 % 7.81 % 29.08 % 6.34 % 37.28 %

Notes: All figures are for the 1997 BR/5500 sample matched to the UI data as well, which is a subset of the overall BR/5500 sample. Corresponds to the sample used in table 13.3, third column.

References Abowd, J. M., P. A. Lengermann, K. L. McKinney. 2002. The Measurement of Human Capital in the U.S. Economy. Longitudinal Employer-Household Dynamics (LEHD) Technical Paper no. TP-2002-09, U.S. Census Bureau. Bulow, J. I. and L. H. Summers. 1986. A theory of dual labor markets with application to industrial policy, discrimination, and keynesian unemployment.” Journal of Labor Economics 4 (3, part 1): 376–414.

The Role of Fringe Benefits in Employer and Workforce Dynamics

505

Burgess, S., J. Lane, and D. Stevens. 2000. Job flows, worker flows, and churning. Journal of Labor Economics 18 (3): 473–502. Decressin, A., J. Lane, K. McCue, and M. Stinson. 2005. Employer-provided benefit plans, workforce composition, and firm outcomes. Technical Paper TP-2005-01, Longitudinal Employer-Household Dynamics (LEHD), U.S. Census Bureau. Decressin, A., K. McCue, and M. Stinson. 2003. Describing the form 5500business register match. Technical Paper TP-2003-05, Longitudinal EmployerHousehold Dynamics (LEHD), U.S. Census Bureau. Dickens, W. T. and K. Lang. 1985. A test of dual labor market theory. American Economic Review 75 (4): 792–805. Dorsey, S., C. Cornwell, and D. Macpherson. 1998. Pensions and Productivity. Kalamazoo, MI: W. E. Upjohn Institute for Employment Research. Ehrenberg, R. and R. Smith. 1996. Modern Labor Economics: Theory and Public Policy, 6th edition. Reading, MA: Addison-Wesley. Even, W. E. and D. A. Macpherson. 1996. Employer size and labor turnover: The role of pensions. Industrial and Labor Relations Review 49 (4): 707–28. ———. 2001. Benefits and productivity. Pension Research Council Working Paper 2001-16. Gustman, A. L., O. S. Mitchell, and T. L. Steinmeier. 1994. The role of pensions in the labor market: A survey of the literature. Industrial and Labor Relations Review 47 (3): 417–38. Gustman, A. L., and T. L. Steinmeier. 1993. Pension portability and labor mobility: Evidence from the SIPP. Journal of Public Economics 50 (2): 299–323. ———. 1995. Pension incentives and job mobility. Kalamazoo, MI: W. E. Upjohn Institute for Employment Research. Ippolito, R. A. 1997. Pension plans and employee performance: Evidence, analysis, and policy. Chicago: The University of Chicago Press. ———. 2002. Stayers as ‘workers’ and ‘savers’: Toward reconciling the pensionquit literature. Journal of Human Resources 37 (2): 275–308. Montgomery, E., and K. Shaw. 1992. Pensions and wage premia. NBER Working Paper no. 3985. Cambridge, MA: National Bureau of Economic Research, January. Staﬀ of the LEHD Program. 2002. The Longitudinal Employer-Household Dynamics Program: Employment dynamics estimates project versions 2.2 and 2.3. Technical Paper TP-2002-05-rev1, LEHD, U.S. Census Bureau.

Comment

Dan A. Black

Decressin, Hill, McCue, and Stinson (DHMS henceforth) should be congratulated for a very good chapter. Moreover, the profession owes them a debt of gratitude for their eﬀorts to create this intriguing data set. Economists are heavy users of data sets, but we seem to undervalue the eﬀort and creativity associated with the creation of data. The authors of this chapter have done an immense amount of work creating an important data set that Dan A. Black is a professor at the Irving B. Harris Graduate School of Public Policy Studies, University of Chicago, and a senior fellow at the National Opinion Research Center (NORC) at the University of Chicago.

506

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

links employers and employees with administrative data on the fringe benefits oﬀered by employers. In addition, their chapter provides some intriguing relationships between the performance of firms and the firms’ decisions about whether to oﬀer fringe benefits. In many ways, in the analysis of this chapter DHMS raise more questions than they answer. In this comment, I would like to focus on two important issues that jumped out at me when I read their chapter. Using these data, the latter part of their chapter examines the association of fringe benefits and firm turnover, employment growth, productivity, and the likelihood the firm goes out of business. While the link between the association and causality is tenuous at best without a better model of the contracts between workers and their firms, these associations are suggestive of an important role of fringe benefits in the labor market. In their analysis, DHMS find that the provision of fringe benefits is associated with lower turnover, higher productivity growth, and a higher probability of surviving. Of course, these are good outcomes for the firms, and a natural concern is that there is another omitted factor that reduces turnover, increases productivity growth, and increases the probability of surviving, as well as allowing firms to oﬀer fringe benefits. Obviously, an important next step would be to develop a model of the provision of fringe benefits and find some variables that would not directly aﬀect the outcomes of interest, but would aﬀect the likelihood that a firm does indeed oﬀer fringe benefits. With such exclusion restrictions, we could then see if the intriguing associations documented in this chapter are in fact causal. In the rest of this comment, I want to focus on the problem of measurement error when one matches employer and employee data. Due to the nature of the matching process, there are a lot of false negatives: many workers who in fact may have fringe benefits are not matched. In table 13C.1, I reproduce a comparison that DHMS make between Bureau of Labor Statistics (BLS) estimates of pension and health insurance with similar estimates from their data. One of the more striking features in DHMS data is the low incidence they find of health insurance. While their pension coverage appears to be measured very well, their measures of health insurance coverage are poor. Because of this measurement error, DHMS aggregate their measure into a single variable: whether the firm oﬀers any fringe benefits. Usually, measurement error in a binary variable is a problem without instruments: ordinary least squares (OLS) estimates are attenuated.1 Because of the peculiar form of the measurement error, however, the situation is much more promising. Because we assured that there are virtually no 1. Instrumental variable (IV) estimates are biased away from zero; see Black, Berger, and Scott (2000). Black, Berger, and Scott, as well as Kane, Rouse, and Staiger (1999) discuss how identification may be achieved with two measures of the binary variable; Frazis and Loewenstein (2003) extend the analysis to any instruments. See Bound, Brown, and Mathiowetz (2001) for an excellent discussion.

The Role of Fringe Benefits in Employer and Workforce Dynamics Table 13C.1

BLS DHMS

507

Comparison of pension and health insurance coverage rates Pension coverage

Health insurance coverage

56% 61%

86% 34%

Source: Decressin, Hill, McCue, and Stinson (chapter 13, this volume).

false positive matches—workers incorrectly matched to firms—we may be assured that virtually all of the matched workers do indeed have the pension and health insurance ascribed to them. This allows the research to measure correctly one of two moment conditions, which allows for identification under much weaker assumptions. To see why, consider the impact of benefit coverage on an outcome, yi. We wish to estimate: (1)

(X ) E( yi |X, C 1) E( yi |X, C 0)

where (X ) are the parameters of interest. We compare the expectation of our outcome when a worker is covered (C 1) to the expectation of our outcome when a worker is not covered (C 0), conditional on the realization of some covariates (X ). When there are no false positives, we may estimate E( yi | X, C 1) from the noisy measure of coverage E( yi | X, C˜ 1) if we are willing to assume that E( yi | X, C˜ 1) E( yi | X, C 1) so that the mismeasurement is uncorrelated with yi. The problem is that E( yi | X, C˜ 0) is contaminated with false negatives. Thus, we have (2)

E( yi |X, C˜ 0) (X )E( yi |X, C 1, C˜ 0) [1 (X )]E( yi |X, C 0).

Fortunately, we have an estimate of E( yi | X, C 1, C˜ 0), which should just be E( yi | X, C˜ 1) under the assumption that E( yi | X, C˜ 1) E( yi |X, C 1). Thus, equation (2) has only two unknown parameters: (X ) and E( yi | X, C 0). If we may use alternative data sets that allow us to estimate, conditional on X, the probability of benefit coverage, then we may obtain estimates of (X ). This would then allow the researcher to recover estimates of E( yi | X, C 0) directly from equation (2). Thus, with auxiliary data on the probability of coverage, it would be possible to identify the parameters of interest nonparametrically if we are, of course, willing to make the assumption that E( yi | X, C˜ 1) E( yi | X, C 1). This is potentially an important result because, as we see the growth of more and more data of the type that DHMS use (matched administrative records), this is likely to become a more common form of measurement error. If the matching problem is severe (as it is in the DHMS measure of the coverage of health insurance), we may at least take some comfort from the

508

Anja Decressin, Tomeka Hill, Kristin McCue, and Martha Stinson

fact that attenuation bias may be corrected through the use of auxiliary data. This result also has some important implications for how data matching should be performed. Often researchers are confronted with tough decisions regarding whether a particular pair is correctly matched. This analysis suggests that researchers should be quite demanding on the data before agreeing that there is a true match. This will insure that we have the necessary one-sided error that allows us to recover the parameters of interest. Of course, if the researcher wants to perform a probabilistic match for observations that may be correctly matched, an indicator variable that documents perfect matches will also allow researchers to recover the parameters of interest, as well as assess the accuracy of their probabilistic matches. Of course, the identification is achieved only if we are able to maintain the assumption that E( yi | X, C˜ 1) E( yi | X, C 1). Assessing the validity of this assumption is, of course, quite diﬃcult. As DHMS document, the match rate varies systematically with firm size, and the variation is larger than what we would expect if the variation were solely due to diﬀerences in the rates of provision of fringe benefits. We know, of course, that there are systematic diﬀerences in outcomes by firm size so it would be prudent to include firm size in the vector of covariates. One fears, however, that conditional on coverage, unmatched firms are simply inferior at filling out paperwork (hence the lack of a match), but these firms are also inferior at running their businesses and hence have worse outcomes on the average. Unfortunately, there is little empirical content to the assumption that E( yi | X, C˜ 1) E( yi | X, C 1). If means are suﬃciently disparate, it is possible that the observed mean, E( yi | X, C˜ 0), will not allow the imposition of E( yi | X, C˜ 1) E( yi | X, C 1), but this is quite unlikely unless the means E( yi | X, C 1, C˜ 0) and E( yi | X, C 0) are greatly diﬀerent than the E( yi | X, C˜ 1). In most applications, however, one suspects that the means are not that greatly diﬀerent and so the assumption will pass this weak test. At some level, the lack of empirical content of the identification assumption is troublesome. Clearly, one would prefer a strong test of the identification assumption before we make heavy use of it. Yet most empirical papers make another, even stronger assumption: there is no measurement error in the data used. More importantly, I think this identification strategy shows that there may be immense value to matching employer and employee data even when the match rates may be quite low. As long as researchers can use the resulting data to estimate accurately one of the two moment conditions, the use of auxiliary data may allow researchers to recover the parameters of interest despite very high levels of measurement error. Again, DHMS should be congratulated for a very good chapter and their immense eﬀorts in the creation of this extremely interesting data set.

The Role of Fringe Benefits in Employer and Workforce Dynamics

509

References Black, D., M. Berger, and F. Scott. 2000. Bounding parameter estimates with nonclassical measurement error. Journal of the American Statistical Association 95: 739–48. Bound, J., C. Brown, and N. Mathiowetz. 2001. Measurement error in survey data. In Handbook of Econometrics, Volume 5, ed. J. J. Heckman and E. Leamer, 3705– 3843. Amsterdam: Elsevier Science, B.V. Frazis, H. and M. A. Lowenstein. 2003. Estimating linear regression with mismeasured, possibly endogenous, binary explanatory variables. Journal of Econometrics 117: 151–78. Kane, T., C. Rouse, and D. Staiger. 1999. Estimating returns to schooling when schooling is misreported. NBER Working Paper no. 7235. Cambridge, MA: National Bureau of Economic Research, June.

V

Producer Dynamics in International Markets

14 Importers, Exporters, and Multinationals A Portrait of Firms in the U.S. that Trade Goods Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

14.1 Introduction “What does (Art Vandelay) do?” “He’s an importer.” “Just imports? No exports?” “He’s an importer-exporter. Okay?” Seinfeld, Episode: The Cadillac (2), aired 1996, NBC Art Vandelay is not alone. In 1993, 38.1 million workers were employed by a firm that was directly engaged in the international trade of goods (see table 14.1). These workers represent 31.7 percent of the entire civilian workforce and 40.0 of employment outside government and education.1 By 2000, the total number of workers at firms that either import or export Andrew B. Bernard is the director of the Center for International Business and the Jack Byrne Professor of International Economics at the Tuck School of Business at Dartmouth College, and a research associate of the National Bureau of Economic Research. J. Bradford Jensen is an associate professor at the McDonough School of Business at Georgetown University, a research associate of the National Bureau of Economic Research, and a senior fellow at the Peterson Institute. Peter K. Schott is a professor of economics at Yale School of Management and a research associate of the National Bureau of Economic Research. We thank our discussant James Harrigan for helpful comments and Evan Gill for research assistance. We thank the National Science Foundation (SES-0241474, SES-0552029, and SES-0550190) for research support. The research in this chapter was conducted while the authors were Special Sworn Status researchers of the U.S. Census Bureau at the Boston Census Research Data Center and the Center for Economic Studies. Results and conclusions expressed are those of the authors and do not necessarily reflect the views of the Census Bureau or the NBER. This chapter has been screened to insure that no confidential data are revealed. 1. These shares are probably an understatement of the employment at firms directly engaged in goods trade, as the linked data employed in this chapter cannot associate every export and import transaction with a firm. We discuss this issue in greater detail in the data appendix. We also provide a more precise definition of the nongovernment, nonagriculture workforce in section 14.3.

513

514

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

Table 14.1

Employment at firms engaged in trade Employment (Mill) at trading firms 1993

Firms that trade Firms that export Firms that import Firms that export and import Firms that just export Firms that just import

2000

Employment

Share (%)

Employment

Share (%)

38.1 34.6 30.8 27.3 7.3 3.5

40.0 36.3 32.3 28.7 7.7 3.7

47.9 45.0 37.7 34.8 10.2 2.9

41.9 39.4 33.0 30.4 8.9 2.5

Notes: Table reports the amount of employment (in millions of workers) and share of total civilian U.S. employment at private firms. For a more detailed description of the firm and employment data see section 14.3 and the appendix. The categories are not mutually exclusive, that is, the bottom three rows sum to the first row, as do the second and the sixth, and similarly for the third and fifth rows.

had risen to 47.9 million, or 35.0 percent of the civilian workforce. Indeed, importing and exporting are closely related—more than 50 percent of the firms in the United States that import also export and these firms account for close to 90 percent of U.S. trade. This chapter oﬀers an integrated perspective on globally engaged firms by exploring a newly developed data set that links international trade transactions to longitudinal data on U.S. enterprises. It extends existing empirical research by examining importers as well as exporters, identifying the activities of multinational firms separately from those of domestic enterprises, and diﬀerentiating between arm’s-length and related-party (i.e., intra-firm) trade. A surge of interest in the microeconomics of international trade and investment has yielded numerous studies of exporters and multinationals. Using firm-level data, empirical researchers have documented that exporting plants and firms represent a small fraction of the total, that firms engaged in exporting have positive performance characteristics (including higher productivity, larger size, greater capital intensity, etc.), that multinational firms pay higher wages than domestic counterparts, and that globally engaged firms undertake more innovation.2 To date, these research streams have proceeded largely in parallel with little integration. This chapter expands our understanding of internationally engaged firms by examining a number of new dimensions of firm activity, including how many products firms trade, how many countries firms transact with, the characteristics of those countries, the concentration of trade across firms, 2. See Bernard and Jensen (1995, 1999), Doms and Jensen (1998), and Criscuolo, Haskel, and Slaughter (2004).

Importers, Exporters, and Multinationals

515

and whether firms import as well as export. We also trace the evolution of these variables, as well as firm survival and employment over time. Our ability to answer these questions is made possible by merging two newly available data sets. The first records U.S. import and exports at the transaction level (i.e., according to the customs documents that accompany every shipment of goods crossing a U.S. border). A unique feature of these documents is that they note whether a transaction takes place at arm’s length or between related parties.3 We merge these data with a second, recently developed longitudinal database of U.S. enterprises that tracks almost all private sector firms in the United States as well as their employment over time (Jarmin and Miranda 2002). The merged data set provides a more complete picture of firm-level U.S. trade than has heretofore been possible. For example, we can examine the trading activity of firms both inside and outside of manufacturing. We also can identify firms that import as well as firms that export or do both. Perhaps most importantly, unlike most other data sources on trade, we can measure how much of each firm’s trade takes place at arm’s length versus with related parties. Our analysis uncovers a wealth of interesting results. Some of these reinforce existing findings, while others are entirely new. We find U.S. trade to be concentrated among a very small number of firms. In 2000, for example, the top 1 percent of trading firms (in terms of their trade flows) account for 81 percent of U.S. trade. In terms of product and trading-partner intensity, we find that most importers as well as exporters tend to trade relatively few products and engage in trade with a relatively small number of high-income countries. However, the small number of firms with the greatest product and trading-partner intensity employ large numbers of workers and account for the preponderance of both exports and imports. Over time, the number of firms that export and the number of firms that import rises substantially, from 2.6 and 1.7 percent of all firms in 1993, respectively, to 3.1 and 2.2 percent of all firms in 2000. For exporters, this increase is matched by greater product and trading-partner intensity: between 1993 and 2000, exporters’ average number of products increases from 6 to 10, while their average number of destination countries increases from 3.3 to 3.5. For importers, there is little change in either product or tradingpartner intensity. By linking trade transactions to a comprehensive database on U.S. em3. “Related party” trade refers to trade between U.S. companies and their foreign subsidiaries as well as trade between U.S. subsidiaries of foreign companies and their foreign aﬃliates. For imports, firms are related if either owns, controls, or holds voting power equivalent to 6 percent of the outstanding voting stock or shares of the other organization (see Section 402[e] of the Tariﬀ Act of 1930). For exports, firms are related if either party owns, directly or indirectly, 10 percent or more of the other party (see Section 30.7[v] of The Foreign Trade Statistics Regulations).

516

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

ployment we are able to explore the composition of trading firms across goods-producing, wholesale and retail, and service sectors. We find that the greatest share of exporting and especially importing firms are found in wholesale and retail trade. However, goods-producing firms account for the majority of exports and imports by value. Multinationals that export are typically goods producers while more than half of multinational importers are in the wholesale and retail sector. Analysis of firm dynamics reveals that both importing and exporting are associated with greater probability of survival. Both importers and exporters are less likely to exit than firms that do not trade, and firms that engage in some form of related-party trade (i.e., multinationals) have even lower failure rates than firms that trade at arm’s length.4 Employment growth also varies by trading status. We find that trading firms increase employment more rapidly than nontrading firms between 1993 and 2000. We also observe that firms switching their trading status during the sample period have more extreme changes in employment growth than firms with constant trade status. The average firm that opens up to trade between 1993 and 2000 experiences employment growth of close to 100 percent, while the average firm that quits trading over this period experiences a decline on the order of 10 percent. By comparison, employment growth at continuing traders and continuing nontraders averages between 20 and 25 percent. The unique characteristics of our data permit identification of a special subset of firms that we refer to as the most globally engaged (MGE). These MGE firms import as well as export and conduct at least a portion of both types of trade with related parties. Thus, these multinationals have the maximum possible links to the global economy. The MGE firms are very influential in U.S. trade and employment. In 2000 they account for nearly 80 percent of U.S. exports and imports, respectively, and employ 18 percent of the entire U.S. civilian workforce. They also stand out in a number of other dimensions. First, they are more likely to export to and import from low-income countries than other U.S. exporters and importers. Second, they experience substantially higher growth in exports and imports per worker than non-MGE traders. Finally, over time the MGEs increase their share of intra-firm trade with low-income countries and increase their share of arm’s-length trade with upper-income countries. The remainder of this chapter is structured as follows. Section 14.2 documents existing empirical research. Section 14.3 and the Data appendix provide a detailed description of our data set. Section 14.4 characterizes U.S. trade according to various dimensions of firm activity. Section 14.5 oﬀers an in-depth view of U.S. multinationals and MGEs. Section 14.6 summarizes trading firm dynamics, and section 14.7 concludes. 4. This definition of a multinational is comparable to that employed by the Bureau of Economic Analysis in its surveys of multinational firms.

Importers, Exporters, and Multinationals

517

14.2 Existing Research We begin by reviewing the existing literature on exporters, importers, and multinationals. Our overview is limited to empirical studies that describe their characteristics and the role they play in U.S. trade and employment. We note that there is virtually no research documenting and analyzing importing firms. In the last decade a substantial body of work has documented the diﬀerences between exporters and firms producing solely for the domestic market. Looking at U.S. manufacturing firms, Bernard and Jensen (1995, 1999) find that exporters are relatively rare and quite large. Even in tradable goods sectors, the majority of plants and firms do not export and nonexporters are an order of magnitude smaller than exporters. In addition, exporters are more productive, more capital-intensive, pay higher wages, employ more technology, and have more skilled workers than nonexporting firms, even when controlling for industry and geography.5 To date, these studies have been largely limited to the manufacturing sector due to the limitations of the underlying data.6 In this chapter, we summarize export participation and the employment evolution of exporters across all sectors of the U.S. economy from 1993 to 2000. Recent work by Eaton et al. (2004) extends the analysis of exporting manufacturing firms. These authors examine French firm-level data in 1986 that include information on the destination markets for exporters as well as information about the manufacturing firms themselves. These data show that 17.4 percent of the 234,300 French manufacturing firms export; among the exporters, 34.5 percent ship to exactly one country while 19.7 percent export to ten or more markets, and only 1.5 percent export to fifty or more countries. We examine the intensity of export and import activity by U.S.-based firms and changes in these intensities over time. In addition, we sort source and destination countries into groups based on income per capita and examine how trading patterns vary according to the global engagement of the firm. Given the increasing attention to exporters, it is surprising how little work has considered the actions of importing firms. There are no systematic studies of the characteristics of importing firms in the U.S. or other developed economies. MacGarvie (2003) reports some features of large importers using French firm data in her study of the patenting behavior of trading firms. In a subsample of 2,757 large firms, she finds diﬀerences between firms that trade and those that do not. Specifically, in her sample she 5. Similar evidence on exporters has been documented for other countries, for example, Bernard and Wagner (1997); Germany: Clerides, Lach, and Tybout (1998); Colombia, Mexico, and Morocco: Aw, Chung, and Roberts (2000); Korea and Taiwan: Delgado, Farinas, and Ruano (2002); Spain, among many others. 6. The general data source for such studies are censuses of manufacturing plants or firms (e.g., the U.S. Census of Manufactures).

518

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

compares exporters and nonexporters and then importers and nonimporters and find that both exporters and importers are larger, more productive, more capital-intensive, and pay higher wages. While she notes that exporters are likely to also be importers, she does not separately examine firms that both export and import. Given the nature of our data, we are able to provide a first look at the extent of importing by U.S. firms, the distribution of activity across importers, and their role in the overall economy. There is also an enormous literature on multinational firms that we cannot hope to adequately summarize here. As our focus is on the exports, imports, and employment of U.S.-based firms, we limit our discussion to studies of multinationals based in the United States, either U.S. parents or U.S. aﬃliates of foreign firms, that also examine these areas. Two recent papers by Slaughter (2004a, 2004b) using aggregate data from the Bureau of Economic Analysis summarize employment trends of multinationals operating in the United States. Although these papers focus on two diﬀerent types of multinationals based in the United States, both report sizable increases in employment at multinationals during the 1990s. Slaughter (2004a) finds that U.S. employment of U.S. multinationals increases from 17.5 million to 23.9 million from 1993 to 2000. Looking at U.S. aﬃliates of foreign parents, Slaughter (2004b) reports that employment rises from 3.9 million in 1992 to 5.4 million in 2002. Using our firmlevel data, we are able to decompose the overall changes in U.S. employment from 1993 to 2000 by the trading activities of the firm.7 Another body of work has documented diﬀerences between multinational and domestic firms. Doms and Jensen (1998) use plant-level data from the Census Bureau and the Bureau of Economic Analysis to examine the characteristics of plants owned by multinational companies. Doms and Jensen find that U.S. plants owned by MNCs (whether U.S. MNCs or foreign-owned MNCs) are larger, more capital intensive, more skill intensive, pay higher wages, are more technology intensive, and are more productive than non-MNC plants. A related literature focuses on multinational trade. Zeile (1997) summarizes the role of multinationals and intra-firm trade in overall U.S. trade using data from firm-level surveys conducted by the Bureau of Economic Analysis. Zeile (1997) reports little trend in the share of intra-firm exports and imports in total U.S. exports and imports from 1977 to 1994. He also reports that U.S. parents have seen their share of trade decrease even as their trade has shifted toward intra-firm activity. Using trade transaction data, we are able to examine the role of multinationals in U.S. exports and imports and we report separate results for total trade and related-party trade throughout the chapter. 7. Our linked trade-firm data does not provide information on the nationality of ownership so we are unable to separately examine the activities of U.S.-based versus foreign-based multinationals.

Importers, Exporters, and Multinationals

519

Another collection of recent papers using firm-level data has examined the decision by U.S. multinationals to export intermediate goods to their foreign aﬃliates. Hanson et al. (2004) find that higher trade costs, higher wages for unskilled labor, and higher corporate tax rates reduce demand for intermediate inputs exported by U.S. parents. Borga and Zeile (2004) also use data on U.S. MNCs collected by the U.S. Bureau of Economic Analysis in the 1994 benchmark survey. They report that the share of intermediate goods exported from U.S. parents to their aﬃliates increased from 8 percent of total U.S. exports in 1977 to 15 percent in 1999. Borga and Zeile (2002) are primarily concerned with analyzing vertical versus horizontal multinational structure and consider the role of firm, industry, and country eﬀects on the share of imported intermediates in total sales of aﬃliates. One of the main goals of this chapter and further research using the transaction-firm linked data is the development of a deeper understanding of the decision to trade at arm’s length or inside the firm. The role of arm’slength versus intra-firm trade has been the focus of several recent theoretical papers. Antràs (2003) develops a trade model with firm boundaries set by incomplete contracts and property rights to examine the variation in intra-firm trade across destinations and sectors in U.S. trade. Antràs and Helpman (2004) study the importance of within-sector heterogeneity and industry characteristics on the prevalence of integrated versus arm’s length organizational forms in a model North-South trade. Grossman and Helpman (2004) develop a model of firm organization and location across borders that focuses on problems in contracting between principals and suppliers or employees in a world with heterogeneous firms. Grossman, Helpman, and Szeidl (2006) develop a model of heterogeneous firms in the presence of variation in industry characteristics, the cost of transport, and regional demand. 14.3 Data This chapter exploits a new data set that links individual trade transactions to U.S.-based firms. This data set is derived from two sources. The first is a database of all U.S. trade transactions assembled by U.S. Customs (imports) and the U.S. Census Bureau (exports). These data cover all shipments of goods that crossed into or out of the United States between 1992 and 2000 inclusive. In this chapter, we make use of data from the years 1993 and 2000. The second data source is the Longitudinal Business Database (LBD) of the Census Bureau.8 These data record employment and survival information for all U.S. establishments outside of agriculture, forestry and fishing, 8. See the data appendix for more information on all the data sources and the sectors covered. See Jarmin and Miranda (2002) for an extensive discussion of the LBD and its construction.

520

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

railroads, the U.S. Postal Service, education, public administration, and several other smaller sectors. Total employment in the sectors covered by the LBD rose from 95 million to 115 million from 1993 to 2000.9 For the firm-level summary that is the focus of this chapter, we aggregate imports and exports for each firm according to (a) product, (b) country (source or destination), (c) relationship (intra-firm or arm’s length), and (d) year.10 We also aggregate the establishment-level employment data in the LBD up to the level of the firm, retaining information on the firm-level distribution of employment across sectors. We link the two data sets at the level of the firm. This link allows us to match the inward and outward trade transactions by the dimensions noted previously to the appropriate firms. This linked data covers more than three quarters of U.S. imports and exports in each year. All of the results reported later are with respect to this linked data set unless otherwise noted. We also note that all dollar amounts reported in this chapter are nominal. Table 14.2 reports the number of trading firms as well as the total number of firms in each year of the sample. Firms are categorized according to whether they export, import, or both export and import, as well as according to whether they engage in these activities as multinationals. We categorize firms as multinationals if at least a portion of their trade is with related parties. Thus, “Multinational Exporters” diﬀer from “Exporters” in that the former have nonzero shares of related-party trade. As indicated in the table, trading firms are relatively rare vis-à-vis all firms, and multinationals are rarer still. The data indicate that firms that export are more prevalent than firms that import, but that the numbers of both types of firms engaged in international trade are increasing two to three times faster than the overall number of firms. In 2000, 3.1 percent of firms export, 2.2 percent of firms import, and 1.1 percent of firms both import and export. Fewer than a quarter of exporters or importers are multinationals. Trade in the United States is heavily concentrated among a very small number of firms. Indeed, trade concentration is much more extreme than either production or employment. Table 14.3 reports the distribution of exports and imports across firm percentiles in both 1993 and again in 2000. The top panel summarizes the share of U.S. trade and employment at firms in the top 1, 5, 10, 25, and 50 percentiles of total trade (i.e., imports plus exports). As indicated in the table, trade concentration is remarkably high, with the top 1 percent of traders (1,732 firms) accounting for 77 percent of 9. Total employment in the United States increases by 16.7 million, from 120.2 million in 1993 to 136.9 million in 2000 (Economic Report of the President 2005). 10. Every export or import transaction records whether the transaction takes place between related parties. See the data appendix for the definition of related-party transactions for exports and imports. We use the terms intra-firm and related-party interchangeably in this chapter. All firms that have a related-party transaction (export, import, or both) during the year are described as multinationals or related-party firms.

Table 14.2

Breakdown of trading firms 1993

Change 1993 to 2000

2000 % of total

% of total

Firm Type

Firms

2000

Exporters Importers Exporters and Importers Multinational exporters Multinational importers Multinational exporters and importers Total firms

130,072 86,294 43,206 23,293 19,141

2.6 1.7 0.9 0.5 0.4

167,217 117,812 60,587 28,281 24,324

7,772 4,987,145

0.2 100.0

9,559 5,474,639

Firms

Percent

3.1 2.2 1.1 0.5 0.4

37,145 31,518 17,381 4,988 5,183

29 37 40 21 27

0.2 100.0

1,787 487,494

23 10

Notes: Table reports the number of trading firms by the type of trade they engage in, as well as the total number of firms for 1993 and 2000. A firm is referred to as a multinational if at least a portion of its trade is conducted via related parties.

Table 14.3

Export and import concentration across firms

Number of firms Firm rank

1993

2000

Top 1 percent Top 5 percent Top 10 percent Top 25 percent Top 50 percent

1,732 8,658 17,316 43,290 86,580

2,245 11,223 22,445 56,111 112,221

Top 1 percent Top 5 percent Top 10 percent Top 25 percent Top 50 percent

1,301 6,504 13,008 32,518 65,036

Top 1 percent Top 5 percent Top 10 percent Top 25 percent Top 50 percent

863 4,315 8,630 21,574 43,147

Percent of all firms 1993

Percent of employment

Percent of trade

2000

1993

2000

1993

2000

Total trade 0.03 0.04 0.17 0.20 0.35 0.41 0.87 1.02 1.74 2.05

15.1 21.2 23.7 28.2 32.4

14.0 21.2 23.9 28.7 34.2

77.1 90.8 95.1 98.7 99.8

80.9 92.7 96.1 99.0 99.8

1,673 8,361 16,722 41,805 83,609

Exports 0.03 0.13 0.26 0.65 1.30

0.03 0.15 0.31 0.76 1.53

11.8 17.7 21.5 26.0 30.5

11.0 17.6 20.8 27.0 32.7

78.2 91.8 95.6 98.7 99.7

80.9 93.0 96.3 98.9 99.8

1,179 5,891 11,782 29,453 58,906

Imports 0.02 0.09 0.17 0.43 0.87

0.02 0.11 0.22 0.54 1.08

11.5 16.7 18.9 22.1 25.6

11.0 16.3 18.5 21.7 25.5

72.7 88.2 93.4 98.2 99.7

77.6 90.8 95.0 98.6 99.8

Notes: Table reports the number of firms, percent of all U.S. firms, percent of employment and percent of U.S. trade for firms that are responsible for the top 1, 5, 10, 25, and 50 percentiles of the total trade, export and import distributions, respectively.

522

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

exports plus imports in 1993.11 These firms are also among the largest in the economy, accounting for 15.1 percent of employment or 14.3 million workers. Over time, trade is becoming increasingly concentrated at the top firms. By 2000, the largest 1 percent of trading firms (2,245 firms) control almost 81 percent of all trade.12 The second and third panels of table 14.2 report concentration among importers and exporters separately. Importers show a similar if slightly smaller degree of concentration than exporters. For both imports and exports, the smallest 75 percent of firms are responsible for less than 2 percent of imports and exports, respectively. 14.4 Importers and Exporters In this section we characterize U.S. firm-level trade according to several dimensions of activity. First we examine firms’ product and tradingpartner intensity, that is, the number of products firms trade and the number of countries with which they trade. We then segment firm trade according to the income level of source and destination countries. Finally, we categorize trading firms’ global engagement and identify the set and influence of firms that we define to be the most globally engaged (MGE). This section highlights several noteworthy trends. First, we show that importers as well as exporters tend to trade relatively few products with a relatively small number of countries. Second, we show that most trading firms import from or export to relatively high-income countries, and that importers are relatively more likely to trade with lower-income countries than exporters. Third, we find that a substantial and growing fraction of trading firms are in service sectors, particularly wholesale and retail, though the majority of MGEs (multinationals that export as well as import) are found in manufacturing. Finally, we demonstrate that MGE firms dominate U.S. trade flows and employment among trading firms. 14.4.1 Firms’ Product-Intensity Exporters generally export fewer products per firm than importers import, but exporters are catching up over time. Between 1993 and 2000, the average number of products exported by exporters rose from 6.1 to 8.9 products per firm. The average importer sources ten products in both periods. Table 14.4 reports the distribution of firms, export and import value, intrafirm trade, and employment according to the number of products firms import or export in each year. Each cell of the table reports the share of one of these variables accounted for by all firms exporting or importing the 11. These firms control equal shares of exports and imports. 12. Note that while the shares of the top 5, 10, 25, and 50 percent of firms rose, these increases were due entirely to growth in shares at the very top of the distribution.

Importers, Exporters, and Multinationals Table 14.4

Share of firms, value, and employment by number of products exported or imported per firm

Share of firms (%) Products

523

1993

2000

Share of value (%) 1993

2000

Related-party value share (%)

Employment share (%)

1993

2000

1993

2000

0.3 0.2 0.6 2.5 96.5

0.1 0.2 0.4 1.1 98.2

63.7 4.0 2.5 2.7 3.9 23.3

60.6 4.4 2.5 2.8 4.9 24.7

0.2 0.4 0.8 2.3 96.3

0.2 0.3 0.8 1.6 97.1

67.7 3.8 1.9 2.3 2.8 21.5

67.0 3.5 2.8 2.4 2.9 21.4

Exports 0 1 2 3–4 5–9 10

41.2 16.8 16.3 14.2 11.6

38.0 16.2 16.1 15.1 14.5

1.1 1.2 2.9 6.0 88.9

0.7 0.8 1.7 3.8 92.9 Imports

0 1 2 3–4 5–9 10

32.1 15.1 15.7 16.3 20.8

31.6 15.2 15.9 16.5 20.8

0.8 1.1 2.5 5.2 90.4

0.7 1.1 1.9 4.1 92.1

Notes: Table reports percent of firms, share of export or import value produced by firms, and share of employment by firms according to the number of products they import and export in 1993 and 2000.

number of products noted at the left. As indicated in the table, exporters are more likely to trade just a single product and are less likely to export more than ten products than importers, though in both cases single-export and single-import firms are in the majority. The vast majority of trade value and related-party trade value, on the other hand, increasingly flows through firms that export or import the largest number of products. In 2000, just 7 percent of exports and 2 percent of related-party exports are accounted for by firms shipping fewer than ten products. Similar figures are reported for imports. Export product intensity is increasing over time while import product intensity is basically flat. The share of firms exporting just one product falls from 41 percent in 1993 to 38 percent in 2000, while the share of firms exporting ten or more products increases from 11.6 percent to 14.5 percent. This shift among exporters occurs even as the number of exporting firms rises by 29 percent and the number of exporters as a fraction of all U.S. firms increases from 2.6 percent to 3.1 percent (see table 14.2). The final block of columns in table 14.4 reports the share of U.S. employment represented by firms that export and import relative to firms that serve the domestic market only. The first row of these columns reveals that the share of workers employed by firms that do not trade, while high in

524

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

Table 14.5

Distribution of per firm and per worker statistics by number of products exported or imported per firm

Workers per firm Products

1993

2000

1 2 3–4 5–9 10

71 107 121 200 1,477

77 104 119 224 1,172

1 2 3–4 5–9 10

131 136 164 192 1,142

108 179 146 170 996

Value per firm ($000) 1993

Value per worker ($000)

2000

1993

2000

Exports 66 182 456 1,093 19,806

69 186 385 918 23,351

0.9 1.7 3.8 5.5 13.4

0.9 1.8 3.2 4.1 19.3

Imports 132 383 812 1,623 22,290

193 619 1,023 2,086 37,172

1.0 2.8 5.0 8.5 19.5

1.8 3.5 7.0 12.3 37.3

Notes: Table reports average employment per firm, export or import value per firm, and export or import value per worker across firms according to the number of products they export or import in 1993 and 2000.

both periods, has fallen with time. This decline is evident across both exporters and importers, but is more pronounced among exporters (a decline of 64 to 61 percent versus 68 to 67 percent). Table 14.5 reports the average employment as well as trading volume per firm and per worker by the number of products firms trade. As expected, average employment per firm is positively correlated with the number of products traded. Firms that export the largest number of products are more than ten times larger than exporters exporting just one or two products. Over time, the average firm size for the most prolific exporters has fallen from 1,477 employees to 1,172 employees. Over the same interval, these firms experience a slight increase in export value per firm (from roughly $20 million to $23 million) and a 44 percent increase in export value per worker, from $13.4 to $19.3 thousand. These results demonstrate that, over time, trade is becoming more concentrated at firms sending and receiving the most products across U.S. borders. This rise in concentration stems both from an increase in the number of firms engaged in multi-product trade as well as a dramatic increase in exports and imports per employee at those same firms. Firm size is actually decreasing for this group. 14.4.2 Firms’ Trading-Partner Intensity This section examines the changing nature of the firms’ global engagement in terms of their trading-partner intensity. The average number of

Importers, Exporters, and Multinationals

525

countries with which exporters trade is rising over the sample period, from 3.3 to 3.5. For importers, trading-partner intensity is flat at an average of 2.8 countries per firm in both years. Table 14.6 summarizes this activity. Here, as with product intensity, there is substantial variation across firms. More than half of both importers and exporters transact with just a single foreign country, while substantially fewer firms transact with ten or more countries. Here, too, the dominant portion of exports and imports as well as related party trade flow through firms transacting with the largest number of countries. Trading partner intensity increases slightly over time for importers and more so for exporters. Between 1993 and 2000 the share of exporters transacting with just a single country declined from 60.3 percent to 56.6 percent, while the analogous movement for importers is a decline from 52.1 percent to 51.3 percent. Similarly, the share of trade, the share of related-party trade, and the share of employment all increase over time for firms trading with more than a single country. Average firm employment, as well as average trading value per firm and per worker by trading-partner intensity, are reported in table 14.7. As above, average employment is positively correlated with the number of countries with which firms trade but is declining with time. For both exporters and importers, average value per firm and per worker for firms trading with the largest number of countries increases substantially between 1993 and 2000. Table 14.6

Share of firms, value and employment by number of source or destination countries Share of firms (%)

Destination or source countries

1993

2000

Share of value (%) 1993

2000

Related-party value share (%)

Employment share (%)

1993

2000

1993

2000

3.4 1.6 2.1 4.1 88.7

1.5 1.2 1.6 2.8 92.6

63.7 7.9 2.7 3.1 3.4 19.2

60.6 7.7 3.1 4.2 5.8 18.6

3.3 2.8 9.3 12.8 71.8

1.7 2.0 4.2 8.2 83.9

67.7 5.0 2.3 3.3 4.0 17.6

67.0 5.1 3.2 3.1 4.9 16.7

Exports 0 1 2 3–4 5–9 10

60.3 13.6 10.5 8.3 7.2

56.6 14.7 11.8 9.3 7.7

5.9 2.7 4.0 5.8 81.7

3.7 2.3 3.2 5.2 85.6

Imports 0 1 2 3–4 5–9 10

52.1 18.2 15.3 10.3 4.1

51.3 18.9 15.4 10.2 4.2

4.2 3.7 9.0 13.5 69.6

3.0 3.2 5.6 10.6 77.7

Notes: Table reports percent of firms, share of export or import value produced by firms, and share of employment by firms according to the number of countries with which they trade in 1993 and 2000.

526

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

Table 14.7

Distribution of per firm and per worker statistics by number of countries with which firms trade Workers per firm

Destination or source countries

1993

2000

1 2 3–4 5–9 10

95 143 218 302 1,944

93 145 242 430 1,652

1 2 3–4 5–9 10

106 141 241 431 4,713

97 163 197 466 3,815

Value per firm ($000) 1993

Value per worker ($000)

2000

1993

2000

241 562 980 2,049 40,675

2.6 3.6 4.4 5.9 15.0

2.6 3.9 4.0 4.8 24.6

487 1,437 3,046 8,710 153,956

3.9 7.4 12.5 15.6 18.3

5.0 8.8 15.5 18.7 40.4

Exports 251 514 964 1,786 29,085 Imports 416 1,041 3,007 6,720 86,412

Notes: Table reports average employment per firm, export or import value per firm, and export or import value per worker for firms according to the number of countries with which they trade in 1993 and 2000.

Trade is also becoming more concentrated at firms with the most trading partners. Again, this rise in concentration stems both from an increase in the number of firms with multiple trading partners as well as a dramatic increase in exports and imports per employee at those firms even as firm size has been shrinking. 14.4.3 The Income Level of Firms’ Trading Partners In this section we examine the types of countries with which firms trade. Our analysis makes use of a classification developed by the World Bank that segments countries according to whether their per capita income is low, lower-middle, upper-middle, or high.13 Use of these groups to classify trading partners is consistent with existing research indicating a strong relationship between income per capita and both variety-driven intraindustry trade and endowment-based comparative advantage. Though most trade is conducted with firms in upper-income countries, a relatively greater share of importers and import value is associated with lowermiddle-income countries. Over time, the share of trade with middle- and low-income countries is rising. 13. We use the 2003 classification for both years of our sample. The income cutoﬀs for the four groups are $765 or less, $766 to $3,035, $3,036 to $9,385 and $9,386 or more. For a list of countries and their World Bank income group, see http://www.worldbank.org/data/ countryclass/countryclass.html (The data appendix describes modifications made to this data.)

Importers, Exporters, and Multinationals

527

The first two columns of table 14.8 report the share of exporters and importers that trade with at least one country of each type in 1993 and 2000. In both years, the largest share of both exporters and importers trade with at least one upper-income country, though these shares decline over time for both groups of firms. In 2000, 85.6 percent of exporters and 79.9 percent of importers transact with at least one upper-income country, down from 88.3 percent and 85.5 percent in 1993, respectively.14 The middle two rows of each panel in table 14.8 reveal that lower-middle-income countries are substantially more important for imports than for exports. More than 30 percent of importers source goods from at least one lower-middle country in 1993, rising to more than 38 percent in 2000. This diﬀerence is likely driven by China, which is defined by the World Bank to be a lower-middle country. The largest shares of export and import value are destined for upperincome countries. In 1993, 65.5 percent of exports and 69.7 percent of imports are accounted for by upper income countries while low-income countries represented just 1.0 percent and 2.6 percent of trading value, respectively.15 Lower-middle income countries are relatively more important for imports than for exports. Over time, the import value shares represented by both middle income groups increases by 8.6 percentage points. The middle four columns of table 14.8 report the employment shares of firms as well as average employment per firm according to the types of countries with which they transact. While most exports and most exporters are engaged in trade with upper-income countries, average employment is greatest for firms shipping to low-income destinations. Average firm size falls systematically as the income of firms’ trading partners increases. This finding suggests that the largest firms are the first to enter markets that are least similar to the United States. 14.4.4 Firms’ Sector Aﬃliation Typically, imports and exports are categorized according to the product being traded. In this section we focus on firms and ask how much trade is controlled by firms in three broad sectors: goods-producing firms, wholesale and retail, and service establishments. We provide the first direct evidence on the distribution of trade by firms across sectors. We first place firms in one of five groups based on the activities of their operations in the United States. Each establishment within a firm is categorized by a primary industry designation (i.e., a four-digit Standard Industrial Classification [SIC] code). We group these codes into three sectors: 14. Note that the cumulative sum of shares in the first two columns of the table do not sum to 100 percent because firms may trade with countries of diﬀerent income levels, and therefore be included in more than one row of the table. 15. Note that export and import value shares do sum to 100 percent because export and import value can be observed at the transaction level.

10.6 38.2 18.2 79.9

8.2 30.7 15.5 85.5

Low Lower-middle Upper-middle Upper 12.5 21.3 19.0 31.1

1993 13.2 22.5 19.9 31.7

2000

1,480 660 591 303

1,684 763 1,358 401

1,202 570 1,062 385

2000

Employment per firm 1993

Importing

1,863 764 766 293

2000

Employment per firm 1993

Exporting

1.0 11.1 19.6 68.3

2000

2.6 14.0 13.6 69.7

1993

3.0 17.5 18.7 60.8

2000

Share of imports (%)

1.0 10.7 18.9 65.1

1993

Share of exports (%)

Notes: Income levels of U. S. trading partners are according to the 2003 World Bank Income Group classification available at www.worldbank.org First two columns report the percent of exporting and importing firms that export to and import from at least one country in the noted country-income groups. Subsequent columns report the share of employment, employment per firm, and export and import value represented by firms that trade with at least one country in the noted groups. The sums of all exporter and importer shares as well as the sums of all employment shares for a given year do not equal 100 because firms may appear in more than one row of the table if they trade with countries of more than one type. The sums of the shares or exports and imports for a given year do sum to 100 because they sum trade flows at the firm-destination country level.

2000

1993

Income level of source country

15.2 21.9 24.7 37.9

2000

Employment share (%)

13.2 21.4 22.4 35.4

7.0 22.7 28.6 85.6

5.2 20.5 21.4 88.3

Low Lower-middle Upper-middle Upper

Share of importers (%)

1993

2000

Employment share (%)

1993

Share of exporters (%)

Share of firms trading with diﬀerent country-income groups

Income level of destination country

Table 14.8

Importers, Exporters, and Multinationals

529

Goods (manufacturing, mining, and agriculture); Wholesale and Retail trade; and Services (all remaining industries). We then calculate the share of employment within the firm that is in each of these three aggregate sectors. Firms are assigned to one of five groups—Goods, Wholesale and Retail, Services, Goods Plus, and Other—depending upon these shares. Firms with at least 75 percent of their employment in manufacturing, mining, and agriculture are designated as Goods. Firms with at least 75 percent of their employment in Wholesale and Retail or Services are assigned to those sectors respectively. Firms with 25 to 75 percent of their employment in manufacturing, mining, and agriculture are assigned to Goods Plus. All remaining firms, that is, firms with less than 25 percent employment in Goods and less than 75 percent employment in either Wholesale and Retail or Services, are assigned to Other. Table 14.9 shows the distribution of firms, employment, and trade by firms’ sector aﬃliation. In 2000, Goods, Wholesale and Retail, and Services account for 99.9 percent of firms (7.3, 23.2, and 69.4, respectively) and 95.5 percent of employment (16.2, 24.9, and 54.4, respectively). Exporters are most likely to be in Goods or Wholesale and Retail (35.2 and 40.8 percent, respectively) with Services accounting for 22.6 percent. However, most exports (by value) originate in firms with a heavy presence in Goods: 62.8 percent at Goods firms and 19.2 percent at Goods Plus firms even though the latter sector comprises a relatively small number of firms. Exports per firm in the Goods Plus category average more than $61 million in 2000. Understandably, a greater share of importers than exporters are in Wholesale and Retail (62.7 percent in 2000), followed by Goods and Services (24.9 and 20.4 percent, respectively). Import value is also increasingly concentrated among Goods and Goods Plus firms (40.1 and 21.6 percent, respectively), though the level of imports due to Wholesale and Retail firms (27.3 percent in 2000) is substantially higher than for export value (10.4 percent). Related-party trade is most heavily concentrated at production-based firms: 90.5 percent of related-party exports and 74.5 percent of related-party imports are at Goods and Goods Plus firms in 2000. Though employment rises over the sample period for firms in all sectors except Other, employment growth is disproportionately large among trading firms in the Wholesale and Retail and Service sectors. While employment in Goods firms rises 3 percent, employment at Wholesale and Retail and Services firms grows by 18 and 30 percent, respectively. These results point to a shift in activity in the tradeable goods sectors. While goods-producing firms still dominate the landscape, trading firms are increasingly engaged in wholesale and retail trade. 14.4.5 Firms’ “Global Engagement” In previous sections we found that the largest firms account for the preponderance of trade and are the most likely to trade with the poorest

385 7.7 49.6 38.1 23.1 26.7 16.8 38.8 10.0 42.7 5.7 29.9 3.5 45.3 202,600 60.5 170,400 38.5 75,120 63.2 96,820 45.2 18,026 18.9 47 399 7.3 58.9 35.2 29.4 24.9 21.7 35.8 12.0 42.6 7.3 30.1 4.5 46.6 382,800 62.8 397,100 40.1 128,700 64.7 248,200 48.5 18,554 16.2 46

2000 4 0.1 1.9 1.5 1.3 1.5 1.2 2.8 0.9 3.7 0.7 3.4 0.5 7.0 50,720 15.1 89,270 20.2 25,770 21.7 58,270 27.2 4,167 4.4 1131

1993 3 0.1 1.9 1.1 1.4 1.2 1.3 2.1 0.9 3.2 0.7 2.9 0.6 6.5 117,100 19.2 214,200 21.6 51,270 25.8 133,000 26.0 4,207 3.7 1354

2000

Goods Plus

1,273 25.5 53.2 40.9 46.8 54.2 20.5 47.4 8.3 35.6 10.0 52.1 3.0 38.0 42,470 12.7 139,200 31.5 9,380 7.9 50,280 23.5 24,023 25.2 19

1993 1,273 23.2 68.2 40.8 62.7 53.2 29.0 47.9 10.1 35.8 12.3 50.6 3.4 35.2 63,510 10.4 269,900 27.3 10,510 5.3 95,260 18.6 28,409 24.9 22

2000

Wholesale and Retail

Sector aﬃliation

3,322 66.6 25.0 19.2 14.9 17.2 4.6 10.6 4.1 17.4 2.7 14.0 0.7 8.8 35,920 10.7 36,560 8.3 7,714 6.5 5,157 2.4 47,849 50.2 14

1993 2000 3,797 69.4 37.7 22.6 24.1 20.4 8.4 13.8 5.0 17.7 3.9 16.0 1.1 10.9 43,330 7.1 86,690 8.8 7,703 3.9 15,400 3.0 62,149 54.4 16

Services

3 0.1 0.4 0.3 0.3 0.3 0.2 0.5 0.1 0.6 0.1 0.6 0.1 0.9 3,239 1.0 6,942 1.6 929 0.8 3,464 1.6 1,187 1.2 372

1993

Other

2 0.0 0.4 0.2 0.3 0.2 0.2 0.3 0.2 0.7 0.1 0.4 0.1 0.8 3,088 0.5 21,980 2.2 772 0.4 19,770 3.9 940 0.8 381

2000

Notes: Table reports the number of trading firms (in thousands), nominal trade values (in millions of dollars) and employment (in thousands) by firms’ sector aﬃliation. Each establishment within a firm possesses a primary industry designation via a four-digit Standard Industrial Classification code. These codes map into three basic firm orientations: Goods (manufacturing, mining, or agriculture). Wholesale and Retail (wholesale or retail trade) and Services (all remaining sectors). Firms with more than 75 percent of their employees in one of these orientations are assigned to it. Firms where employment in Goods is between 25 percent and 75 percent are assigned to “Goods Plus,” and all other firms are assigned to “Other.” Firms are “E and I” if they both export and import. Firms are multinationals if at least part of their trade is with related parties. Italicized numbers represent the fraction of that column in the total (across all columns) for the row immediately above.

Average employment/firm

Employment

Related-party imports

Related-party exports

Import value

Export value

Multinational E and I

Multinational importers

Multinational exporters

E and I firms

Importing firms

Exporting firms

Firms

1993

Goods

Breakdown of firms, trade, and employment by firm activity

Firms (000), Trade Value ($Mill) or Employment (000)

Table 14.9

Importers, Exporters, and Multinationals

531

countries. In this section we define firms’ global engagement according to the breadth and depth of their global interaction. Firms may export, import, do both, or neither. Firms that both export and import have greater breadth of global engagement than firms that do not trade or firms that just export or just import. Trading firms may also trade via arm’s-length transactions or with related parties, with the latter reflecting greater depth of global engagement than purely domestic firms. We define the most globally engaged (MGE) firms as those that both export to and import from a related foreign aﬃliate. Table 14.10 reports the distribution of exporters and importers according to their export and import relationships. Results are reported in two panels, with the upper panel summarizing all firms that export and the lower panel summarizing all firms that import. The export and import relationships noted in the first two columns roughly characterize increasing global engagement. For example, arm’s-length (AL) exporters that do not

Table 14.10

Distribution of trading firms according to their export and import relationships Exporters Firms

Export relationship AL AL RP AL RP RP

Import relationship None AL None RP AL RP

Firms (%)

1993

2000

1993

2000

77,329 23,588 9,537 5,862 5,984 7,772 130,072

94,954 34,231 10,551 8,548 8,171 9,559 166,014

59 18 7 5 5 6 100

57 21 6 5 5 6 100

Importers Firms Import relationship AL AL RP AL RP RP

Export relationship None AL None RP AL RP

Firms (%)

1993

2000

1993

2000

37,581 23,588 5,507 5,984 5,862 7,772 86,294

51,017 34,231 6,208 8,171 8,548 9,559 117,812

44 27 6 7 7 9 100

43 29 5 7 7 8 100

Notes: Table summarizes the distribution of exporters and importers according to their export and import relationships. These relationships can be either arm’s-length (AL) or via related-parties (RP).

532

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

import are the least globally-engaged exporters; that is, they are less globally engaged than exporters that also import and have at least some part of one of their relationships encompassing trade with related parties. As indicated in the table, the MGE firms comprise a very small share of trading firms, with 6 percent of exporters and 9 percent of importers. The overall global engagement of exporters is increasing with time. Between 1993 and 2000, the share of exclusively arm’s-length exporters declined from 59 percent to 57 percent. Exclusively arm’s-length importers are 44 percent and 43 percent of all importers, respectively, in the two years. Table 14.11 summarizes trading firms according to both their level of global engagement and the income level of countries with which they trade. The first block of columns reports results for exporters and the countries to which they send goods, while the second block of columns reports results for importers and the countries from which they source products. In 1993, for example, 3 percent of exporters that only export and only via arm’s-length trade shipped goods to at least one country with the lowestlevel of income. The analogous number for importers is 7 percent.16 Table 14.11 shows that trading firms are most likely to transact with upper-income countries regardless of their level of global engagement, reinforcing the message of table 14.8. More interestingly, the table reveals that the most globally engaged firms (MGEs), that is, those that both import and export and engage in at least some trade with related parties, are the most likely to export to countries of all types. While just 4 percent of exclusively arm’s-length exporters export to a low-income country in 2000, for example, 28 percent of the most globally engaged firms do so that year. These diﬀerences between the least and most globally engaged firms are generally more pronounced for exporters than for importers, but are present for both groups of trading firms. Table 14.11 also shows that the greater proclivity of importers to trade with lower-middle income countries increases with their global engagement. Table 14.12 reports export and import value shares according to the same typology used in table 14.11.17 As expected, upper-income countries account for the largest share of trade value. However, an interesting diﬀerence emerges between low and low-middle trading partners versus upper and upper-middle partners. Looking across types of firms, we find that poorer countries account for a relatively larger share of trade at the least globally engaged firms. In 2000, arm’s-length exporters ship 17 percent of their goods to the two lowest income groups and arm’s-length importers source 40 percent of their imports from the same countries. In contrast, the 16. As noted in the table, the percentages for any given level of global engagement do not sum to 100 percent because firms may trade with countries of more than one income level. 17. As noted in the table, the export or import value percentages for each export and import relationship pair sum to 100 percent because trade can be observed at the firmtransaction level.

Importers, Exporters, and Multinationals Table 14.11

Global engagement and trading partner characteristics Exporter type

Trading partner income level Low

Lower-middle

Upper-middle

Upper

533

Exporters (%)

Importer type

Importers (%)

Export relationship

Import relationship

1993

2000

Import relationship

Export relationship

1993

2000

AL AL RP AL RP RP AL AL RP AL RP RP AL AL RP AL RP RP AL AL RP AL RP RP

None AL None RP AL RP None AL None RP AL RP None AL None RP AL RP None AL None RP AL RP

3 5 7 4 16 21 13 24 26 23 49 51 14 24 26 26 49 57 87 88 88 90 93 96

4 6 9 6 21 28 14 26 29 26 53 59 20 30 37 33 60 71 82 85 86 87 93 95

AL AL RP AL RP RP AL AL RP AL RP RP AL AL RP AL RP RP AL AL RP AL RP RP

None AL None RP AL RP None AL None RP AL RP None AL None RP AL RP None AL None RP AL RP

7 8 8 6 11 13 29 31 29 27 36 40 10 14 16 19 23 37 80 87 86 91 92 95

10 10 11 10 13 17 36 38 34 37 41 51 12 17 16 24 24 47 72 82 83 87 90 95

Notes: Table reports the distribution of trading firms according to both their export and import relationships and the income level of their trading partners. Exporting and importing firms are allocated to one of six mutually exclusive categories according to their export and import relationships, which can be either arm’s-length (AL) or related-party (RP). The first block of columns reports results for exporters and the countries to which they export while the second block of columns reports results for importers and the countries from which they import. The percentages reported in columns 4, 5, 8, and 9 represent the percent of trading firms of each type that export to (columns 3 and 4) or import from (columns 8 and 9) at least one country of the noted type. The percentages for any given export and import relationship pair may not sum to 100 percent because firms may trade with countries of more than one income level.

most globally-engaged multinationals send just 11 percent of their exports and source 16 percent of their imports from these same countries. 14.5 Multinationals Multinationals play a key role in U.S. employment and trade patterns. Employment at multinationals accounts for 31.3 million workers, or 27.4 percent of the nongovernmental workforce in 2000, up from 25.5 million

534

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

Table 14.12

Export and import value by firms’ global engagement and trading partner characteristics Exporter type

Trading partner income level Low

Lower-middle

Upper-middle

Upper

Export value (%)

Importer type

Import value (%)

Export relationship

Import relationship

1993

2000

Import relationship

Export relationship

1993

2000

AL AL RP AL RP RP AL AL RP AL RP RP AL AL RP AL RP RP AL AL RP AL RP RP

None AL None RP AL RP None AL None RP AL RP None AL None RP AL RP None AL None RP AL RP

2 2 2 2 1 1 16 16 10 16 9 11 17 12 18 13 10 15 66 70 70 69 79 71

3 3 2 2 1 1 14 16 14 12 12 10 16 17 20 20 20 19 66 65 64 67 66 65

AL AL RP AL RP RP AL AL RP AL RP RP AL AL RP AL RP RP AL AL RP AL RP RP

None AL None RP AL RP None AL None RP AL RP None AL None RP AL RP None AL None RP AL RP

6 4 9 3 2 2 32 32 15 29 21 10 7 13 10 11 16 14 55 50 67 58 61 74

8 6 4 3 7 2 32 37 23 38 26 14 9 11 13 16 16 20 51 45 60 43 51 63

Notes: Table reports the distribution of export and import value according to firms’ export and import relationships and the income level of their trading partners. Exporting and importing firms are allocated to one of six mutually exclusive categories according to their export and import relationships, which can be either arm’s-length (AL) or related-party (RP). The first block of columns reports results for exporters and the countries to which they export while the second block of columns reports results for importers and the countries from which they import. The percentages reported in columns 4, 5, 8, and 9 represent the share of value traded by firms of each type that export to (columns 3 and 4) or import from (columns 8 and 9) at least one country of the noted type. The percentages for any given export and import relationship pair sum to 100 percent (e.g., rows 1, 7, 13, and 19) because export and import value are observed at the transaction level.

workers and 26.7 percent in 1993 (table 14.13). The increase of employment at multinational firms represents more than a third of the net job creation in the private sector over the period, highlighting the disproportionate role of multinationals as a source of job creation. Multinationals also mediate a substantial majority of U.S. trade. This role is highlighted by figure 14.1, which reveals that roughly 90 percent of U.S. exports and imports in our sample flow through multinational firms. Each column in the figure reports the total trade by either exclusively

Importers, Exporters, and Multinationals Table 14.13

535

Employment at multinationals engaged in trade Multinational employment (mill) 1993

2000 a

Employment Share (%) Employment Sharea (%) Multinationals – that export to a related party – that import from a related party – that export to and import from a related party – that just export to a related party – that just import from a related party

25.5 23.4 19.5

26.7 24.5 20.4

31.3 27.5 23.3

27.4 24.1 20.4

17.4 6.0 2.1

18.2 6.3 2.2

19.4 8.1 3.8

17.0 7.1 3.3

Notes: Table reports the amount of employment (in millions of workers) at multinational firms in 1993 and 2000. The categories are not mutually exclusive (i.e., the bottom three rows sum to the first row, as do the second and the sixth, and similarly for the third and fifth rows). a Employment shares are with respect to total civilian U. S. employment as reported in the Economic Report of the President.

arm’s-length trading firms or multinationals in 1993 or 2000. The first four columns summarize imports while the second four columns summarize exports. The columns for multinationals note the share of their trade that is conducted at arm’s length as well as the share conducted inside the firm. As indicated in figure 14.1, multinationals’ share of total trade in our sample increases over time, rising 2.0 percent for imports and 4.0 percent for exports. Within multinationals, the breakdown of trade between intrafirm and arm’s-length transactions remains relatively constant over time. For imports, the share of intra-firm trade in the linked data set rises slightly from 47.9 percent in 1993 to 50.2 percent in 2000. For exports, it falls from 35.2 to 31.7 percent. Figures 14.2 and 14.3 break down U.S. exports and imports, respectively, by the global engagement categories employed in section 14.4.5. A large majority of both exports and imports are due to firms that both export to and import from related-parties (i.e., MGEs). In both cases these shares increase over time, from more than 70 percent in 1993 to about 80 percent in 2000. The role of MGEs in both employment and, especially, trade is on the rise, driven in large part by a large increase in the number of these most globally engaged firms. Within multinationals, the share of trade that is with related parties varies widely. Table 14.14 reports the distribution of multinational firms and related-party trade according to related-party trade intensity, that is, whether related-party trade accounts for less than 25 percent, between 25 percent and 75 percent, or more than 75 percent of multinationals’ trade, respectively. For a large share of multinationals, related-party trade makes up less than a quarter of total trade.

Fig. 14.1

The share of U.S. trade that flows through multinational firms

Fig. 14.2

Global engagements and exports

Fig. 14.3

Global engagement and imports

Importers, Exporters, and Multinationals Table 14.14

Distribution of multinational firms and related-party trade by multinationals’ related-party trade intensity Exports 1993

Related-party share of trade (%) 0.25 0.25–0.75 0.75

539

Imports 2000

1993

2000

Firms

Value

Firms

Value

Firms

Value

Firms

Value

53.0 24.6 22.4

6.7 56.6 36.7

62.4 22.6 15.1

8.8 63.5 27.7

41.9 25.1 33.0

3.3 30.8 66.0

43.1 25.0 31.9

3.5 25.9 70.6

Notes: Table reports the distribution of firms and related-party trade according to the share of trade within multinationals that is with related parties. The percentages in each columns sum to 100.

Among firms with higher related-party trade intensity, there are substantial diﬀerences between exporters and importers. About a quarter of multinationals have intra-firm trade shares between 0.25 and 0.75. Exporters in this group account for a majority of related-party trade (56.6 percent in 1993), while importers in this group, by contrast, account for a much smaller share of intra-firm trade, 30.8 percent. The roles are reversed for multinationals reporting the highest level of related-party trade intensity. Exporters with intra-firm trade shares greater than 75 percent are only 22 percent of all exporting multinationals in 1993 and their share of overall intra-firm exports is relatively low, 36.7 percent. Firms with intra-firm import shares greater than 75 percent are about one-third of importing multinationals but dominate overall intra-firm imports, 66.0 percent of total related-party imports in 1993. There are significant changes over time in the share of firms and intrafirm trade in the three groups of multinationals. In addition, we find diﬀerent trends for exports and imports. Between 1993 and 2000, the share of multinationals in the lowest related-party trade intensity category increases from 53.0 and 41.9 percent to 62.4 and 43.1 percent for exporters and importers, respectively. However, these firms are responsible for a relatively small, albeit rising, amount of related-party trade in both years, less than 10 percent for exports and less than 4 percent for imports. One potential explanation for these trends is the substantial increase in the numbers of multinationals during the period. New multinationals may have smaller share of related-party trade than established firms. The share of exports among firms with intermediate related-party trade intensity rises to 63.5 percent in 2000, while importers in this group account for a smaller share of imports in 2000, 25.9 percent. The roles are reversed for multinationals reporting the highest level of related-party trade intensity with the share of intra-firm trade falling to 27.7 for exporters and rising to 70.6 percent for importers in 2000.

540 Table 14.15

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott A breakdown of the most globally engaged firms by activity Sector Aﬃliation

Goods

Firms Export value Import value Related-party exports Related-party imports Employment

Goods Plus

Wholesale and Retail

Services

Other

1993

2000

1993

2000

1993

2000

1993

2000

1993

2000

3,523 45.3 173 69.9 155 46.1 72 70.0 95 47.7 8,018 46.2

4,486 46.9 341 68.0 363 47.1 125 66.1 244 50.2 8,346 42.9

541 7.0 42 17.1 82 24.4 20 19.4 56 28.1 3,131 18.0

603 6.3 114 22.7 202 26.2 51 26.9 133 27.3 3,313 17.0

2,955 38.0 17 6.9 73 21.6 6 6.1 41 20.6 3,232 18.6

3,387 35.4 26 5.2 141 18.3 7 3.8 79 16.2 2,949 15.2

682 8.8 13 5.3 20 6.0 4 3.8 4 1.9 2,349 13.5

1,008 10.5 18 3.6 43 5.6 5 2.7 11 2.3 4,471 23.0

71 0.9 2 0.9 6 1.9 1 0.8 3 1.7 625 3.6

75 0.8 2 0.4 21 2.8 1 0.4 20 4.0 360 1.9

Notes: Table breaks out the number of firms, trading value, and employment of the most globally engaged (MGE) firms according to their sector aﬃliation. Each establishment within a firm possesses a primary industry designation via a four-digit Standard Industrial Classification code. These codes map into three basic firm orientations: Goods (manufacturing, mining, or agriculture), Wholesale and Retail (wholesale or retail trade) and Services (all remaining sectors). Firms with more than 75 percent of their employees in one of these orientations are assigned to it. Firms where employment in Goods is between 25 percent and 75 percent are assigned to “Goods Plus,” and all other firms are assigned to “Other.”

14.5.1 The Most Globally Engaged Firms (MGEs) The most globally engaged firms are multinationals that both import and export with related parties. In this section we describe the activities of this set of firms in greater detail. Table 14.15 breaks out the number of firms, trading value, and employment of the most globally engaged firms according to the sectoral activity of the firm. The distribution of MGEs across sectors is sharply diﬀerent from the overall distribution of firms reported in table 14.9. Firms with a major presence in goods production, either Goods or Goods Plus, account for more than 50 percent of MGE firms. In contrast, goods-producing firms account for under 10 percent of all U.S. firms and 35 percent of nonmultinational firms that import and export. Wholesale and Retail and Services firms are 35.4 percent and 10 percent of MGEs, respectively, in 2000. The importance of Goods and Goods Plus firms among the most globally engaged firms is even more evident when we consider their share of trade flows. Goods-producing firms control an increasing share of total trade by MGEs, 91 percent of exports and 73 percent of imports in 2000. Intra-firm trade by MGEs is even more concentrated at Goods and Goods

Importers, Exporters, and Multinationals

541

Plus firms. Their share of MGE intra-firm imports rises to 77 percent in 2000 while their export share increases to 93 percent. These increases in export and import shares occur even as employment is shifting towards MGEs in the Wholesale and Retail sector. The overall picture painted by table 14.15 is of the continued and increasing importance of goodsproducing firms in U.S. trade flows controlled by MGEs. Table 14.16 provides a view of the distribution of MGE activity across country-income groups. The first two columns report the share of MGE intra-firm exports and imports by source or destination country where, as before, countries are grouped by per capita income. The last two columns report the share of total U.S. exports and imports controlled by MGEs. Looking across country groups, we find that intra-firm trade shares for MGEs generally are rising with the income of the source or destination country. However, there have been several notable changes over time. For both exports and imports, intra-firm trade shares are rising for the lowerincome countries. In contrast, intra-firm exports to upper-income destinations fall for MGEs, while imports show small increases in intra-firm trade even for the upper-income source countries. At the same time, table 14.16 reveals that while the importance of trade with the most globally engaged firms is falling for low-income countries, it is rising for middle- and highincome countries. Throughout this chapter, we have found that multinationals that both Table 14.16

Intra-firm trade of the most globally-engaged firms Related-party share (%) 1993

Most-globally engaged share (%) 2000

1993

2000

All countries Low–income Lower–middle Upper–middle Upper

42 14 19 53 43

Export value 37 15 22 42 38

74 70 73 76 73

82 64 78 83 82

All countries Low–income Lower–middle Upper–middle Upper

59 14 27 63 64

Import value 61 22 35 68 66

76 61 56 78 80

80 55 61 84 82

Notes: Table summarizes the activity of multinational firms that both export to and import from related parties (i.e., the most globally engaged firms). Table reports the share of trade by these firms that is intra-firm to the particular country-income group as well as the share of total trade to that country-income group accounted for by the most globally engaged firms.

542

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

export to and import from a related party play a large role in total U.S. trade. The results here suggest these firms are still heavily associated with goods production and that the extent of their intra-firm trade varies substantially with the characteristics of the source or destination country. 14.6 Importer and Exporter Dynamics In this section we examine trading-firm versus nontrading-firm survival and employment growth rates as well as changes in firms’ trading status between 1993 and 2000. We find that both importing and exporting are positively associated with survival and that multinationals have an even higher probability of survival than the larger group of trading firms. We also show that employment growth varies by trading status, with firms that transition from being nontraders to traders expanding the fastest. 14.6.1 Firm Survival Dynamics Table 14.17 decomposes the overall growth of trading firms between 1993 and 2000 into several categories. Each row of the table focuses on a diﬀerent, nonmutually exclusive subset of trading firms. In the upper panel, the first and last columns of the table report the number of firms in each subset of firms at the beginning and end of the sample period. The second and third columns of the top panel report the number of 1993 firms that shut down and the number of new firms that enter between 1993 and 2000, respectively. The fourth, fifth, and sixth columns of the upper panel report on firms that exist in both years according to their trading status: trade in both years, start trading, and stop trading, respectively. The final row of the upper panel reports an analogous breakdown for all firms. The lower panel of the table expresses all of these firm counts as percentages of their 1993 values. As indicated in the table 14.7, survival rates for firms vary according to their trading status. Exit rates for every type of trading firm (35 to 39 percent) are significantly lower than the failure rate for all firms (47 percent). Among trading firms, multinationals have higher survival probabilities than their nonmultinational counterparts, while MGE firms, (i.e., multinationals that both import and export) have the highest survival rate of all. The relatively low failure of MGE firms is one contributor to the rising share of MGE firms over time. 14.6.2 Firm Trading-Status Dynamics Table 14.17 reveals that another factor in the rising share of globally engaged firms over the sample period is the transition of some continuing firms from nontrading to trading status between 1993 and 2000. The first row of the table, for example, indicates that 49,035 firms, or 1.9 percent of the 2.6 million continuing firms that did not trade in 1993, become

100 100 100 100 100 100 100

Firms that export Firms that import Firms that both E and I Multinational exporters Multinational importers Multinational E and I All firms 37 39 35 36 37 33 47

48,269 33,273 15,106 8,401 7,119 2,570 2,354,216

Exiting firms Trade in both years

Share of firms relative to 1993 level (%) 41 49 42 61 44 52 30 40 38 54 42 39 53 57

Number of firms 64,352 53,830 52,698 36,458 22,299 18,987 9,417 7,053 10,406 7,212 2,997 3,235 2,841,710 2,632,929

New entrants

38 33 45 51 35 43 n.a.

49,035 28,656 19,301 11,811 6,706 3,327 n.a.

Nontraders that become traders

Continuing firms

22 19 21 32 25 25 n.a.

27,973 16,563 9,113 7,338 4,810 1,926 n.a.

Traders that become nontraders

129 137 140 121 127 123 110

167,217 117,812 60,587 28,281 24,324 9,559 5,474,639

2000

Notes: Table summarizes dynamics across diﬀerent subsets of firms between 1993 and 2000. The overall growth in the number of firms of each type is decomposed across columns. Upper panel displays firm counts while lower panel displays the share of each count relative to the 1993 total. Columns 1 and 7 report the number of firms of each type in 1993 and 2000, respectively. Note that the subsets of firms reported in each row are not mutually exclusive (i.e., some of the firms that export also import, and vice versa). Columns 2, 3, and 4 report the number of 1993 firms that exit, the number of new firms entering between 1993 and 2000, and the number of 1993 firms present in both years, respectively. Column 5 and 6 report the number of firms that switch their trading status between 1993 and 2000. Column 5 indicates the number of continuing firms that did not engage in the noted activity in 1993 but start doing so by 2000. Column 6 reports the opposite. n.a. not applicable.

130,072 86,294 43,206 23,293 19,141 7,772 4,987,145

1993

Decomposition of the number of trading firms, 1993 to 2000

Firms that export Firms that import Firms that both E and I Multinational exporters Multinational importers Multinational E and I All firms

Subset of firms

Table 14.17

544

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

exporters over the sample period. The share of continuing firms that move in the opposite direction, that is, that shift from being exporters in 1993 to being nonexporters in 2000, by contrast, constitute a much smaller percentage (1 percent). Similar relative magnitudes are found for all forms of global engagement—the share of continuing firms that disengage from international trade ranges from roughly one-third to three-quarters of the share of continuing firms that start trading. Furthermore, the levels and shares of firms that start engaging in international trade exceed the number of international traders that exit. Both the higher likelihood of firms switching into trade relative to switching out and the higher number of new entrants engaged in international trade spur increases in the overall share of globally engaged firms. 14.6.3 Firm Employment Dynamics Table 14.18 decomposes 1993 to 2000 employment growth along the same dimensions as table 14.17. As indicated in the last row of each panel, aggregate employment grows by 19 million workers, or 20 percent, over the sample period. Employment growth at multinationals is lower than the average, with multinational importers having the highest employment growth among multinationals. Employment growth at nonmultinational trading firms is higher, with arm’s-length exporters experiencing 30.2 percent growth, arm’s-length importers experiencing 22.3 percent growth, and firms that import and export at arm’s-length experiencing 27.4 percent growth. Table 14.19 shows the employment growth at firms by trading status. The most striking feature is the employment growth rates at firms that change their trading status. Firms that switch from being nontraders in 1993 to traders in 2000 experience the largest gains in employment growth. This growth is highlighted in table 14.19, which reveals that firms that become exporters over the sample period increase their employment by 94.3 percent, from 3.9 million to 7.4 million.18 Firms that become importers or switch into both importing and exporting experience similar increases. Table 14.19 also reports the employment declines experienced by firms that exit international markets. Firms that quit exporting, quit importing, and quit both importing and exporting witness declines of 12.3, 16.6, and 10.1 percent, respectively. Table 14.19 also reports the employment growth rates at firms that maintained the same status in both periods. For continuers, trading firms that maintain their trading status typically have lower employment growth rates than nontrading firms that maintain their trading status.

18. This is consistent with the findings of Bernard and Jensen (1999, 2004) that exporters grow significantly faster than nonexporters.

19.2 18.2 16.8 14.3 13.6 11.9 29.6

6.6 5.6 4.6 3.3 2.6 2.1 28.2

Exiting firms Trade in both years

7.5 7.0 6.8 5.1 5.6 4.3 n.a.

Nontraders that become traders

Change in employment relative to 1993 level (%) 17.7 17.0 21.7 15.2 15.3 22.8 14.3 14.9 25.0 13.3 11.8 21.8 12.0 7.0 28.6 6.6 9.2 24.8 19.0 30.5 n.a.

Change in employment 6.1 5.9 4.7 4.7 3.9 4.1 3.1 2.7 1.4 2.3 1.2 1.6 18.1 29.1

New entrants

Continuing firms

7.1 12.8 9.9 14.6 14.4 16.3 n.a.

2.5 4.0 2.7 3.4 2.8 2.8 n.a.

Traders that become nontraders

130.2 122.3 127.4 117.8 119.5 111.8 120.0

45.0 37.7 34.8 27.5 23.3 19.4 114.3

2000

Notes: Table decomposes overall employment growth across the noted subsets of firms between 1993 and 2000. The upper panel displays changes in employment for the noted subset of firms while the lower panel normalizes these employment changes according to their respective 1993 levels. Columns 1 and 7 report total employment by the noted subset of firms in 1993 and 2000, respectively. Note that the subsets of firms reported in each row are not mutually exclusive (i.e., some of the firms that export also import, and vice versa). Columns 2, 3, and 4 report the number of workers employed by firms that exit, by firms that enter between 1993 and 2000 and by firms present in both years, respectively. Columns 5 and 6 report the number of workers employed by firms that switch their trading status between 1993 and 2000. Column 5 is computed for firms that did not engage in the noted activity in 1993 but start doing so by 2000. Column 6 reports the opposite. n.a. not applicable.

100 100 100 100 100 100 100

34.6 30.8 27.3 23.4 19.5 17.4 95.3

Firms that export Firms that import Firms that both E and I Multinational exporters Multinational importers Multinational E and I All firms

Firms that export Firms that import Firms that both E and I Multinational exporters Multinational importers Multinational E and I All firms

1993

Decomposition of employment across trading firm types, 1993 to 2000

Subset of firms

Table 14.18

546

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

Table 14.19

Employment growth by firms’ trading status, 1993 to 2000 Employment (Mill)

Transition type

1993

2000

Change

% Change

Not exporting to exporting Not importing to importing Not E and I to E and I

3.9 3.6 3.3

7.5 7.0 6.8

3.6 3.4 3.5

94.3 93.9 108.3

Exporting to not exporting Importing to not importing E and I to not E and I

2.5 4.0 2.7

2.2 3.3 2.4

–0.3 –0.7 –0.3

–12.3 –16.6 –10.1

Continuing exporters Continuing importers Continuing E and I

25.5 21.3 20.0

31.6 25.9 23.9

6.1 4.7 3.9

24.0 22.0 19.5

Continuing nonexporters Continuing nonimporters Continuing non-E and I

35.3 38.2 41.1

43.9 48.9 52.0

8.6 10.7 10.9

24.5 27.9 26.6

Notes: Table reports the employment level of surviving firms that continue trading or switch to being traders of the noted type from being nontraders, and vice versa, between 1993 and 2000. E and I refers to firms that both import and export.

14.7 Conclusions This chapter provides a new integrated portrait of firms in the United States that trade goods. We document the increasing globalization of U.S. firms by linking data on U.S. international trade transactions to a comprehensive census of U.S. enterprises. The U.S. firms’ global engagement is increasing in a number of dimensions. First, there is substantial growth in the number of firms that export, import, and trade with related parties. Second, firms increasingly send a greater number of products to a larger set of more diverse countries. Third, trading firms are becoming increasingly more import- and export-intensive in terms of their dollar value of trade per worker. We show that the most globally engaged firms, that is, those that export as well as import from related parties, have substantial influence: they both account for a significant share of U.S. employment and mediate a dominant portion of U.S. trade flows. The data employed in this chapter can be used to answer a wide-ranging set of questions about the decisions of firms engaged in international commerce. By being able to separately identify arm’s-length and intra-firm transactions, we can understand the response of multinationals to financial crises, transfer pricing inside the firm, the role of firm, product, and country characteristics in the decision to outsource, pricing-to-market and pass-through responses to exchange rate movements, the role of multinationals in job creation, and the importance of imports and exports in firm performance.

Importers, Exporters, and Multinationals

547

Appendix Data Sources In this chapter, we make use of transaction-level import and export data linked to information on firms in operation in the United States. The transaction data used in this chapter are compiled from administrative records from the oﬃcial U.S. import and export merchandise trade statistics. The merchandise trade data are a complete enumeration of documentation collected by the U.S. Customs Service and are not subject to sampling error. Quality assurance procedures are performed at every stage of collection, processing, and tabulation; however, the data are subject to nonsampling errors, including undocumented shipments, timeliness, and data capture errors. The establishment and firm data used in this chapter are compiled from administrative records and the Census Bureau’s Company Organization Survey program. The establishment-level data should represent a complete enumeration of all establishments in scope for the Economic Census and not subject to sampling error. However, the data are subject to nonsampling errors. Export Transaction Data We make use of transaction-level data on exports collected by the U.S. Census Bureau via the Shippers Export Declaration (currently U.S. Department of Commerce Form 7525-V). The Census Bureau collects export shipments data for all export shipments above $2,500. The Shippers Export Declaration (SED) contains information on the firm that ships the exports (Employer Identification Number), detailed ten digit Harmonized System product code, value, quantity, export destination, date of the transaction, port, mode of transport, and whether the transaction is between related parties.19 The number of export transactions range from 13 million in 1993 to 23 million in 2000 and represent the universe of export shipments greater than $2,500. The Census Bureau imputes a total value for low-value exports. We exclude these imputed records. Canada Data Exchange The data for exports to Canada is not collected through the Shippers Export Declaration. To reduce reporting burden for U.S. and Canadian firms, the United States and Canada exchange import transaction information. The United States uses Canadian import transaction from the United 19. For exports, Foreign Trade Statistics Regulations, 30.7(v), define a related-party transaction as one between a U.S. exporter and a foreign consignee, where either party owns, directly or indirectly, 10 percent or more of the other party.

548

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

States as export transaction to Canada. These transactions contain the same information as the SEDs with the exception of Employer Identification Number. The Canadian transactions do not contain EIN but instead contain a firm name field. Exports to Canada account for approximately 35 percent of total transaction volume and approximately 20 percent of total transaction value. Import Transaction Data We make use of transaction-level data on imports collected by U.S. Customs and Border Protection via import declarations (including current U.S. Customs Forms 7501 and 7533). The U.S. Customs collects import shipments data for all import shipments above $2,000 ($250 for certain quota items). The Customs forms contain information on the firm that imports (Employer Identification Number), detailed ten-digit Harmonized System product code, value, quantity, country of origin, date of the transaction, port, mode of transport, and whether the transaction is between related parties.20 The number of import transactions range from 16 million in 1993 to 33 million in 2000 and represent the universe of import shipments greater than $2,000. The Census Bureau imputes a total value for low-value imports. We exclude these imputed records. Standard Statistical Establishment List (SSEL)/Business Register We make use of Employer Identification information and business name information from the Census Bureau Business Register (also called the Standard Statistical Establishment List [SSEL]). The SSEL contains records for all private entities except households. The SSEL carries information on the business name, address, Employer Identification Number (EIN), and information on the industry and employment at the entity. The SSEL also contains information on the firm or enterprise that owns the entity. We make use of the EIN and name information to match firm identifiers to the import and export transaction data. We use the SSEL because it contains name, EIN, and firm-level information and because it represents the largest possible universe of firms. Longitudinal Business Database (LBD) To construct firm information (employment and industrial activity), we use the Longitudinal Business Database (LBD). The LBD is a longitudinal 20. For imports, Section 402(e) of the Tariﬀ Act of 1930 defines related-party trade to include transactions between parties with various types of relationships, including “any person directly or indirectly, owning, controlling or holding power to vote, 6 percent of the outstanding voting stock or shares of any organization.”

Importers, Exporters, and Multinationals

549

version of the information contained in the SSEL. The LBD represents a significant improvement on the raw information contained in the SSEL in that it constructs longitudinal linkages for all establishments and enhances industry code information (among other improvements). See Jarmin and Miranda (2002) for more details. We use establishments in the LBD that are considered in-scope for the Economic Censuses and the County Business Patterns program. We restrict our analysis to industries that are in-scope to the Economic Census/ CBP program because industries that are not in-scope for the Economic Censuses are not broken out into establishments and the Census Bureau does not devote the same resources to these industries, so the data quality is more suspect. Jarmin and Miranda report that currently, out-of-scope industries include: Agriculture, Forestry and Fishing (SIC Division A), railroads (SIC 40), U.S. Postal Service (SIC 43), Certified Passenger Air Carriers (part of SIC 4512), Elementary and Secondary Schools (SIC 821), Colleges and Universities (SIC 822), Labor Organizations (SIC 863), Political Organizations (SIC 865), Religious Organizations (SIC 866), and Public Administration (SIC Division J). Most government owned or operated entities are outside the scope of the Economic Census. While some import and export trade transactions are matched to SSEL entities that are not in-scope for the Economic Census, the value of trade associated with these entities is quite small (approximately 3 to 5 percent). We use information from the LBD to construct firm-level measures of employment and industrial activity and exploit the longitudinal nature of the LBD to examine firm birth and death rates. Import Transaction Matching The import transaction data contain a field for the Employer Identification Number (EIN), so matching to the SSEL is relatively straightforward. The match rates of import transactions to the SSEL are typically in the 80 percent range and the share of matched import value is typically above 80 percent. The largest classes of unmatched import transactions are import transactions where the EIN is not in the SSEL or the EIN field is blank. Nonemployers are not included in the SSEL, so import transactions with Social Security Numbers (SSN) as the firm identifier will not match to the SSEL. The other large category of nonmatches is import transaction, where the EIN field is blank, representing about 3 to 5 percent of import transactions and import value. Once the match to the SSEL is made via the EIN, firm-level identifiers are applied to the import transaction data. These firm-level identifiers are then used to match to firm-level information constructed from the LBD. Detailed match rate information on import transactions and import value is presented in the top panel of table 14A.1.

550

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

Table 14A.1

Matching statistics for imports and exports Transactions 1993

Value (Bill$)

2000

1993

2000

Matched to the LBD Matched to the SSEL but not the LBD Unmatched

Imports 12,578,893 783,269 3,099,433

24,984,001 2,103,087 6,271,552

442.4 28.4 82.5

989.9 75.6 228.0

Matched to the LBD Matched by hand Matched to the SSEL but not the LBD Unmatched

Exports 8,561,733 221,226 1,335,973 3,848,122

15,430,000 410,935 2,663,119 5,370,931

328 10 27 116

601 20 64 170

Export Transaction Matching Exports to countries other than Canada contain EIN information and are relatively straightforward to match to the SSEL. For exports to Canada, we first perform an automated name match using the name field on the export transaction and the business name field on the SSEL. Subsequent to the automated matching, we do hand matching for nonmatched high value exporters to Canada. After these three phases of matching, we match approximately 70 to 75 percent of transactions and 75 to 80 percent of value.21 The largest classes of unmatched export transactions are again export transactions where the EIN is not in the SSEL or the EIN field is blank. The unmatched export transactions where the EIN field is blank represent about 7 to 10 percent of export transactions and export value. Detailed match rate information on export transactions and export value is presented in the bottom panel of table 14A.1. Country-Income Groups We use the 2003 World Bank classification of countries by their per capita income for both years of our sample. The per capita income cutoﬀs for the four groups are $765 or less, $766 to $3,035, $3,036, to $9,385, and $9,386 or more. For a list of countries and their World Bank income group see http://www.worldbank.org/data/countryclass/countryclass.html. Taiwan, Israel, and Czechoslovakia (1993 only) were not in the World 21. These match rates represent slightly lower volume match rates than the Census Bureau’s Foreign Trade Division reports for its “Profile of U.S. Exporting Companies” program. The Foreign Trade Division reports that it matches approximately 78 percent of value in 1992. We do not have access to the algorithm used by FTD or the matched files they produced; however, based on conversations with FTD staﬀ, we believe that our algorithm is more conservative than theirs (reducing the number of false positive matches). For our analytical purposes, we believe that a more conservative approach is appropriate.

Importers, Exporters, and Multinationals

551

Bank listing and were allocated to the upper-middle, upper, and lowermiddle country income groups, respectively. Smaller trading partners of the United States, that is, some small countries and country subdivisions (e.g. territories) that were missing per capita income information in the World Bank data were omitted from the country income group analysis.

References Antràs, P. 2003. Firms, contracts, and trade structure. Quarterly Journal of Economics 118 (4): 1374–1418. Antràs, P., and E. Helpman. 2004. Global sourcing. Journal of Political Economy 112:552–80. Aw, B. Y., S. Chung, and M. J. Roberts. 2000. Productivity and turnover in the export market: Micro-level evidence from the Republic of Korea and Taiwan (China). World Bank Economic Review 14 (1): 65–90. Bernard, A. B., and J. B. Jensen. 1995. Exporters, jobs, and wages in U.S. manufacturing, 1976–1987. Brookings Papers on Economic Activity, Microeconomics: 67–119. ———. 1999. Exceptional exporter performance: Cause, eﬀect, or both? Journal of International Economics 47 (1): 1–25. ———. 2004. Exporting and productivity in the U.S.A. Oxford Review of Economic Policy 20 (3): 343–57. Bernard, A. B., and J. Wagner. 1997. Exports and success in German manufacturing. Weltwirtschaftliches Archiv 133 (1): 134–57. Borga, M., and W. J. Zeile. 2004. International fragmentation of production and the intrafirm trade of U.S. multinational companies. Bureau of Economic Analysis Working Paper 2004-02. Clerides, S. K., S. Lach, and J. R. Tybout. 1998. Is learning by exporting important? Micro-dynamic evidence from Colombia, Mexico, and Morocco. Quarterly Journal of Economics 113 (3): 903–47. Criscuolo, C., J. E. Haskel, and M. J. Slaughter. 2005. Global engagement and the innovation activities of firms. NBER Working Paper no. 11479. Cambridge, MA: National Bureau of Economic Research, July. Delgado, M. A., J. C. Farinas, and S. Ruano. 2002. Firm productivity and export markets: A non-parametric approach. Journal of International Economics 57 (2): 397–422. Doms, M. E., and J. B. Jensen. 1998. Comparing wages, skills, and productivity between domestically and foreign-owned manufacturing establishments in the United States. In Geography and ownership as bases for economic accounting, ed. R. E. Lipsey, R. E. Baldwin, and J. D. Richardson, 235–58. Chicago: The University of Chicago Press. Eaton, J., S. Kortum, and F. Kramarz. 2004. Dissecting trade: firms, industries, and export destinations. American Economic Review Papers and Proceedings 94 (2): 150–54. Economic Report of the President. 2005. U.S. Government Printing Oﬃce, Washington, D. C. 20402-0001. Grossman, G. M., and E. Helpman. 2004. Managerial incentives and the international organization of production. Journal of International Economics 63:237–62.

552

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

Grossman, G. M., E. Helpman, and A. Szeidl. 2006. Optimal integration strategies for the multinational firm. Journal of International Economics 70:216–38. Hanson, G. H., R. J. Mataloni, and M. J. Slaughter. 2004. Vertical production networks in multinational firms. Revision of NBER Working Paper no. 9723. Cambridge, MA: National Bureau of Economic Research, April. Jarmin, R., and J. Miranda. 2002. The longitudinal business database. Center for Economic Studies Working Paper 02-17. MacGarvie, M. 2003. Do firms learn from international trade? The Review of Economics and Statistics 88 (1): 46–60. Slaughter, M. J. 2004a. Globalization and employment by U.S. multinationals: A framework and some facts. Daily Tax Report, March 26. ———. 2004b. Insourcing jobs: Making the global economy work for America. Organization for International Investment, October. Zeile, W. J. 1997. U.S. intrafirm trade in goods. Survey of Current Business 77 (2): 23–38.

Comment

James Harrigan

Once upon a time, trade economists did not pay much attention to firms, and when they did they ignored within-industry heterogeneity. Bernard and Jensen challenged this orthodoxy in a series of influential papers that first appeared in draft form in the early 1990s. Using the Census Bureau’s Longitudinal Research Database, the Bernard-Jensen papers focused on the exporting behavior of plants, and found tremendous heterogeneity: a small minority of plants exported, and they diﬀered dramatically from nonexporters, with exporters generally being larger and more productive. Adding co-authors (including Peter Schott) along the way, the BernardJensen research program has continued and been taken up by many others, and with Marc Melitz’s seminal 2003 Econometrica paper the study of heterogeneity in export behavior was put into a tractable general equilibrium framework. As the theory and empirics of heterogeneous exporters advanced rapidly in the 1990s and early 2000s, the microfoundations of importing were almost entirely ignored. For example, in Melitz’s paper, the sophisticated and insightful treatment of exporters is complemented by the conventional, and utterly uninteresting, assumption that the demand for imports comes from the CES utility function of a representative consumer. I recount this brief intellectual history to help explain why this new chapter by Bernard, Jensen, and Schott is so important. It makes several contributions that should, and I think will, have a profound impact on how trade economists think. It should be required reading (or at least skimJames Harrigan is a professor of economics at the University of Virginia and a research associate of the National Bureau of Economic Research.

Importers, Exporters, and Multinationals

553

ming) for graduate international trade courses, and textbook authors should incorporate its findings into their next editions. There is much that is new in this chapter, but the newest is that it identifies imports with firms. As the authors note, prior to their paper there was “virtually no research documenting and analyzing importing firms.” I would add that there is virtually no theory that takes importing firms seriously, with the partial exception of the literature on vertical multinationals (e.g., Markusen 2002, Chapter 9). If importing firms do nothing more than add some retail and/or wholesale services to imported goods before handing them over to consumers, then thinking carefully about importing firms might not be important. However, this type of importing firm accounts for less than a third of total imports: the bulk of imports are bought by firms in goods or services categories (table 14.9). This suggests that the direct demand for imports comes largely from the production process, and the modeling of imports should reflect this. Before proceeding further, it is worth defining what it means for a firm to trade, and examining how closely the measurement categories of the chapter match up with what we might ideally want to know. A firm is deemed an exporter if it produces a good that is sold to a foreign customer. Of course, manufacturers buy intermediate inputs as well as hiring labor and capital. There are several implications of this observation: • Attributing the gross value of an export transaction to a final product firm may be misleading about which firms are globally engaged. • Export involvement may be less concentrated than the chapter’s figures suggest. • The measurement of export involvement is sensitive to the organization and boundaries of firms. To illustrate my point, consider Boeing, the largest U.S. exporter. Boeing purchases parts from U.S. suppliers, yet these parts suppliers will not be counted as exporting firms unless they sell parts directly to foreign buyers. Similarly, if Boeing buys services from accounting and legal firms these firms will not be counted as globally engaged. The point is not that Boeing’s parts suppliers or accounting firm should be counted as exporting firms, but simply to observe that the correct measurement depends on the question being asked. For importing firms, many of the same concerns apply, and the appropriate definition is even more problematic. In the data, a firm is an importer if it purchases a product from a foreign supplier, with no reference to what happens next. As with the definition of an exporting firm, such a classification probably understates the extent of global engagement, perhaps dramatically. For example, big retailers such as Wal-Mart buy directly from foreign suppliers, while smaller retailers are more likely to buy from a distributor or broker. Thus, an electronics retailer that primarily sells im-

554

Andrew B. Bernard, J. Bradford Jensen, and Peter K. Schott

ported goods will not be recorded as an importer even if its wholesale suppliers buy exclusively from foreign manufacturers and do no more than add a small markup. With the above caveats in mind, the most striking finding of this chapter is the extraordinary skewness in trading behavior: a small minority of firms import and export, and they are big. Skewness in firm size is a longestablished fact (see, for example, Axtell 2001), but what is remarkable about this chapter is that what the authors call global engagement is even more skewed than size. This is illustrated in table 14.3: in 2000, the top 1 percent of firms employed 14 percent of the labor force and conducted 80 percent of trade. They also show that the vast majority of trade is conducted by huge multinationals that trade multiple products (table 14.4) with multiple countries (table 14.6) and conduct much of their trade with their aﬃliates (table 14.14), often in developing countries (tables 14.12 and 14.16). The implication is clear: if you want to think about U.S. trade, you need to think about big multinationals. The dimensionality and scale of the data set assembled by the authors is nearly overwhelming. Dimensions of variations include • time (1992–2000) • firms (5 million!) • sectors (manufacturing, wholesale/retail, etc.) • types of firms (most globally engaged, other) • trading partners (Mexico, China, Europe, . . .) • products (within firms, one to ???) • type of transaction (arms length, related party, . . .) • size of firms (corner store to Wal-Mart) • mode of shipment (air, surface) In contemplating what questions can be addressed with such a data set, it is useful to think about what questions can be asked with more aggregate (and more publicly available) data. For example, as the authors note, trends in within-firm trade and multinational activity have been usefully analyzed by previous authors using Bureau of Economic Analysis data. Similarly, variation in highly disaggregated import and export data has recently been analyzed by a number of authors, including Schott (2004), Hummels and Skiba (2004), Hallak (2004), and (if I may be permitted some self-promotion) Harrigan (2005). For researchers, the key innovation in the data set is the linkage between product-level trade and individual firms. This is the link that permits identification of what I take to be the major finding of this chapter, the extraordinary concentration of U.S. trade in a tiny fraction of firms. The firm-trade link will also permit the analysis of important issues such as transfer pricing (an opportunity already taken up by the authors in a paper-in-progress), exchange rate pass-through, and economies of scope that it was not possible to study with earlier data sets.

Importers, Exporters, and Multinationals

555

Economic analysis of international trade is an intellectually healthier field than it was twenty years ago because of its embrace of empirical work and its receptiveness to surprising empirical findings. This paper has already been met with great interest, and just as the earlier Bernard-Jensen work has been influential, I expect this new Bernard-Jensen-Schott chapter to be the stimulus for exciting future research. References Axtell, R. L. 2001. Zipf distribution of U.S. firm sizes. Science 293: 1818–20 (7 September 2001). Hallak, J. C. 2004. Product quality, linder, and the direction of trade. NBER Working Paper no. 10877. Cambridge, MA: National Bureau of Economic Research, October. Harrigan, J. 2005. Airplanes and comparative advantage. NBER Working paper no. 11688. Cambridge, MA: National Bureau of Economic Research, September. Hummels, D., and A. Skiba. 2004. Shipping the good apples out? An empirical confirmation of the Alchian-Allen conjecture. Journal of Political Economy 112 (6): 1384–1402. Markusen, J. R. 2002. Multinational firms and the theory of international trade. Cambridge, MA: MIT Press. Melitz, M. J. 2003. The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica 71 (6): 1695–1725. Schott, P. K. 2004. Across-product versus within-product specialization in international trade. Quarterly Journal of Economics 119 (2): 647–78.

15 The Impact of Trade on Plant Scale, Production-Run Length, and Diversification John Baldwin and Wulong Gu

15.1 Introduction This chapter examines the impact of trade on product diversification and plant size. The issue has dominated discussions on potential benefits of trade liberalization in Canada. Operating behind tariﬀ barriers and limited market size, Canadian plants have been described as having production runs that were too short to exploit economies of large-scale production. Tariﬀ reductions were predicted to reduce product diversification at the plant level and to improve the length of production runs. However, there is little empirical evidence on the link between tariﬀ reductions and increases in product specialization. This chapter attempts to fill this research gap. Shorter production runs can arise either from suboptimal plant size or excessive product line diversity. Earlier studies by Daly, Keys, and Spence (1968) and Caves (1975) argued that Canadian plants suﬀered from excessive levels of diversity. And a number of Canadian studies have attributed lower productivity to shorter production runs. For example, Safarian’s survey on the relative costs of foreign multinationals operating in Canada (1966, ch. 7) reported that most foreign aﬃliates operating in Canada had higher unit costs than parent companies’ plants located in the United John Baldwin is director of the Microeconomic Studies and Analysis Division of Statistics Canada. Wulong Gu is chief of research in the Microeconomic Studies and Analysis Division of Statistics Canada. We would like to thank Tim Dunne, J. Bradford Jensen, Mark Roberts, and James Tybout for helpful comments. Views expressed in this chapter do not necessarily reflect those of Statistics Canada.

557

558

John Baldwin and Wulong Gu

States. These higher costs were attributed by the firms to a variety of sources, but shorter production runs was the most common response for those reporting higher unit costs. In the same vein, a study by Scherer et al. (1975) reported that Canadian textile makers claimed that their unit costs on style-sensitive dress goods and decorative fabrics were 20 to 30 percent higher than the costs of comparable U.S. manufacturers, primarily because of a ten-fold diﬀerence in market size and the attenuated but still substantial diﬀerences in lot sizes. Paint manufacturers reported that average batch sizes in Canada were onefifth to one-half those experienced in the United States. Both the Canadian Economic Council (1967, 1975) and the Royal Commission on Corporate Concentration (1978) predicted that the lowering of Canadian tariﬀ barriers would increase Canadian average plant size and that it would reduce product diversity at the plant level and improve the length of production runs. Starting in 1989, two major changes occurred in the trading environment that faced Canadian manufacturers that should have influenced the length of production runs. First, the Canada-United States Free Trade Agreement (FTA) guaranteed a new type of open-border arrangement between these two countries. Then the North American Free Trade Agreement (NAFTA) in 1994 brought together Canada, Mexico, and the United States. These agreements continued a process that extended back to the post-World War II commitments to reduce tariﬀs and expand international trade. The average tariﬀ collected continued its downward trend during the 1990s—from 3.3 percent in 1989 to 1.1 percent in 1996. But the FTA and NAFTA changes marked a turning point in that they set a timetable for the elimination of tariﬀs and a framework for the resolution of trade disputes that was intended to give companies greater certainty for foreign direct investment. The result was an increase during the 1990s in both the export intensity and the import intensity of the Canadian manufacturing sector. Both export intensity and import intensity increased from around 31 percent in 1990 to 47 percent in 1997. The FTA allowed a process that had begun in the 1970s and 1980s to continue into the 1990s. Manufacturing activity shifted from primarily facing import competition to being more exportoriented; this transition provided the link between trade liberalization and the expected impact of increased market size on diversity. The importcompeting segments of Canadian manufacturing may also have responded to trade liberalization, as there would be increased competition in an enlarged domestic market. Previous empirical work suggests that trade liberalization in the early 1990s might have been expected to increase plant specialization. Earlier studies by Baldwin and Gorecki (1983b, 1986) made use of data for the 1970s to study whether the reduction in tariﬀs that occurred following the

Trade and Plant Scale, Production-Run Length, and Diversification

559

Kennedy round was associated with an increase in plant specialization. During this period of gradual tariﬀ reductions, plant specialization increased slightly, as did the length of the production run. Increases in the latter, though not the former, were greater in those industries where tariﬀs declined the most. Baldwin, Beckstead, and Caves (2001) examined longerrun trends in both firm and plant specialization. This chapter extends our work that examines trends in specialization in the Canadian manufacturing sector. We have two objectives. First we develop a model of trade in diﬀerentiated goods with multi-product plants to structure our analysis. The model contributes to the recent development of firm-based models that highlights diﬀerences in the responses of individual firms to trade policies (Bernard et al. 2003; Melitz 2003; Yeaple 2002). Second, we provide empirical evidence on the model’s prediction regarding the impact of tariﬀ reductions on product diversification, productionrun length, and plant size using Canadian experience over the 1980s and the 1990s. Melitz (2003) has developed a model of trade in diﬀerentiated products with producer heterogeneity to examine the eﬀect of trade on firm/plant turnover (entry, exit, and output reallocation) in domestic and export markets. Melitz and Ottaviano (2005) examine the eﬀect of market size on firm size, firm productivity, and firm turnover. In this chapter, we develop a model of trade with multi-product firm/plants to examine the eﬀect of market size and trade on product specialization and production-run length. Our model generates a number of predictions on the eﬀect of market size and trade integration on product specialization, production-run length, plant size, and plant turnover in domestic and export markets. The most novel finding relates to the eﬀect of market size and trade on product diversification, production-run length and plant size. Our model predicts that plants in a smaller market tend to be more diversified and have shorter production runs. Bilateral trade liberalization reduces the number of products supplied by plants, and the rate of decline is smaller for larger and exporting plants. It increases the production-run length of exporters while having no eﬀect on the production-run length of nonexporters. The eﬀect of bilateral tariﬀ reductions on plant size depends on the export status of a plant. Bilateral tariﬀ cuts reduce the plant size of nonexporters as nonexporters reduces the number of products while keeping the production-run length unchanged. The eﬀect of tariﬀ cuts on the plant size of exporters is ambiguous. On the one hand, tariﬀ cuts increase the plant size of exporters by increasing the production-run length of the portion of the product line that is exported. One the other hand, tariﬀ cuts reduce the plant size of exporters by reducing the total number of products produced. The net eﬀect of bilateral tariﬀ cuts on plant size depends on the size of those two oﬀsetting factors. The predictions of our model on the eﬀect of trade and market size on

560

John Baldwin and Wulong Gu

plant size, plant productivity, and plant turnover are similar to those in Melitz (2003) and Melitz and Ottaviano (2005). First, plants in a smaller and less competitive market tend to be smaller and less productive than those in a larger and more competitive market. These predictions are similar to those in Melitz and Ottaviano (2005) and have been confirmed in a number of previous empirical studies (Scherer et al. 1975; Caves 1975; Syverson 2003). Second, tariﬀ barriers induce only the most productive plants to enter the export market. As trade costs fall, the least productive plants exit and the most productive of nonexporters enter the export market and expand their output. In our empirical analysis, we focus on the model’s prediction on the effect of bilateral tariﬀ cuts on product diversification, production-run length and plant size. To this end, we use a sample of Canadian manufacturing plants in the 1980s and 1990s. The Canadian experience with tariﬀ reductions as a result of the 1989 Canada-U.S. Free Trade Agreement (FTA) and its extension to Mexico provides us with an opportunity to examine how the plants in a market of limited size respond to trade liberalization. The Canada–U.S. Free Trade Agreement (FTA) committed two countries to gradually eliminate all manufacturing tariﬀ rates over a ten-year period beginning in 1989. The tariﬀ reductions in the two countries are highly correlated (Head and Ries 1999). In addition, the political economy that governed tariﬀ reductions has produced similar cross-industry reductions in the two countries that make it diﬃcult to separate out the eﬀect of each set of tariﬀ reductions. As such, the Canada–U.S. tariﬀ cuts resemble the case of bilateral trade liberalization examined in the model. 15.2 A Model of Closed Economy In this section, we will develop a model of a closed economy to examine the eﬀect of market size on product diversification and firm size. The model also serves as a building block for the open-economy model that will be developed in the next section. It is similar to the one in Melitz and Ottaviano (2005) with one distinction. Here we assume multi-product firms while Melitz and Ottaviano (2005) assume single-product firms. 15.2.1 Demand Consider an economy with L identical consumers. The consumer’s preferences are described by a quasi-linear utility function that is defined over a continuum of diﬀerentiated varieties, and a homogeneous good chosen as numéraire: (1)

1 1 q()d q()2d 2 ∈ 2 ∈

U

q .

∈q()d

2

O

Trade and Plant Scale, Production-Run Length, and Diversification

561

where qo and q() represent the individual consumption levels of the numéraire good and variety . The set of varieties supplied by firms is . The demand parameters , , and are all positive. The parameter indexes the degree of product diﬀerentiation between the varieties. The degree of product diﬀerentiation increases with as consumers give increasing weights to the dispersed consumption of the varieties. An increase in implies a decline in substitutability between the varieties, thus limiting the response of a consumer’s consumption pattern over the varieties to changes in the price of particular variety. In the limit when 0, the varieties are perfect substitutes and the consumers care only about their total consumption level over the varieties ∫ ∈ q()d. The parameters and indexes the substitution between the diﬀerentiated varieties and the numéraire. Increases in and decreases in increase the demand for the diﬀerentiated varieties relative to the numéraire. Each consumer is endowed with one unit of labor. The budget constraint for the consumer can be written as:

∈ p()q() d q

(2)

o

w

where w is the wage and p() is the price of variety . Solving (2) for the numéraire consumption, substituting the corresponding expression into (1), and solving the first order conditions with respect to q(), yields the inverse demand for variety supplied by firm i: pi() qi() Q

(3)

where Q ∫ ∈M ∫ ∈ qi()ddi is the total market demand of the diﬀerentiated product. The total market demand for variety of firm i can be expressed by the inverse demand function: i

(4)

Q qi() pi() . L L

The quasi-linear utility function (1) we choose in our model has a desirable feature that the elasticity of demand is not fixed. Instead, it is related to the intensity or toughness of competition. Increases in the toughness of competition due to a larger market (L), a lower degree of product diﬀerentiation () leads to increases in the elasticity of demand. In contrast, the C.E.S. preferences used in previous studies (e.g., Melitz 2003) yield a demand system in which the price elasticity of demand is constant. Though convenient from the analytical point of view, such a result is at odds with empirical findings that more intensive competition is associated with a higher elasticity of demand (Campbell and Hopenhayn 2002; Greenhut, Norman, and Hung 1987; Roberts and Tybout 1996; Syverson 2003; Tybout 2003).

562

John Baldwin and Wulong Gu

15.2.2 Production and Firm Behavior To examine the impact of trade and market size on product diversification, we depart from previous monopolistic competition models of trade in diﬀerentiated products. In all those models, production exhibits economies of scale within varieties but no economies of scope across varieties. As such, each firm supplies one variety, and there is a one-to-one relationship between firms and varieties. In our model, we assume that production exhibits economies of scale within varieties but economies of scope across varieties. To enter the diﬀerentiated product sector, a firm must bear fixed costs of entry E regardless of the size of its product range, thus implying that economies of scope are present. An entrant then learns about the marginal cost of the production of a variety. We assume that this is drawn from a common distribution G(c) with support on [0,cM] and it is the same across varieties within a firm. The production technology of a variety requires fixed overhead costs F in order to produce any amount of a variety, thus implying economies of scale within varieties. We assume that this overhead cost is known and it is the same across all varieties. As the entry cost is sunk, an entering firm would immediately exit if its profit gross of entry costs were negative. The surviving firm first chooses its product range, then, the quantity and price of each variety it supplies. Let M be a given number of multiproduct firms. Let i R denote the set of varieties produced by firm i ( 1, . . . , M ) and qi() the quantity of variety . The total production cost of firm i is given by Ci

(5)

[ci qi () F ]d

∈i

and the total revenue is Ri

(6)

pi ()qi ()d.

∈i

Firm i maximizes its profit (7)

Πi

[pi ()qi () ci qi () F ]d

∈i

where the demand for variety is defined in equation (4). Because we have symmetry among varieties with each firm’s product line, the quantity and price that a firm chooses is the same across its varieties. In other words, we have pi() pi and qi() qi for the varieties supplied by firm i. The strategic behavior of surviving multi-product firms has been studied in Ottaviano and Thisse (1999). The rest of this section follows closely the

Trade and Plant Scale, Production-Run Length, and Diversification

563

analysis in that paper.1 Ottaviano and Thisse (1999) argue that firms should behave like oligopolists as those firms are large actors and control a nonnegligible set of varieties. When choosing its product range and the length of production runs, a firm no longer neglects its impact on the market as in monopolistic competition models of trade.2 The firm must account for the impact of its choice on the demand for its varieties through its eﬀect on total market demand Q, which is the sum of the demand for the varieties of firm i and those of its competitors (Q–1). These discussions suggest that the total market demand is: (8)

Q qi i Qi

and the profit of firm i can be rewritten as: (9)

Πi ( pi qi ci qi F )i,

and the inverse demand (4) becomes: (10)

pi qi Q, Q qi i Q1. L L

This is a two-stage game. A firm chooses its product range i in the first stage and then the quantity and price of its varieties pi and qi in the second stage. The solution of the second stage subgame is obtained from the diﬀerentiation of the profit function with respect to qi. Solving for these firstorder conditions, we have the optimum output and price of each variety provided by firm i: (11)

( ci )L Qi qi

, 2( i)

and (12)

( ci)L Qi pi

. 2L

These results show that the firms in a larger market choose longer production runs and set lower prices for their products as a result of higher demand elasticity for their products. Substituting (11) and (12) into (9) yields the second-stage equilibrium profit of firm i: 1. But there is a diﬀerence. Ottaviano and Thisse (1999) assume that firms are identical and have the same marginal cost. We introduce firm heterogeneity and assume that the marginal cost of producing a product is drawn from a common distribution. 2. In monopolistic competition models of trade in diﬀerentiated products, each firm produce one variety as there is no economies of scope across varieties. In these models, each firm correctly neglects its impact on the market.

564

(13)

John Baldwin and Wulong Gu

[( ci)L Qi]2 Πi

i Fi . 4L ( i)

The expression (13) describes the payoﬀ of firm i in the first stage game. To find the solution of the second stage subgame, we diﬀerentiate (13) with respect to i and obtain the first order conditions for the equilibrium product range i.3 (14)

[( ci)L Qi] ( i)

2

. FL

Equations (11), (12), and (13) provide a unique solution (pi, qi, i) for M firms. For the rest of the section, we will obtain an analytical solution for (pi, qi, i). The results will be used to conduct a comparative analysis on the impact of market on firm size and product diversification. Substituting the expression for ( i) in (14) into (11) gives the equilibrium output of each variety supplied by firm i: (15)

q∗i

FL

q∗.

This shows that the lengths of production runs are the same across individual products within a firm. Furthermore, it is the same across all firms. This implies that the sum of the output Q–i for the varieties of firm i’s competitors can be written as q∗( – i), where ΣM i1i is the total number of varieties in the market. The first order condition (14) can be rewritten as: (16)

[( ci)L q∗( i)] ( i)

2

. FL

Summarizing (16) over all firms and solving for the total number of varieties :

(17)

L ( c)M

2M F ∗

(M 1)

where c Σici /M is the average cost of M firms. Substituting (17) into (16) and solving for i yields the equilibrium product range supplied by firm i:

(18)

L [( ci) M(c ci)]

2 F ∗i ∗(ci)

. (M 1)

3. The payoﬀ function (13) is concave in i. Therefore, the equilibrium product range implicit in (14) is unique maximum.

Trade and Plant Scale, Production-Run Length, and Diversification

565

Substituting the expressions (15), (17), and (18) for q∗i , ∗ and ∗i into (13) gives the maximum profit of firm i: (19)

L

2 .

F

F Π∗(ci)

2 [ M c (M 1)ci ] (M 1)

2

Finally, solving (14) for Q–1 and substituting the resulting expression into (12), we obtain the equilibrium price of each variety supplied by firm i: ∗i FL (20) p∗i ci

. L

This implies that firms use an absolute markup instead of relative markup when choosing prices. In sum, we have derived the analytical solutions for the number of varieties ∗(ci), the quantity q∗i and price p∗i of each variety, the maximum profit Π∗(ci) for each of the M firms. These results show that (a) firms in a larger market have longer production runs for individual products; (b) product diversification declines with the economies of scale within individual products (or increases in fixed overhead costs F ); (c) firms with lower costs set lower price, earn higher profits, and are larger. 15.2.3 Free Entry Equilibrium in a Closed Economy After entering a market by making an initial investment E, a firm learns about the marginal cost of the production of variety. Let cD denote the cost of a firm who earns zero profits. All firms with costs above the cutoﬀ cost cD would make negative profits and choose to exit. All firms with cost level below cD earn positive profits and remain in the market. The cutoﬀ cost cD is determined by the zero profit condition: (21)

Π∗ (cD) 0, or [ M c (M 1)cD]

L

2 0 F

where c ∫c0D cdG(c)/G(cD) is the average cost of surviving firms, and G(cD) is the survival rate of entering firms. We can now determine the number of firms M in equilibrium. Before enc tering the market, the expected profit is ∫0D Π∗(c)dG(c) – E, where Π∗(c) is given in (19). If this profit were positive, more firms would enter. Therefore, the number of firms in equilibrium must satisfy the following condition: c

(22)

0 DΠ∗(c)dG(c) E 0.

For the rest of the chapter, we will assume that productivity draws 1/c follow a Pareto distribution with lower productivity bound 1/cM and shape parameter k 1. This implies a distribution of cost c:4 4. The logarithm of labor productivity log(1/c) follows an exponential distribution with a standard deviation equal to 1/k.

566

(23)

John Baldwin and Wulong Gu

c k G(c) , c ∈ [0, cM]. cM

When k 1, costs follow a uniform distribution. An increase in k implies a decline in the dispersion of the costs. Solving the zero profit and free entry conditions (21) and (22) yields the solutions for cD and M: (24)

(25)

E cD ckM(k 1)(k 2)

2L

1/(k 2)

, and

F cD 2

L M (k 1)

. cD

These results show that there are more firms in a larger market. The cutoﬀ cost in a larger market is lower and the exit rate for entrants (equals 1 – G(cD)) is higher as competition is more intense in the larger market. Given these expressions for cD and M, the performance measures of firm i in (15), (18), (19), and (20) can be rewritten as: (26)

(cD ci) ∗(ci)

L

, F

F

, L

q∗i

p∗i cD

FL

,

L Π∗(ci) (cD ci)2. And the average performance measures across all firms can be written as: (27)

L

, F F FL

, q∗

, p∗ c L cD ∗

(k 1) D

2c2D L ∗

Π (k 1)(k 2) The total number of product varieties is: (28)

L

2 . F

1 ∗ ( cD)

Compared with an average firm in a smaller market, the one in a larger market supplies a larger number of varieties (with a higher degree of prod-

Trade and Plant Scale, Production-Run Length, and Diversification

567

uct diversification). It has a longer production run and sets a lower price for its product varieties. It is larger and more productive, and has higher profits.5 There are more product varieties and more firms in a larger market. Equations (27) also provide intuitive results on the impact of scale and scope economies on product diversification, production run length, firm size, and firm profits. The existence of strong scale economies within individual products (high F ) is related to higher product specialization, longer production run length, and higher profits. However, it has no eﬀect on firm size and productivity. The existence of strong scope economies at the firm level (high E) is related to higher product diversification, larger firm size, lower productivity, and higher profits. But it has no eﬀect on the lengths of production runs for individual products. The result relating to the degree of product diﬀerentiation () is straightforward. A low degree of product diﬀerentiation leads to narrow product lines, long production runs, low price and low profits. It has no eﬀect on firm size and productivity. 15.3 A Model of Open Economy In this section, we examine the impact of trade on product diversification and firm size. We will consider two economies of the type that was examined in the last section. We assume that two economies are integrated through trade with positive trade cost. If the two economies are perfectly integrated and there are no trade costs, trade allows individual countries to replicate the outcome of an integrated world as in the model of section 15.2.1. 15.3.1 Model We now consider two economies h and f where there are trade costs. To simplify our analysis, we assume that the two countries are symmetric. Each country has L consumers. Trade costs are modeled in the standard iceberg formulation, where 1 units of a good must be shipped in order for one unit to arrive at destination. The firms in the two markets are of the type modeled in section 15.2. To enter, a firm must first make an irreversible investment E. The firm then learns about the cost of the production of a variety that is drawn from a common distribution. After learning about the cost, the least productive firms choose to exit. The more productive firms choose to remain in the domestic market. These firms will also have to decide whether to serve the ex5. Firm size is defined as the real output of the firm that is equal to the number of varieties times the output of each variety.

568

John Baldwin and Wulong Gu

port market at the same time. All these remaining firms will then choose their product range, the price and quantity of a variety for the domestic market and for the export market if they also decide to serve the export market. As in Melitz (2003), we assume that there is no additional uncertainty for the decision to enter the export market. The firms maximize the sum of profits earned from domestic and export sales. As the markets are segmented, the firms must maximize the profits from domestic sales and from export sales. The results in the section 15.2.1 show that the number of varieties D(c), the quantity and price of each variety qD(c) and pD(c), and profits ΠD(c) for a firm that produces for the domestic market can be written as:

(29)

L [( c) M(c c)]

2 F D(c)

, (M 1) qD(c)

FL (c) FL

, p (c) c

L D

D

L

2

F

F ΠD(c)

2 [ M c (M 1)c] (M 1)

2

where M is the total number of firms that sells in an economy that includes both domestic firms and foreign exporters that sell in the country. For the firms that sell in a foreign market, number of varieties X (c) supplied for the export market, the quantity and price of each variety qX (c) and pX (c), and the profits ΠX (c) can be rewritten as:

(30)

L [( c) M(c c)]

2 F X (c)

, (M 1) qX (c)

FL (c) FL

, p (c) c

L X

X

L

2

F

F ΠX (c)

2 [ M c (M 1) c] (M 1)

2

where c is the delivered cost of exporters. Upon entry and learning about its cost, a firm with cost below cD makes positive profits and stays in the market. Otherwise the firm will exit. The firm with cost below cX will enter the export market. The cutoﬀ cost levels

Trade and Plant Scale, Production-Run Length, and Diversification

569

cD and cX are determined from zero profit conditions for domestic sales and export sales: L

2 0,

F F L

2 0. Π (c ) 0:

[ M c (M 1) c ]

F (M 1)

F (31) ΠD(cD) 0:

2 [ M c (M 1)cD] (M 1)

2

2

X

X

X

2

Equations in (18) show that the two cutoﬀ cost levels satisfy the condition: cD cX .

(32)

As 1, we have cX cD. The two cutoﬀ cost levels provide a portioning of firms into exiting, nonexporting, and exporting firms. The least productive firms, those firms with cost above cD exit the market. The firms with cost between cX and cD produce exclusively for the domestic market. The most productive firms with the cost below cX enter the export market and produce for both domestic and export markets. Given the relationship between the cutoﬀs for domestic and foreign sales in (31), the cost of surviving domestic firms c ∈ [0, cD] and the delivered cost of exporting firms c ∈ [0, cX] have identical distributions. The average cost of all firms that sell in a market (that includes domestic firms and foreign exporters) is: cD

c cDG(c). 0

(33)

Free entry drives the expected profit to zero: c

(34)

0 D Π

cX

(c)dG(c) ΠX (c)dG(c) E 0.

D

0

Solving for cD and cX, we have: (35)

E cD ckM(k 1)(k 2)

2L(1 k)

1/(k 2)

cD E cX ckM(k 1)(k 2)

2L( k 2 2)

,

1/(k 2)

.

The results show that a reduction in trade costs leads to a decline in cD and an increase in cX. As tariﬀ barriers fall, the least productive firms exit. Of the remaining nonexporters, the more productive enter the export market. Using the zero profit conditions (31), the product range and the price and quantity of each variety supplied by a firm in the domestic market in (29) can be rewritten as:

570

(36)

John Baldwin and Wulong Gu

L

, F FL F

.

, p (c) c q (c) L (cD c) D(c)

D

D

D

Similarly, the product range and the price and quantity of each variety supplied by a firm in the foreign market can be rewritten as: (37)

L

, F FL F

.

, p (c) c q (c) L (cD c) X (c)

X

X

D

We have X(c) D(c). For a firm that produces for both domestic and export markets, the product range supplied for the domestic market is wider than the one supplied for the export market. An exporting firm always exports a subset of its product varieties to the foreign market. 15.3.2 The Comparative Statistics of Bilateral Trade Liberalization Our model generates a number of testable implications on firm size and product diversification of bilateral trade liberalization, or the decline in common trade cost in the two countries. We will focus on the case of bilateral trade liberalization as the Canada-U.S. FTA tariﬀ cuts should be more appropriately modeled as a case of bilateral liberalization.6 The Canada-U.S. FTA committed the two countries to eliminate manufacturing tariﬀs in a ten-year period beginning in 1989. The tariﬀ rates are similar in level and their changes over time are highly correlated in the two countries. In addition, the political economy that governed tariﬀ reductions has produced similar cross-industry reductions in the two countries that make it diﬃcult to separate out the eﬀect of each set of tariﬀ reductions. The Eﬀect on the Number of Products The total number of products that a firm produces is given by (36). The expression (36) for a firm’s product range shows that the number of products is a negative function of tariﬀ rates. A lower tariﬀ rate reduces the number of products supplied by firms. In addition, the marginal eﬀect of tariﬀ cuts on log changes in the number of products decline with c. As tariﬀ rates fall, the rate of decline in the number of products should be smaller 6. An important extension of the model is to examine the implications of unilateral trade liberalization. The eﬀect of unilateral liberalization and other industrial and trade policy has been the focus of an extensive literature (see, e.g., Helpman and Krugman 1989).

Trade and Plant Scale, Production-Run Length, and Diversification

571

for firms that are larger and exporters. We have the first testable implication for product diversification from our model: HYPOTHESIS 1. A decline in tariﬀ rates is related to a decline in the number of products supplied by individual firms. The decline is smaller at exporting and larger firms than at nonexporting and smaller firms. The Eﬀect on the Index of Product Diversification In our empirical section, we will use an entropy index to measure product diversification. The entropy index of product diversification is defined as E Σi1si log(1/si), where is the number of products and si is the share of a product. The index of product diversification of nonexporters is ln(D)—the number of products in log, where D is given by (36). This will decline as tariﬀ rates fall. For exporters, tariﬀ changes have an ambiguous eﬀect on the product diversification index. On the one hand, exporters produce a smaller number of products. On the other hand, exporters expand the range of products that are shipped abroad. The former leads to a decline in the index of firm diversification while the latter leads to an increase in the index of firm diversification. These discussions provide the second testable implication from the model: HYPOTHESIS 2. A decline in tariﬀ rates reduces the product diversification index of nonexporting firms. It has an ambiguous eﬀect on the product diversification index of exporting firms. The Eﬀect on Firm Size We define firm size as real output calculated as the number of products times the output of each product. The size of nonexporters is DqD, where D and qD are given by (36). The size of nonexporters declines with lower tariﬀ rates. The size of exporters is DqD X qX. The decline in tariﬀ rates reduces D, increases X, and has no eﬀect on qD and qX at exporters. This suggests that tariﬀ reductions increase export sales and lowers domestic sales at existing exporters. The overall eﬀect of tariﬀ cuts on the size of exporters depends on the relative magnitude of those two oﬀsetting factors. These discussions provide a third testable implication from our model: HYPOTHESIS 3. A decline in tariﬀ rates reduces the size of nonexporters. It has an ambiguous eﬀect on the size of exporters. The Eﬀect on Production-Run Length The production-run length of individual products for nonexporters is qD in (36), which is independent of tariﬀ changes. The exporters improve the

572

John Baldwin and Wulong Gu

production-run length of the products that they begin to export as a result of lower tariﬀs. We have a fourth implication from our model: HYPOTHESIS 4. A decline in trade costs increases the production-run length of exporters and has no eﬀect on the production-run length of nonexporters. In addition to its prediction on the eﬀect of tariﬀ cuts on product diversification, plant size and production-run length of existing exporters relative to nonexporters, our model has implications for the entrants to the export market. Tariﬀ cuts will reduce the product diversification index and increase the production-run length of entrants to the export market compared with nonentrants to the export market. The eﬀect of tariﬀ cuts on the size of entrants to the export market depends on the magnitude of two oﬀsetting factors: increased export sales and the reduced product ranges. A proof of these results is similar to the one for our results on the responses of exporters versus nonexporters as a result of tariﬀ cuts. The implication of bilateral tariﬀ cuts on firm turnover in domestic and export markets are similar to those in the Melitz model of trade (Melitz 2003). As tariﬀ rates fall, the least productive firms exit and the most productive of nonexporters enter the export market. Current exporters increase export/shipment ratios with lower tariﬀ rates. This is a result of a decline in domestic shipments and an increase in foreign shipments at current exporters. These predictions have been confirmed in a number of previous empirical studies (Bernard, Jensen, and Schott 2003; Baldwin and Gu 2004; Bernard et al. 2003).7 15.4 Data The empirical analysis will be carried out at the plant level. The data used for the analysis come from a longitudinal data file on all plants in the Canadian manufacturing industry over the period 1973 to 1997. This longitudinal file is based on data that are derived from both survey and administrative sources that provide plant-level data for the universe of plants in the manufacturing sector. The survey data are derived from long-form questionnaires (generally filled in by the largest plants) that contain the most detailed information, including commodity data and short-form questionnaires (generally filled in by smaller plants) that are much less detailed. In addition, for the very smallest plants, administrative data on sales and employment come from tax records. In this database, a plant’s sales are classified to one industry. Each plant is identified as being part of a firm. Detailed information at the plant level includes the 1980 SIC, employment, value of shipments and value added, 7. Tariﬀ reductions have a bigger impact on the export/shipment ratios of exporters for the industries with a larger dispersion of productivity levels (Helpman, Melitz, and Yeaple 2004)

Trade and Plant Scale, Production-Run Length, and Diversification

573

nationality of control, age of plant, exports, the SIC of the industry to which the plant is classified, and whether the owning firm possess multiple plants. Information on export status is also available for plants that are given a long-form (detailed) questionnaire for the years 1979, 1984, 1990, 1993, 1996, and 1997. In addition, annual commodity data for all products produced (both primary and secondary) are available for all plants that received a long-form questionnaire. The survey collects data on the value of shipments and quantity of each commodity produced in these long-form plants. We use these commodity data to calculate an index of diversity across commodities for plants. In this chapter, we use a diversification measure that takes into account both the number of commodities that a firm produces and the distribution of its activity across commodities. The commodity dimension utilizes over 7,000 commodities. We use an entropy measure of product diversification that measures how concentrated a plant’s sales are at the product level (see Jacquemin and Berry 1979). The entropy diversification index takes a value of zero when sales are concentrated within a single product line. At the other extreme, if the plant’s activity is spread evenly across products, the plant’s entropy is maximized at E(s) log(). Production-run length is defined as plant production divided by number of products. We also experimented with an alternative—production divided by the numbers equivalent derived from the entropy diversification measure.8 The results were similar. In our model, we have considered the case of symmetric bilateral trade liberalization where tariﬀ reductions are symmetric in the two countries. Our previous discussion suggests that tariﬀ cuts in Canada and the United States resemble symmetric bilateral trade liberalization, particularly during the FTA period. In our empirical analysis, we will use as independent variable the sum of Canadian tariﬀ reductions against U.S. imports and U.S. tariﬀ reductions against Canadian exports. The coeﬃcient on the combined tariﬀ cuts should capture the model’s prediction on the eﬀect of bilateral tariﬀ cuts. The Canadian tariﬀ rates against U.S. imports are based on duties paid that are collected by commodity. These commodities are assigned to industries based on the primary industry of production. Average industry tariﬀs are then calculated using import values as weights. The U.S. tariﬀ rates against Canadian imports are once again based on import duties by commodity, which are assigned to an industry using the same Canadian 8. This is derived from the entropy measure of diversification by taking its antilog, which is referred to as the numbers-equivalent entropy. Its values are bounded between one and K: it equals one when 100 percent of a plant’s activity is in one commodity and it equals K when a plant’s production is spread equally across K products.

574

John Baldwin and Wulong Gu

concordance table used for Canadian commodity duties, and then aggregated to industries based on U.S. import weights.9 15.5 Empirical Results In this section, we provide empirical evidence on the eﬀect of tariﬀ rates on product diversification, production-run length, and plant size as summarized in the four hypotheses in section 15.3. We estimate the following specification that expresses changes in product diversification, production-run length, or plant size as a function of tariﬀ changes, export status, plant size, and a set of plant characteristics: (38)

Ypt i i 1 it 2Ept1 3Spt1 4[Ept1 it] 5[Spt1 it] 6Xpt εpt

where denotes the change between periods t – 1 and t, Ypt is the dependent variable denoting the number of products in log for plant p during period t, the index of product diversification, the output of a plant in log, or the average length of production runs for individual products in log; it is the average annual change in tariﬀ rates; Ept–1 is a variable indicating whether the plant is an exporter in period t – 1; Spt–1 is relative plant size; Xpt is a set of plant characteristics that includes the value of the dependent variable in period t – 1 (Ypt–1), a variable indicating whether a plant entered the export market between t – 1 and t, and a dummy variable indicating whether a plant is a young plant (less than five years old) in period t – 1. The relative size of a plant is defined as the log diﬀerence between the plant and the mean plant in the SIC four-digit industry to which the plant belongs. Industry fixed-eﬀects i are included to control for diﬀerences in changes in product ranges across industries. Time fixed-eﬀects t control for diﬀerences over time, which arise from changes in production technologies, organizational structures, or business conditions. Our choice of sample for estimating (38) is driven by the availability of data on plant export status and industry tariﬀ rates. The longitudinal ASM plant sample provides data on exports for the plants given long forms for the following years, 1979, 1984, 1990, 1993, and 1996 and 1997. Tariﬀs are available for the period 1980 to 1996. As such, we use two panels of continuing long form plants, one over the period 1984 to 1990 and the other over the period 1990 to 1996. We further restrict the sample to those plants that produce more than one product at the start of each period. We have a 9. We are grateful to Professor Dan Trefler for providing us with Canadian and U.S. tariﬀ rates (for details on the sources and construction of the tariﬀ data, see the appendix in Trefler 2004).

Trade and Plant Scale, Production-Run Length, and Diversification

575

total of 7,074 plants for the period 1984 to 1990 and 5,966 plants for the period 1990 to 1996.10 We ask whether plants in industries with larger tariﬀ changes had larger changes in product diversification, production-run length, and plant size. A positive coeﬃcient on the tariﬀ change variable indicates that the plants in the industries with large tariﬀ cuts have a bigger decline in plant performance variable Y. The plant characteristics are included to provide us with evidence on the changes that were taking place within industries in terms of product ranges. They allow us to determine whether changes in plant size, production-run length, and product diversification took place in subsets of plants and thereby to infer what the basic underlying forces behind changes might have been. The initial value of plant size, production-run length, and product diversification is included to control for the natural process of the regression to mean. There are two empirical issues in estimating equation (39). First, the estimated equation includes a lagged dependent variable to control for the regression to the mean. This may introduce a bias in the estimates. Second, the sample for estimation consists of all plants that produce more than one product in the initial period. This may introduce a sample selection bias due to the exclusion of single-product plants. We will address those issues in our estimation. We begin with summary statistics on the extent and trend of product diversification for Canadian manufacturing plants. In figure 15.1, we plot the average number of products per plant both for multiproduct plants and then for all plants, including those producing just a single product. The two curves exhibit the same pattern. Plant-level diversification is relatively constant from the early 1970s to 1987, but then begins to decline.11 Over the period 1987 to 1997, the number of products per plant at multi-product plants falls by 16 percent. The number of products per plant among all plants falls by about 28 percent over the same period. The decline in plant diversification among all plants is a result both of a decline in the share of plants that produce more than one product and a decline in the diversification of the multi-product plants.12 In figures 15.2 and 15.3, we plot the average number of products at exporters and nonexporters.13 Figure 15.2 includes all plants, and figure 15.3 10. The exact number of observations for estimation may diﬀer slightly across specifications as a result of missing values on some variables. 11. As with the number of plants per firm, the number of products per plant starts to decline two years before the FTA with the United States. 12. For more detail, see Baldwin, Beckstead, and Caves (2001). 13. As data on exports are only available for the following years, 1974, 1979, 1984, 1990, 1993, 1996, and 1997, we compare exporters and nonexporters in those years in figures 15.2 and 15.3.

576

John Baldwin and Wulong Gu

Fig. 15.1

Product diversification of manufacturing plants

Fig. 15.2

Product diversification of all exporters and nonexporters

includes only multiproduct plants. The number of products declined in both exporters and nonexporters. But the decline was faster at exporters. In 1973, exporters tended to have a higher level of product diversification than nonexporters. In 1997, there was little diﬀerence between exporters and nonexporters. Figure 15.4 shows the average size (real gross output) of Canadian manufacturing plants, normalized to 100 for multi-product plants in 1973. The

Trade and Plant Scale, Production-Run Length, and Diversification

Fig. 15.3

Product diversification of multiproduct exporters and nonexporters

Fig. 15.4

Average size of manufacturing plants

577

average plant size increased over time and showed large fluctuations over business cycles. It declined during the recessions of the early 1980s and early 1990s. In figures 15.5 and 15.6, we plot the average size of exporters and nonexporters. Figure 15.5 includes all plants and figure 15.6 includes only multi-product plants. The average size tended to be larger for exporters than for nonexporters. During the 1990s, average plant size increased for

578

John Baldwin and Wulong Gu

Fig. 15.5

Average size of all exporters and nonexporters

Fig. 15.6

Average size of multiproduct exporters and nonexporters

both exporters and nonexporters. In addition, the growth in the size of exporters increased in the 1990s compared with that of nonexporters. Figure 15.7 shows the average production-run length of Canadian manufacturing plants, normalized to 100 for multiproduct plants in 1973. The average production-run length increased over time. The average production-run length of manufacturing plants showed large fluctuations over business cycles. It declined during the recessions in the early 1980s and early 1990s. This is in contrast to the pattern of change for product diversification, which shows little cyclical change. In figures 15.8 and 15.9, we plot the average production-run length of ex-

Trade and Plant Scale, Production-Run Length, and Diversification

579

porters and nonexporters. Figure 15.8 includes all plants and figure 15.9 includes only multiproduct plants. The average production-run length tended to be longer for exporters than for nonexporters. The length of production run increased over time, but the increase was much faster in the 1990s following the Canada–U.S. FTA. The increase in production-run length was faster at exporters than at nonexporters. Table 15.1 presents the mean changes in tariﬀ rates, product diversification, and plant size from our sample of plants. Tariﬀ rates and product diversification declined in both periods 1984 to 1990 and 1990 to 1996. Product diversification showed a much larger decline in the 1990 to 1996 period

Fig. 15.7

Production-run length of manufacturing plants

Fig. 15.8

Production-run length of all exporters and nonexporters

580

John Baldwin and Wulong Gu

Fig. 15.9 Table 15.1

Production-run length of multiproduct exporters and nonexporters Annual average changes in tariﬀs, product diversification, and plant size 1984–1990

1990–1996

–0.0036 –0.0020 –0.0346 –0.0083 0.0157 0.0504

–0.0076 –0.0034 –0.0420 –0.0130 0.0195 0.0615

Log changes in the number of products Changes in product diversification index Changes in real output Changes in production-run length

–0.0422 –0.0105 0.0139 0.0561

–0.0403 –0.0123 0.0264 0.0667

Nonexporters Log changes in the number of products Changes in product diversification index Changes in real output Changes in production-run length

–0.0298 –0.0070 0.0168 0.0467

–0.0441 –0.0140 0.0110 0.0551

Canadian tariﬀ changes U. S. tariﬀ changes Log changes in the number of products Changes in product diversification index Changes in real output Changes in production-run length Exporters

Note: The length of production runs in a plant is defined as plant output divided by the number of products.

as tariﬀ reductions became larger. The rate of decline in the number of products rose from 3.4 to 4.2 percent per year from the 1984 to 1990 to 1990 to 1996 period. The rate of decline in product diversification index increased from 0.8 to 1.3 percent per year. Average plant size and average production-run length increased in both the 1980s and 1990s. The rate of growth was faster during the 1990s as tariﬀ

Trade and Plant Scale, Production-Run Length, and Diversification

581

cuts deepened. These results are encouraging and consistent with the model’s predictions about plant size and product diversification. Table 15.1 also shows that product diversification (product counts and product diversification) declined at both exporters and nonexporters during the 1980s and 1990s. The rate of decline became much larger at nonexporters in the 1990s as tariﬀ cuts deepened. There were increases in production-run length and plant size among both exporters and nonexporters, and the rate of growth showed a somewhat larger acceleration in the 1990s among exporters. The evidence is consistent with the model’s prediction about the diﬀerence in the impact of tariﬀ changes between exporters and nonexporters. 15.5.1 Number of Products Our model has a specific implication for the relationship between tariﬀ barriers and the product range of plants. The number of products will decline as tariﬀ rates fall. The rate of decline in the number of products should be smaller for larger and exporting plants. The evidence in table 15.2 shows that the eﬀect of tariﬀ cuts on the number of products is diﬀerent between exporters and nonexporters and be-

Table 15.2

Changes in the number of products

Tariﬀ changes # of products in log Exporter tariﬀ changes Relative plant size tariﬀ changes New exporter Young plants Dummy for period 1990–1996 Observations R2

(1)

(2)

(3)

0.5737∗∗∗ (2.95) –0.0674∗∗∗ (–33.83) –0.0108∗∗∗ (–3.37) –0.7451∗∗∗ (–3.39) 0.0112∗∗∗ (12.59) — — 0.0015 (0.53) –0.0032 (–1.11) –0.0085∗∗∗ (–4.10)

0.2650 (1.52) –0.0676∗∗∗ (–33.92) –0.0044 (–1.66) — — 0.0098∗∗∗ (8.88) –0.1721 (–1.79) 0.0013 (0.46) –0.0030 (–1.03) –0.0089∗∗∗ (–4.27)

0.5611∗∗∗ (2.90) –0.0675∗∗∗ (–33.86) –0.0103∗∗∗ (–3.22) –0.6889∗∗∗ (–3.07) 0.0103∗∗∗ (9.24) –0.1130 (–1.16) 0.0014 (0.51) –0.0030 (–1.03) –0.0086∗∗∗ (–4.12)

12,034 0.16

12,034 0.16

12,034 0.16

Notes: Numbers in parentheses are robust t-statistics. Regressions cover two panels: 1984–1990 and 1990–1996. All specifications include fixed eﬀects for 4-digit industries. ∗∗∗Significant at the 1 percent level.

582

John Baldwin and Wulong Gu

tween large and smaller plants. The results in column (1) suggest that lower tariﬀs reduce the number of products produced by nonexporters. A onepercentage-point decline in tariﬀs is associated with a 0.6 percent decline in the number of products at nonexporters. But tariﬀ cuts have little eﬀect on the number of products at exporters, as the sum of the coeﬃcient on tariﬀ changes and its interaction with exporter is not significant at the 5 percent level. These results are consistent with those reported in Baldwin, Caves, and Gu (2004). In column (2), we examine the diﬀerence in the eﬀect of tariﬀ cuts on the number of products produced between large and small plants. We find that tariﬀ cuts reduce the number of products that a large plant produces. Our evidence suggests a 1 percentage point tariﬀ cut is associated with a 5 percent decline in the number of products at the plants that are 1 standard deviation smaller than an average plant. But it does not have statistically significant eﬀect on the number of products at the plants that are that are 1 standard deviation larger than an average plant. The results in column (3) show that tariﬀ cuts are associated with a larger rate of decline in the number of products at smaller nonexporters than at larger nonexporters. Overall, the evidence from nonexporting plants in table 15.2 is consistent with the prediction of our model. But, the evidence from exporting plants appears to be at odds with our model. The evidence in table 15.2 shows that while exporters reduce product ranges relative to nonexporters, the decline in the number of products is not related to tariﬀ cuts. For exporters, the eﬀect of tariﬀ cuts on the number of products is not significant at the 5 percent level. This suggests that once in the export markets, plants respond to forces other than tariﬀ cuts, such as learning-by-exporting, the competitive force in the export market, and opportunities aﬀorded with an access to larger markets (Baldwin and Gu 2004). For those exporting plants, additional tariﬀ cuts may not be an important factor in the choice of product ranges. Baldwin and Gu (2004) also find that exporters increase product specialization relative to nonexporters and interpret this as evidence that exporting raises productivity growth through increased product specification. Nevertheless, it should be noted that the sign on plant size is opposite to that on exporters and about the same magnitude, which implies that the eﬀect of being an exporter exists for smaller plants but is unimportant for large plants. The results in table 15.2 also show that larger plants also add new products in order to expand their market for their products. 15.5.2 Product Diversification Our model predicts that lower tariﬀ rates reduce the product diversification index of nonexporters. It has an ambiguous eﬀect on the diversification index of existing and new exporters. For the exporters, lower tariﬀ

Trade and Plant Scale, Production-Run Length, and Diversification

583

rates leads to decline in the number of products and an increase in the portion of its product line shipped abroad. These two eﬀects are oﬀsetting and generate an ambiguous eﬀect of tariﬀ cuts on the product diversification index of exporters. Table 15.3 presents empirical evidence on the eﬀects of tariﬀ cuts on the product diversification index of a plant. The results in column (1) suggest that the reduction in tariﬀ rates is associated with a decline in the product diversification index of nonexporting plants. The eﬀect of lower tariﬀ rates on the product diversification index of exporting plants, which is the sum of the coeﬃcients on tariﬀ changes and its interaction with plant export status, is not significant at the 10 percent level. This implies that tariﬀ reductions have little eﬀect on the product diversification of exporters. In column (2), we examine the diﬀerence in the eﬀect of lower tariﬀs on product diversification across plant sizes. The results show that tariﬀ reductions have less of an impact on the diversification of larger plants than on that of smaller plants. A 1 percentage point decline in tariﬀ rates is associated with 0.2 percent decline in the plant diversification index for plants that are 1 standard deviation smaller than an average plant. The Table 15.3

Changes in product diversification index

Tariﬀ changes Product diversification index Exporter tariﬀ changes Relative plant size tariﬀ changes New exporter Young plants Dummy for period 1990–1996 Observations R2

(1)

(2)

(3)

0.1281 (1.88) –0.0725∗∗∗ (–38.72) –0.0029∗∗∗ (–2.67) –0.2120∗∗∗ (–2.80) 0.0034∗∗∗ (11.60) — — 0.0005 (0.53) –0.0007 (–0.71) –0.0049∗∗∗ (–6.98)

0.0457 (0.76) –0.0726∗∗∗ (–38.79) –0.0011 (–1.24) — — 0.0027∗∗∗ (7.25) –0.0984∗∗∗ (–2.98) 0.0004 (0.45) –0.0005 (–0.54) –0.0050∗∗∗ (–7.13)

0.1189 (1.75) –0.0726∗∗∗ (–38.77) –0.0026∗∗ (–2.36) –0.1704∗∗ (–2.22) 0.0028∗∗∗ (7.51) –0.0838∗∗ (–2.49) 0.0005 (0.48) –0.0006 (–0.55) –0.0049∗∗∗ (–7.02)

12,037 0.20

12,037 0.20

12,037 0.20

Notes: Numbers in parentheses are robust t-statistics. Regressions cover two panels: 1984–1990 and 1990–1996. All specifications include fixed eﬀects for 4-digit industries. ∗∗∗Significant at the 1 percent level. ∗∗Significant at the 5 percent level.

584

John Baldwin and Wulong Gu

eﬀect is significant at the 5 percent level. In contrast, the eﬀect of the tariﬀ cuts on the product diversification of plants that are 1 standard deviation larger is not statistically significant at the 5 percent level. This is consistent with the finding on the number of products in the previous section, where we find that lower tariﬀs reduce the number of products of larger plants less than that of smaller plants. The results in column (3) show that tariﬀ cuts are associated with a larger rate of decline in the product diversification index at small nonexporters than at larger nonexporters. Overall, the results in table 15.3 are consistent with the prediction of our model regarding the eﬀect of tariﬀ cuts on product diversification. The coeﬃcient estimates on the export status variable suggest that exporters reduce product diversification relative to nonexporters, a finding that is consistent with the one in Baldwin and Gu (2004). Once more, this impact exists primarily for small exporters. To examine the eﬀect of tariﬀ cuts on the product diversification of new exporters, we have introduced an interaction term of the variables for new exporters and tariﬀ changes. The evidence suggests that the eﬀect of tariﬀ cuts on the product diversification of entrants to the export market is not significant at the 5 percent level. This is consistent with the model’s prediction that tariﬀ cuts have an ambiguous eﬀect on the product diversification of the entrants to the export market relative to nonexporters. 15.5.3 Plant Size Our model has implications for plant size. The decline in tariﬀ barriers will reduce the size of nonexporting plants as these plants reduce the range of their product lines. But it has an ambiguous eﬀect on the size of existing and new exporters. For those plants, the tariﬀ reduction leads to an increase in export sales and an oﬀsetting decline in domestic sales. The results in table 15.4 provide empirical evidence that is consistent with our model’s prediction about plant size. The coeﬃcient on tariﬀ changes in column (1) is positive and significant at the 1 percent level. Lower tariﬀs lead to a decline in the size of nonexporters. The eﬀect of tariﬀ changes on the plant size of exporters, which is the sum of the coeﬃcients on tariﬀ changes and its interaction with plant export status, is not significant. This suggests that the tariﬀ reduction does not have a significant eﬀect on the size of exporters.14 To examine the eﬀect of tariﬀ cuts on the size of new exporters, we have introduced an interaction term of the variables for tariﬀ cuts and new exporters. We find that the tariﬀ reduction does not have an eﬀect on the size of new exporters. 14. When we introduce the interaction of tariﬀ changes with the dummies for current and new exporters separately, we find that the diﬀerence in the coeﬃcients on the two interactions terms is not significant.

Trade and Plant Scale, Production-Run Length, and Diversification Table 15.4

Changes in plant size

Tariﬀ changes Exporter tariﬀ changes Relative plant size tariﬀ changes New exporter Young plants Dummy for period 1990–1996 Observations R2

585

(1)

(2)

(3)

0.4688∗∗ (2.29) 0.0185∗∗∗ (5.58) –0.1637 (–0.70) –0.0171∗∗∗ (–17.74) — — 0.0195∗∗∗ (7.32) 0.0175∗∗∗ (5.75) 0.0033 (1.53)

0.3706∗∗ (1.97) 0.0199∗∗∗ (7.49) — — –0.0153∗∗∗ (–12.15) 0.2445∗∗ (2.33) 0.0196∗∗∗ (7.36) 0.0170∗∗∗ (5.58) 0.0032 (1.51)

0.4984∗∗ (2.44) 0.0174∗∗∗ (5.24) –0.2975 (–1.26) –0.0150∗∗∗ (–11.87) 0.2699∗∗ (2.54) 0.0196∗∗∗ (7.37) 0.0170∗∗∗ (5.58) 0.0034 (1.57)

12,034 0.09

12,034 0.09

12,034 0.09

Notes: Numbers in parentheses are robust t-statistics. Regressions cover two panels 1984–1990 and 1990–1996. All specifications include fixed eﬀects for 4-digit industries. ∗∗∗Significant at the 1 percent level. ∗∗Significant at the 5 percent level.

The evidence in column (2) suggests that tariﬀ cuts have more of a negative eﬀect on the size of larger plants than on that of smaller plants. A 1 percentage point decline in tariﬀ rates is associated with a 0.6 percent decline in the size of plants that are one standard deviation larger than an average plant. But the eﬀect of tariﬀ cuts on plant size is not significant at the 5 percent level for plants that are 1 standard deviation smaller than an average plant. The evidence in column (3) suggests that the negative eﬀect of tariﬀ cuts on the size of nonexporters increase with plant size. The rate of decline in plant size as a result of tariﬀ cuts is larger for larger nonexporters that for smaller nonexporters. While the tariﬀ cut does not have a significant eﬀect on the size of average exporters, the evidence in column (3) shows that it reduces the size of larger exporters. The coeﬃcients on the exporters and new exporter variables are positive and significant at the 5 percent level. The exporting plants increase their size relative to nonexporters. Baldwin and Gu (2004) finds a similar result. One of the predictions of policy advocates for free trade was that plant size would increase as a result of free trade. A number of previous studies have examined the relationship between tariﬀ barriers and plant size and found little evidence that tariﬀ cuts increased plant size (Head and Ries

586

John Baldwin and Wulong Gu

1999). The firm-based approach to models of trade used in this chapter and other papers (Melitz 2003) highlights the diﬀerences in the responses to tariﬀ reductions that should be expected across plants. Our model and that of Melitz (2003) show that tariﬀ reductions have a diﬀerent eﬀect on the size of exporters and nonexporters. 15.5.4 Production-Run Length Our model has implications for the length of production runs within individual producers. As tariﬀ rates fall, the length of production runs will increase for existing and new exporters as a result of declines in product ranges and increases in the foreign sales of their products. For nonexporters, the length of production run will remain the same. We define the length of production run of individual products for a plant as the ratio of the real output of the plant to the number of products of the plant. The estimated length of production runs represents an average across products, as output distribution is not uniform across individual products. Consistent with the model, the evidence in column (1) of table 15.5 suggests that tariﬀ cuts do not have statistically significant eﬀects on the production-run length of nonexporters. However, the evidence on the eﬀect of tariﬀ cuts on the production-run length of exporters is at odds with the model’s prediction. The eﬀect of tariﬀ cuts on the production-run length of exporters, as calculated as the sum of the coeﬃcients on the tariﬀ change and exporter variables, is not significant at the 5 percent level. In addition, the eﬀect of tariﬀ cuts on the production-run length of new exporters is not found to be statistically significant. This suggests that the tariﬀ cuts do not increase the production-run length of exporters as the model predicts. While tariﬀ reductions do not increase the production-run length of exporters and entrants to the export market, the evidence shows that those exporting plants increased the production-run length compared with nonexporters. We interpret this evidence as suggesting that plants, once in the export markets, do not consider additional tariﬀ cuts as an important determinant in the choice of production-run length. For exporters and entrants to the export market, learning-by-exporting, competition in the export market and continued access to the export market are much more important factors in their production decision. 15.5.5 Discussions of the Results In this section, we discuss two main empirical issues in our estimation. The first relates to our choice of regression specification and the second relates to possible sample selection bias due to our choice of the sample. To estimate the eﬀects of tariﬀ cuts on product diversification, productionrun length, and plant size, we have used an empirical specification that includes a lagged dependent variable as a control variable. If the lagged de-

Trade and Plant Scale, Production-Run Length, and Diversification Table 15.5

Changes in production-run length

Tariﬀ changes Product run in log Exporter tariﬀ changes Relative plant size tariﬀ changes New exporter Young plants Dummy for period 1990–1996 Observations R2

587

(1)

(2)

(3)

–0.1415 (–0.53) –0.0633∗∗∗ (–26.30) 0.0299∗∗∗ (7.07) 0.5997∗∗ (2.00) 0.0351∗∗∗ (14.61) — — 0.0184∗∗∗ (5.14) 0.0196∗∗∗ (4.94) 0.0179∗∗∗ (6.28)

0.0766 (0.32) –0.0637∗∗∗ (–26.39) 0.0248∗∗∗ (7.14) — — 0.0386∗∗∗ (14.59) 0.4203∗∗∗ (3.10) 0.0187∗∗∗ (5.22) 0.0188∗∗∗ (4.75) 0.0183∗∗∗ (6.40)

–0.0989 (–0.37) –0.0637∗∗∗ (–26.36) 0.0284∗∗∗ (6.69) 0.4085 (1.33) 0.0382∗∗∗ (14.34) 0.3854∗∗∗ (2.77) 0.0186∗∗∗ (5.19) 0.0188∗∗∗ (4.75) 0.0181∗∗∗ (6.34)

12,034 0.15

12,034 0.15

12,034 0.15

Notes: Numbers in parentheses are robust t-statistics. Regressions cover two panels: 1984–1990 and 1990–1996. All specifications include fixed eﬀects for 4-digit industries. ∗∗∗Significant at the 1 percent level. ∗∗Significant at the 5 percent level.

pendent variable is predetermined, the ordinary least squares (OLS) estimators are consistent. However, if the lagged dependent variable is correlated with error terms, the OLS estimation will yield a biased estimate of the coeﬃcient on the lagged dependent variable. But it will yield consistent estimates of the coeﬃcients on the variables of interest, such as tariﬀ changes and plant export status. To examine the robustness of our findings on the eﬀects of tariﬀ cuts, we have also estimated a specification that excludes the lagged dependent variable. The results are presented in table 15.6. Overall, the results are similar to those obtained using specifications that include the lagged dependent variable. The sample for the estimation consists of the plants that produce more than one product in the initial period. This may introduce sample selection bias due to the exclusion of single-product plants. To address the issue of sample selection bias, we have estimated the regression equation using a sample that also includes the single-product plants. As shown in table 15.7, the evidence from the full sample shows that tariﬀ cuts reduce the product diversification and size of nonexporting

588

John Baldwin and Wulong Gu

Table 15.6

Alternative estimates of the eﬀect of tariﬀ changes on product diversification, plant size, and production-run length Dependent variables

Tariﬀ changes Exporter tariﬀ changes Relative plant size tariﬀ changes New exporter Young plants Dummy for period 1990–1996 Observations R2

No. of products

Product div. index

Plant size

PR length

0.6808∗∗∗ (3.28) –0.0093∗∗∗ (–2.74) –0.8322∗∗∗ (–3.42) 0.0054∗∗∗ (4.59) 0.0355 (0.34) 0.0043 (1.48) 0.0020 (0.66) –0.0055∗∗ (–2.50)

0.1306 (1.77) –0.0017 (–1.46) –0.1844∗∗ (–2.21) 0.0016∗∗∗ (3.92) –0.0382 (–1.06) 0.0009 (0.89) 0.0003 (0.31) –0.0048∗∗∗ (–6.36)

0.4984∗∗ (2.44) 0.0174∗∗∗ (5.24) –0.2975 (–1.26) –0.0150∗∗∗ (–11.87) 0.2699∗∗ (2.54) 0.0196∗∗∗ (7.37) 0.0170∗∗∗ (5.58) 0.0034 (1.57)

–0.1820 (–0.66) 0.0266∗∗∗ (6.09) 0.5329 (1.67) –0.0204∗∗∗ (–12.48) 0.2340 (1.63) 0.0152∗∗∗ (4.09) 0.0149∗∗∗ (3.66) 0.0089∗∗∗ (3.01)

12,034 0.05

12,034 0.05

12,034 0.09

12,034 0.09

Notes: Numbers in parentheses are robust t-statistics. Regressions cover two panels: 1984–1990 and 1990–1996. All specifications include fixed eﬀects for 4-digit industries. ∗∗∗Significant at the 1 percent level. ∗∗Significant at the 5 percent level.

plants, and has no eﬀect on the production-run length of those plants. Exporting plants reduce product diversification and increase production-run length and plant size, but those changes do not appear to be related to tariﬀ cuts. Overall, these results are qualitatively similar to those obtained using the multi-product plant sample. But as the changes in product diversification are left-censored for single-product plants, the estimated eﬀect of tariﬀ changes on product diversification is lower than the estimated eﬀect using the multi-product plant sample. 15.6 Conclusions Microdata on business populations provide a rich picture of heterogeneity within firm populations. They provide new information on the variety of change going on within industries. Initially, studies of change focused primarily on describing the nature of diﬀerent groups—those that were gaining and losing market share, those that entered and exited versus incumbents, and those that gained and lost

Trade and Plant Scale, Production-Run Length, and Diversification Table 15.7

589

The eﬀect of tariﬀ changes on product diversification, plant size and production-run length from a sample of all continuing plants Dependent variables

Tariﬀ changes Exporter tariﬀ changes Relative plant size tariﬀ changes New exporter Young plants Dummy for period 1990–1996 Observations R2

No. of products

Product div. index

0.4743∗∗ (2.50) –0.0068∗∗ (–2.30) –0.6165∗∗∗ (–2.73) –0.0032∗∗∗ (–3.30) –0.0007 (–0.01) 0.0047 (1.87) 0.0143∗∗∗ (5.91) –0.0061∗∗∗ (–3.19)

0.0875 (1.35) –0.0015 (–1.47) –0.1221 (–1.63) –0.0010∗∗∗ (–2.97) –0.0246 (–0.80) 0.0011 (1.29) 0.0039∗∗∗ (4.78) –0.0042∗∗∗ (–6.73)

17,211 0.03

17,211 0.04

Plant size

PR length

0.7379∗∗∗ 0.2638 (3.96) (1.05) 0.0206∗∗ 0.0274∗∗∗ (7.16) (7.10) –0.4186 0.1970 (–1.95) (0.67) –0.0187∗∗∗ –0.0155∗∗∗ (–14.58) (–10.38) 0.1402 0.1412 (1.43) (1.12) 0.0238∗∗∗ 0.0190∗∗∗ (10.11) (5.84) 0.0206∗∗∗ 0.0063∗∗ (8.52) (1.96) 0.0027 0.0088∗∗∗ (1.44) (3.45) 17,205 0.11

17,205 0.06

Notes: Numbers in parentheses are robust t-statistics. Regressions cover two panels: 1984–1990 and 1990–1996. All specifications include fixed eﬀects for 4-digit industries. ∗∗∗Significant at the 1 percent level. ∗∗Significant at the 5 percent level.

relative productivity. The picture that these studies provided is one of heterogeneous populations, with diﬀerent types of producers existing side by side. Studies using business microdata have begun to outline the ramifications of heterogeneity in producer characteristics. For example, some members contribute more to productivity growth than others. Equally important, heterogeneous producers might be expected to respond diﬀerently to exogenous shocks. This chapter has focused on one such response to outside shocks—the response of diﬀerent manufacturers to trade liberalization. Others have focused on the reaction of industries as a whole to trade liberalization, treating industries as a homogeneous set of producers. In contrast, the approach adopted here has focused on developing a model of heterogeneous producers that diﬀer in terms of costs and asking whether the reaction of producers to trade liberalization might be expected to diﬀer in a systematic way. To do so, we present a model that suggests that two groups of firms, distinguished here as nonexporters and exporters, would be expected to diﬀer

590

John Baldwin and Wulong Gu

substantially in terms of their reaction to trade liberalization with respect to the number of products produced, product specialization, plant size, and, finally, the length of production-run. The stylized model predicts that tariﬀ reductions should increase product specialization and decrease plant size in nonexporters. Its eﬀect on specialization of existing exporters is ambiguous—though it is expected to have a positive eﬀect on the length of production runs in exporters. The empirical evidence on nonexporting plants provides broad support for the model. The evidence on exporting plants shows that exporters reduce product diversification, and increase production-run length and plant size, but those changes do not appear to be related to tariﬀ cuts. Once in the export markets, plants respond to forces other than tariﬀ cuts. Baldwin and Gu (2004) identified learning by exporting, competition in the export market, and access to the larger market as important factors in the production decision of exporters. These findings support the need to think of producer populations as heterogeneous units whose reactions are likely to be diverse. They also stress the need to be cautious about generalizations based on representative plants or firms. While the chapter helps to shed light on the reaction to tariﬀ changes, it also suggests that other changes were taking place within the population of manufacturers. Testing stylized models is diﬃcult when those models have diﬃculty in taking into account changing circumstances. While our findings on the eﬀects of tariﬀ changes accord broadly with expectations, other results suggest the need to expand our research. In particular, the reaction of exporters relative to nonexporters suggests that the underlying technology was not staying constant. Small exporters were more likely to specialize or reduce diversity than large exporters. Similarly, small exporters were more likely to increase their plant size. This suggests that the technology conditions of smaller plants that resulted in increased diversification— possibly to take advantage of scale economies—changed over the time period studied. One explanation for this is that the attraction of scale changed across plant size classes—that is, the advantages of incremental improvements in size increased for larger plants relative to smaller plants. This suggests a shift in the nature of technologies or capital intensity between small and large plants in favor of large plants that led to increased opportunities to exploit scale economies via diversification in the 1990s. In related work, we have found evidence of this occurring. Baldwin, Rama, and Sabourin (1999) report that the gap in advanced technology use between small and large plants increased in the 1990s. Baldwin and Dhaliwahl (2001) report that output per worker in larger plants has increased relative to smaller plants throughout the period. Baldwin, Jarmin, and Tang (2002) report the same phenomenon can be found in both Canada and the

Trade and Plant Scale, Production-Run Length, and Diversification

591

United States. These studies suggest that the degree of scope economies that provide the incentive to increase diversification probably increased in large plants at the same time as trade liberalization was occurring. Our study has also shown that there is a dynamic aspect to the growth of producers that our analytical models have not fully captured. In our models, producers diﬀer at a point in time by their level of unit costs. But this distribution is subject to change. Just as producers grow by increasing their capital intensity, they also do so by learning how to combine more than one product within an establishment to take advantage of scale and scope economies. Both transitions require a learning process that ultimately needs to be incorporated into a more dynamic framework.

References Baldwin, J. R. 1995. The dynamics of industrial competition. Cambridge: Cambridge University Press. Baldwin, J. R., D. Beckstead, and R. E. Caves. 2001. Changes in diversification of Canadian manufacturing firms and plants (1973–1997): A move to specialization. Analytical Studies Branch Research Paper Series no. 179. Ottawa: Statistics Canada. Baldwin, J. R., R. E. Caves, and W. Gu. 2004. Responses to trade liberalization: Changes in product diversification in foreign and domestic controlled plants. In Governance, multinationals and growth, ed. L. Eden and W. Dobson, 206–46. Northampton, MA: Edward Elgar Publishing. Baldwin, J. R., and N. Dhaliwal. 2001. Heterogeneity in labour productivity growth in manufacturing: Diﬀerences between domestic and foreign-controlled establishments. In Productivity Growth in Canada. Ottawa: Statistics Canada. Baldwin, J. R., and P. K. Gorecki. 1983. Trade, tariﬀs, relative plant scale in Canadian manufacturing industries, 1976–79. Discussion Paper no. 232. Ottawa: Economic Council of Canada. ———. 1986. The role of scale in Canada-U.S. productivity diﬀerences in the manufacturing sector: 1970–79. Toronto: University of Toronto Press. Baldwin, J. R., and W. Gu. 2003. Participation in export markets and productivity performance in Canadian manufacturing. Canadian Journal of Economics 36 (3): 634–57. ———. 2004. Trade liberalization: Export-market participation, productivity growth, and innovation. Oxford Review of Economic Policy 20 (3): 372–92. Baldwin, J. R., R. S. Jarmin, and J. Tang. 2002. Small North American producers give ground in the 1990s. Journal of Small Business Economics 23 (4): 349–61. Baldwin, J. R., E. Rama, and D. Sabourin. 1999. Growth of advanced technology use in Canadian manufacturing during the 1990s. Research Paper no. 105, Analytical Studies Branch. Ottawa: Statistics Canada. Bernard, A. B., J. Eaton, B. J. Jensen, and S. Kortum. 2003. Plants and productivity in international trade. American Economic Review 93 (4): 1268–90. Bernard, A. B., J. B. Jensen, and P. K. Schott. 2003. Falling trade costs, heterogeneous firms, and industry dynamics. NBER Working Paper no. 9639. Cambridge, MA: National Bureau of Economic Research, March.

592

John Baldwin and Wulong Gu

Campbell, J. R., and H. A. Hopenhayn. 2002. Market size matters. NBER Working Paper no. 9113. Cambridge, MA: National Bureau of Economic Research, July. Caves, R. E. 1975. Diversification, foreign investment, and scale in North American manufacturing industries. Ottawa: Economic Council of Canada. Daly, D. J., B. A. Keys, and E. J. Spence. 1968. Scale and specialization in Canadian manufacturing. Economic Council Staﬀ Study no. 21. Ottawa: Queen’s Printer. Economic Council of Canada. 1967. Fourth annual review: The Canadian economy from the 1960s to the 1970s. Ottawa: Queen’s Printer. ———. 1975. Looking outward. Ottawa: Information Canada. Greenhut, M. L., G. Norman, and C.-S. Hung. 1987. The economics of imperfect competition: A spatial approach. Cambridge: Cambridge University Press. Harris, R. 1984. Applied general equilibrium analysis of small open economies with scale economies and imperfect competition. American Economic Review 74 (5): 1016–32. Head, K., and J. Ries. 1999. Rationalization eﬀects of tariﬀ reductions. Journal of International Economics 47:295–320. Helpman, E., and P. Krugman. 1989. Trade policy and market structure. Cambridge, MA: MIT Press. Helpman, E., M. J. Melitz, and S. R. Yeaple. 2004. Export versus FDI with heterogeneous firms. American Economic Review 94 (10): 300–16. Jacquemin, A. P., and C. H. Berry. 1979. Entropy measures of corporate growth. The Journal of Industrial Economics 27:359–69. Melitz, M. J. 2003. The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica 71 (6): 1695–1725. Melitz, M. J., and G. I. P. Ottaviano. 2005. Market size, trade, and productivity. NBER Working Paper no. 11393. Cambridge, MA: National Bureau of Economic Research, May. Ottaviano, G. I. P., T. Tabuchi, and J.-F. Thisse. 2002. Agglomeration and trade revisited. International Economic Review 43 (2): 409–35. Ottaviano, G. I. P., and J.-F. Thisse. 1999. Monopolistic competition, multiproduct firms, and optimum product diversity. CORE Discussion Paper no. 9919. Roberts, M., and J. Tybout, eds. 1996. Industrial evolution in developing countries. New York: Oxford University Press. Royal Commission on Corporate Concentration. 1978. Report. Ottawa: Minister of Supply and Services Canada. Safarian, E. 1966. Foreign ownership in Canadian industry. Toronto: McGraw-Hill of Canada. Scherer, F. M., A. Beckenstein, E. Kaufer, R. D. Murphy, and F. BougeonMaassen. 1975. The economics of multi-plant operation. Cambridge, MA: Harvard University Press. Syverson, C. 2003. Product substitutability and productivity dispersion. NBER Working Paper no. 10049. Cambridge, MA: National Bureau of Economic Research, September. Trefler, D. 2004. The long and short of the Canada-U.S. Free Trade Agreement. The American Economic Review 94 (4): 870–95. Tybout, J. 2003. Plant and firm-level evidence on new trade theories. In Handbook of International Trade ed. J. Harrigan and K. E. Choi, 388–415. Oxford: BasilBlackwell. Yeaple, S. R. 2005. A simple model of firm heterogeneity, international trade, and wages. Journal of International Economics 65 (1): 1–20.

Trade and Plant Scale, Production-Run Length, and Diversification

Comment

593

James Tybout

Almost two decades ago, trade economists began to study the eﬀects of openness when product markets are imperfectly competitive. The resulting new trade literature emphasized that openness increases welfare by increasing the menu of products available in each country. It also emphasized that openness can make product markets more competitive, and can thus induce firms to exploit scale economies more fully. The former message has remained central to modern trade theory, but the view that openness leads to significant gains in firm-level scale eﬃciency is no longer widely held. There are several reasons that the link between trade and scale eﬃciency has been deemphasized. One is that trade economists have found new ways to link openness with welfare.1 But equally important, the early empirical evidence for trade-induced gains in scale eﬃciency was less than compelling. Enthusiasm for large scale eﬃciency eﬀects was initially stoked by simulation studies that found commercial policy reforms might generate eﬃciency gains on the order of 5 percent or larger (Tybout 1993). However, these simulations ignored intraindustry heterogeneity. Thus, all firms within an industry were treated as being of average size for that industry, and since most industries are populated by many small firms and a few large ones, this average size was typically ineﬃciently small. This meant that modest increases in firm size could generate substantial eﬃciency gains, even though most production really came from plants well above minimum eﬃcient scale. Another problem with many simulation models was that they were hardwired to ensure that increases in the scale of production took place at all firms with trade liberalization. But as econometric studies linking import competition and firm size emerged, it became clear that plants in import-competing industries tend to contract when exposed to heightened competition from abroad.2 The chapter by Baldwin and Gu (hereafter BG) is interesting because it goes some way toward resuscitating the notion that openness might generate significant welfare gains through simple scale eﬀects. Because it focuses on the length of production runs rather than firm size, scale eﬃciency gains are possible at both small and large firms. Thus, BG’s model emphasizes a kind of scale eﬀect that can be reconciled with the fact that large firms account for most output. Similarly, it reconciles reductions in the aggregate scale of output with increasing production-run-level scale eﬃciency. Perhaps most importantly, BG show that changes in trade policy are sigJames Tybout is a professor of economics at the Pennsylvania State University and a research associate of the National Bureau of Economic Research. 1. The new linkages included induced innovation, agglomeration economy eﬀects, and induced market share reallocations across firms with diﬀering marginal production costs. 2. For Canada, a well-known example is Head and Reis (1999).

594

John Baldwin and Wulong Gu

nificantly correlated with product diversity, with production run length, and with firms’ relative positions in the size distribution. Thus, they convincingly demonstrate that trade-induced production run eﬀects are worth worrying about. Their findings complement earlier studies that established a correlation between openness and firm-level eﬃciency but were unable to control for production run length. They also complement recent work by Bernard, Redding, and Schott (2006b), who show that two-thirds of manufacturing plants in the United States alter their mix of 5-digit products over a five-year period, and this reallocation process appears to be eﬃciencyenhancing. The theory developed by BG also constitutes a useful contribution, and nicely complements some recent works on trade, heterogeneous firms, and endogenous scope. Perhaps the one closest to BG is Bernard, Redding, and Schott’s (2006a, hereafter BRS), which characterizes each firm’s productivity in each product as dependent on both firm-level ability and firmproduct-level expertise. Higher firm-level ability raises a firm’s productivity across all products, which induces a positive correlation between a firm’s intensive (output per product) and extensive (number of products) margins. The BRS model diﬀers from BG’s because (a) it assumes DixitStiglitz preferences and thus fixes firms’ mark-ups parametrically, and (b) it adds an additional dimension of firm heterogeneity. Nonetheless, its predictions are quite similar: “Trade liberalization fosters productivity growth within and across firms and in aggregate by inducing firms to shed marginally productive products and forcing the lowest-productivity firms to exit. Though exporters produce a smaller range of products after liberalization, they increase the share of products sold abroad as well as exports per product” (BRS, 2006a, abstract). Also relevant is Nocke and Yeaple’s (2006) model. It too allows for firm heterogeneity and endogenous scope. However, unlike BRS and BG, Nocke and Yeaple assume that span of control problems cause marginal production costs to rise as additional product varieties are added at a given firm. Nonetheless, as in BRS and BG, reductions in trade costs cause the most eﬃcient firms—that is, those with greatest managerial ability—to shed product lines. The evidence reported in BG and BRS goes some way toward establishing that firms’ scope and production run lengths aﬀect eﬃciency and are related to trade. But there is much more to explore, and these papers will hopefully inspire further research. One unresolved issue is the role of global fragmentation of production in driving scope. When firms can have a stage of their production done abroad, does this mean they produce less product varieties at home, and to what extent does this phenomenon explain the reduction in firms’ scope observed in the Canadian data? A second issue is how unilateral changes in trade policies aﬀect patterns of scale, scope, and eﬃciency. As Melitz and Ottaviano (2008), BRS, and Nocke

Trade and Plant Scale, Production-Run Length, and Diversification

595

and Yeaple (2006) have shown, model predictions can depend critically upon the balance between domestic firms’ access to foreign markets, and foreign firms’ access to domestic markets. Third, it would be worthwhile to explore the sensitivity of firm-level responses to trade policy to both the time horizons involved and the entry costs that new firms face. In the short term, or when entry/exit barriers are prohibitive, all of the response to a trade policy change must be accomplished through adjustment by incumbent firms. But over longer time horizons, entry and exit can dampen pressure to adjust on remaining firms. Finally, the sheer magnitude of the tariﬀ eﬀects documented in BG is puzzling, and merits further investigation. Although tariﬀs only changed a few percentage points on average, they are quite significant in many of the regressions BG report.3 References Bernard, A., S. Redding, and P. Schott. 2006a. Multiproduct firms and product switching. NBER Working Paper no. 12293. Cambridge, MA: National Bureau of Economic Research, May. ———. 2006b. Multiproduct firms and trade liberalization. NBER Working Paper no. 12782. Cambridge, MA: National Bureau of Economic Research, November. Head, K. and J. Reis. 1999. Rationalization eﬀects of tariﬀ reductions. Journal of International Economics 47: 295–320. Melitz, M., and G. Ottaviano. 2008. Market size, trade, and productivity. Review of Economic Studies 75 (1): 295–316. Nocke, V., and S. Yeaple. 2006. Globalization and endogenous Firm scope. NBER Working Paper no. 12322. Cambridge, MA: National Bureau of Economic Research, May. Trefler, D. 2004. The long and short of the Canada-U.S. Free Trade Agreement. American Economic Review 94 (4): 870–95. Tybout, J. 1993. Returns to scale as a source of comparative advantage. American Economic Review: Papers and Proceedings 83 (2): 440–44.

3. Similar large eﬀects appear in other studies of the Canada–U.S. FTA (e.g., Head and Reis 1999 and Trefler 2005).

Contributors

John M. Abowd School of Industrial and Labor Relations 358 East Ives Hall Cornell University Ithaca, NY 14850

Eric Bartelsman Department of Economics and Business Administration Free University Amsterdam De Boelelaan 1105 1081 HV Amsterdam, The Netherlands

Katharine G. Abraham Joint Program in Survey Methodology 1218 LeFrak Hall University of Maryland College Park, MD 20742

Andrew B. Bernard Tuck School of Business at Dartmouth 100 Tuck Hall Hanover, NH 03755

Mary Clare Ahearn Economic Research Service U. S. Department of Agriculture 1800 M Street, NW, rm. 4158 Washington, D. C. 20036 Fredrik Andersson Longitudinal Employer-Household Dynamics (LEHD) U.S. Census Bureau 4600 Silver Hill Road Washington, D. C. 20233 John Baldwin Micro-Economic Analysis Division Statistics Canada R. H. Coats Building Ottawa, Ontario K1A 0T6 Canada

Dan A. Black The Harris School The University of Chicago 1155 East 60th Street Chicago, IL 60637 Jeﬀrey R. Campbell Economic Research Department Federal Reserve Bank of Chicago 230 South LaSalle Street Chicago, IL 60604-1413 Richard L. Clayton U.S. Bureau of Labor Statistics Postal Square Building 2 Massachusetts Avenue, NE Washington, D. C. 20212-0001

597

598

Contributors

Steven J. Davis Graduate School of Business The University of Chicago 5807 South Woodlawn Avenue Chicago, IL 60637 Anja Decressin U.S. Department of Labor Frances Perkins Building 200 Constitution Avenue, NW Washington, D.C. 20210 Timothy Dunne Federal Reserve Bank of Cleveland P.O. Box 6387 Cleveland, OH 44101-1387 R. Jason Faberman Federal Reserve Bank of Philadelphia Ten Independence Mall Philadelphia, PA 19106-1574 Wulong Gu Micro-Economic Analysis Division Statistics Canada R. H. Coats Building Ottawa, Ontario K1A 0T6 Canada John Haltiwanger Department of Economics University of Maryland College Park, MD 20742 James Harrigan Department of Economics University of Virginia Charlottesville, VA 22904 [email protected] Jonathan Haskel Department of Economics Queen Mary, University of London Mile End Road London E1 4NS, England Judith K. Hellerstein Department of Economics Tydings Hall University of Maryland College Park, MD 20742

Tomeka Hill Watson Wyatt Worldwide 901 N. Glebe Road Arlington, VA 22203 Thomas J. Holmes Department of Economics 1035 Heller Hall University of Minnesota 271 19th Avenue South Minneapolis, MN 55455 Ronald S. Jarmin Center for Economic Studies U.S. Census Bureau 4600 Silver Hill Road Washington, D.C. 20233 J. Bradford Jensen McDonough School of Business Georgetown University Washington, D.C. 20057 Shawn D. Klimek Center for Economic Studies U.S. Census Bureau 4600 Silver Hill Road Washington, D.C. 20233 Penni Korb Economic Research Service U.S. Department of Agriculture 1800 M Street, NW Washington, D.C. 20036 C. J. Krizan Center for Economic Studies U.S. Census Bureau 4600 Silver Hill Road Washington, D.C. 20233 Tomas Lindstrom Economic Research Skandinaviska Enskilda Banken SE-196 40 Stockholm, Sweden Kristin McCue Center for Economic Studies U.S. Census Bureau 4600 Silver Hill Road Washington, D.C. 20233

Contributors Kevin L. McKinney California Census Research Data Center University of California, Los Angeles 4250 Public Policy Building Box 951484 Los Angeles, CA 90095 Javier Miranda Center for Economic Studies U.S. Census Bureau 4600 Silver Hill Road Washington, D.C. 20233 Éva Nagypál Department of Economics Northwestern University 2001 Sheridan Road Evanston, IL 60208 Alfred Nucci Center for Economic Studies U.S. Census Bureau 4600 Silver Hill Road Washington, D.C. 20233 Mark J. Roberts Department of Economics 513 Kern Graduate Building Pennsylvania State University University Park, PA 16802 Marc Roemer U.S. Census Bureau 4600 Silver Hill Road Washington, D.C. 20233 Raﬀaella Sadun Centre for Economic Performance London School of Economics and Political Science Houghton Street London WC2A 2AE England Kristin Sandusky U.S. Census Bureau 4600 Silver Hill Road Washington, D.C. 20233

599

Stefano Scarpetta Head of Division, Country Studies 3 OECD Economics Department 2, rue André-Pascal 75775 Paris Cedex 16, France Peter K. Schott Yale School of Management 135 Prospect Street New Haven, CT 06520-8200 Donald S. Siegel Department of Management and Marketing A. Gary Anderson Graduate School of Management University of California, Riverside 221 Anderson Hall Riverside, CA 92521 Kenneth L. Simons Department of Economics 3407 Russell Sage Rensselaer Polytechnic Institute 110 8th Street Troy, NY 12180-3590 James R. Spletzer U.S. Bureau of Labor Statistics Postal Square Building 2 Massachusetts Avenue, NE Washington, D.C. 20212-0001 Spiro E. Stefanou Department of Agricultural Economics and Rural Sociology The Pennsylvania State University 208-B Armsby University Park, PA 16802 Bryce E. Stephens Bates White 1300 Eye Street, NW Suite 600 Washington, D.C. 20005 Martha Stinson U.S. Census Bureau 4600 Silver Hill Road Washington, D.C. 20233

600

Contributors

James Tybout Department of Economics Pennsylvania State University 517 Kern Graduate Building University Park, PA 16802 Lars Vilhuber School of Industrial and Labor Relations 358 East Ives Hall Cornell University Ithaca, NY 14850 Simon Woodcock Department of Economics Simon Fraser University 8888 University Drive Burnaby, British Columbia V5A 1S6 Canada

Daniel Yi Xu Department of Economics New York University New York, NY 10012 Jet Yee [email protected]

Author Index

Abbring, J. H., 263, 265 Abowd, J. M., 150, 151, 156, 183, 193, 195, 197, 199n26, 200, 231, 443n1, 448, 449, 450, 450n1, 452n2, 454, 466, 468 Abraham, K., 84, 106 Aguirregabiria, V., 308 Ahearn, M., 371, 373, 374, 375 Akerlof, G. A., 106 Allen, D. W., 392 Anderson, P., 98, 107, 447, 452n3 Antràs, P., 519 Asplund, M., 78, 261, 303, 304, 307 Audretsch, D. B., 43, 53n35, 448 Aw, B. Y., 16n2, 517n5 Axtell, R. L., 554 Bagwell, K., 242 Bailey, M., 2, 59, 298 Baldwin, J. R., 69, 79, 384, 402, 558, 572, 575n12, 582, 584, 590 Barkley, A. P., 370 Barnes, M., 277n10 Barro, R. J., 16n3 Bartelsman, E. J., 3, 4, 21, 64n41, 398n1 Basker, E., 143, 242 Bastos, F., 20n8 Basu, S., 269, 271, 300 Bayard, K., 239n1 Becker, G. S., 373 Beckstead, D., 575n12 Benedetto, G., 173, 451 Bentley, S. E., 370

Berger, M., 506n1 Bernard, A. B., 2, 20, 27, 448, 459, 514n2, 517n5, 544n18, 559, 572, 594 Berry, C. H., 573 Berry, S., 303, 304, 304n1, 305, 306n2, 307n3, 308, 319, 320 Betancourt, R., 286, 287 Bhagat, S., 402 Black, D., 506n1 Blanchard, O. J., 84, 106 Boden, R., 332 Bollman, R., 370n2 Boone, Z., 137 Borga, M., 519 Bosworth, B. P., 286 Bound, J., 506n1 Bowlus, A., 447, 451 Boylaud, O., 46n30 Bresnahan, T., 303, 304, 305, 310 Brown, C., 506n1 Brown, D. J., 16n2 Brown, S., 136 Bruno, M., 287 Bulow, J. I., 475 Burgess, S., 98, 107, 126, 479 Butani, S., 135, 136 Caballero, R., 2 Campbell, D., 2, 59 Campbell, J., 143, 242, 263, 265, 303, 306, 319, 321, 561 Carneiro, A., 448, 449, 468

601

602

Author Index

Caves, R. E., 372, 377, 387, 398n1, 557, 560, 572, 575n12, 582 Chavas, J. P., 392 Chiang, H., 201 Chung, S., 16n2, 517n5 Clark, K. A., 84, 84n1, 137 Clayton, R. L., 131, 136 Clerides, S. K., 517n5 Cochrane, W., 372 Collard-Wexler, A., 308 Corbel, P., 448 Cornwell, C., 476 Crankshaw, M., 85n6 Creecy, R. H., 450n1 Criscuolo, C., 273n4 Daly, D. J., 557 Dardia, M., 447, 451 Das, S., 2, 308 Davis, S. J., 2, 43n26, 84, 84n1, 84n3, 87, 87n8, 92n11, 94, 99, 101n15, 106, 125, 136, 137, 139, 145, 210, 229, 356n9, 360nn10–11, 447, 448 Debertin, D., 370 Decressin, A., 480, 481–482n9 Delgado, M. A., 517n5 Dhaliwal, N., 590 Diamond, P., 84, 106, 109 Dickens, W. T., 21n10, 475 Dinlersoz, E. M., 242 Dixit, A., 307 Dollar, D., 20n8 Doms, M., 2, 242, 398n1, 518 Doppelhoffer, G., 16n3 Dorsey, S., 476 Dumas, M., 242 Dunne, T., 1, 2, 53n35, 69, 79, 99, 125, 252, 252n9, 326, 360n10, 377, 383, 448, 468 Earle, J. S., 16n2 Eaton, J., 2, 517 Edwards, C., 370 Ehrenberg, R., 475n1 Ellickson, P. B., 242 El-Osta, H., 373 Engel, E. M. R. A., 2 Ericson, R., 372 Eslava, M., 16n2 Evans, D. S., 367 Even, W. E., 476 Faberman, R. J., 84, 87, 87n8, 88, 92n11, 94, 99, 101n15, 106, 136, 137

Fallick, B., 98, 447 Farinas, J. C., 517n5 Fernald, J. G., 269 Findeis, J., 373 Fisher, F. M., 398 Fleischman, C. A., 98 Foltz, J., 370, 393 Foote, C. L., 126 Foster, L., 59, 61n40, 64n41, 69, 242, 246, 289, 290, 373, 448 Frazis, H., 506n1 Freeman, R. B., 468 Friedman, B., 242 Fuglie, K., 371 Gardner, B. L., 372 Gautschi, D., 286 Geroski, P., 43 Goetz, S., 370 Goodwin, B., 373 Gorecki, P. K., 558 Gort, M., 404 Greenhut, M. L., 561 Griffith, R., 20, 272, 295, 298, 299 Griliches, Z., 2, 30, 60 Groshen, E. L., 21n10 Grossman, G. M., 519 Grossman, N., 447 Gu, W., 69, 572, 582, 584, 590 Gugler, K., 397 Gustman, A. L., 476 Hall, R. E., 84, 98, 106 Hallak, J. C., 554 Hallberg, M., 373 Hallward-Driemeier, M., 20n8 Haltiwanger, J., 2, 3, 4, 11, 20n8, 59, 61n40, 64n41, 69, 79, 84, 84n1, 84n3, 87, 87n8, 92n11, 94, 99, 101, 101n15, 106, 125, 136, 137, 139, 145, 150, 242, 246, 289, 290, 356n9, 360nn10–11, 373, 448 Hambrick, D. C., 404 Hanson, G., 375, 519 Harmgart, H., 272, 295, 298, 299 Harrigan, J., 554 Harris, R. D., 398 Haskel, J., 272, 272n1, 272n2, 273n4 Hayward, M. L. A., 404 Head, K., 560, 585–86, 593n2 Helpman, E., 519, 570n6, 572n7 Henrekson, M., 43n26 Holmes, T. J., 242, 365, 367, 405

Author Index Holtz-Eakin, D., 367 Holzer, H. J., 106 Hopenhayn, H., 78, 143, 242, 263, 303, 306, 308, 319, 321, 372, 561 Hoppe, R., 370 Hulten, C., 2, 59 Hummels, D., 554 Hung, C.-S., 561 Hyson, R., 84n1, 136, 137

603

Lengermann, P. A., 443, 443n2, 448, 449, 450, 454, 466 Lentz, R., 58 Levins, R., 372 Levinsohn, A., 58 Lichtenberg, F. R., 398, 401, 402, 405, 413 Link, A., 405n2 Lowenstein, M. A., 506n1 Lueck, D., 392 Lynch, L. M., 11

Ippolito, R. A., 476 Jacobson, L. S., 443n2, 447, 452n3 Jacquemin, A. P., 573 Jarmin, R., 64n41, 242, 244, 246n5, 296, 299, 310, 310n6, 332, 332n1, 515, 519n8, 590 Jensen, J. B., 2, 448, 459, 514n2, 518, 544n18, 572 Jensen, M. C., 397, 404 Jones, C. J., 20, 27 Jones, G., 274 Joulfaian, D., 367 Jovanovic, B., 372, 405 Kane, T., 506n1 Kapani, V., 136 Keys, B. A., 557 Khawaja, N., 272n1 Kimhi, A., 370n2 Klapper, L., 48n32 Kleiner, M. M., 468 Kletzer, L. G., 447, 448 Klimek, S. D., 239n1, 242, 296, 299, 468 Knaup, A. E., 136 Konigsberg, S., 232 Korb, P., 370, 374 Kortum, S. S., 2 Kramarz, F., 443n1, 448, 450n1, 452n2 Krizan, C. J., 59, 61n40, 64n41, 69, 242, 246, 289, 290, 373, 448 Krugman, P., 570n6 Kuhn, P., 447 Lach, S., 517n5 Laeven, L., 48n32 LaLonde, R., 443n2, 447, 452n3 Lambson, V., 372 Lancaster, T., 117 Lane, J., 98, 107, 126, 150, 479 Lang, K., 475 Lass, D., 373 Leighton, L. S., 367

MacGarvie, M., 517, 553 Mackie, C., 11 Macpherson, D., 476 Maksimovic, V., 398, 402, 422 Malinoski, M., 287 Manne, H., 404 Markusen, J. R., 553 Martin, R., 30, 273n4, 277n10, 292 Mathiowetz, N., 506n1 Mazzeo, M. J., 306n2 McCue, K., 481–82n9 McGowan, J., 398 McGuckin, R. H., 398, 401, 402, 413, 448 McKinney, K. L., 449, 450, 454, 466 McLaughlin, K. J., 106 McWilliams, A., 397, 398 Meade, J. E., 404 Melitz, M. J., 552, 559, 560, 568, 572, 572n7, 586, 594 Mengistae, T., 20n8 Meyer, B., 98, 107, 447, 452n3 Micco, A., 48, 48n32 Miller, R., 16n3 Mira, P., 308 Miranda, J., 244, 246n5, 247n7, 296, 299, 310, 310n6, 332, 332n1, 515, 519n8 Mishra, A., 373 Montgomery, E., 475 Mortensen, D. T., 58, 84nn2–3, 109 Mousa, J. A., 136 Mueller, D. C., 404 Nagypál, E., 98, 106, 116 Nasir, J., 20n8 Nguyen, S. V., 398, 401, 402, 413, 448 Nicoletti, G., 20, 46n30 Nocke, V., 78, 261, 304, 307, 594–95 Norman, G., 561 Nucci, A., 332 Oi, W. Y., 286 Olley, G. S., 71

604

Author Index

Ostrovsky, M., 304, 308 Ottaviano, G. J. P., 559, 560, 562, 563, 563n1, 594 Oulton, N., 269 Oviedo, A. M., 61n39 Pages, C., 48, 48n32 Pakes, A., 71, 304, 308, 372 Partington, J., 277n11, 285n15 Pesendorfer, M., 308 Peterson, R. N., 370 Petrin, J., 58 Petrongolo, B., 120 Phillips, G., 398, 402, 422 Piazza, M. C., 136 Pindyck, R., 307 Pinkston, J. C., 135, 136 Pissarides, C., 84nn2–3, 109, 120 Portugal, P., 448, 449, 468 Rajan, R., 48n32 Rama, E., 590 Ramey, G., 242 Ravenscraft, D. J., 397 Redding, S., 20, 594 Regev, H., 2, 30, 60 Reiss, P., 303, 304, 304n1, 305, 310, 319, 320 Reznek, A. P., 413 Ries, J., 560, 585–86, 593n2 Roberts, M., 1, 2, 16n2, 53n35, 99, 125, 252, 252n9, 308, 360n10, 377, 383, 448, 468, 517n5, 561 Robertson, K., 131 Rogerson, R., 84n3 Roll, R., 404 Rose, A. K., 106 Rosen, H. S., 367 Rouse, C., 506n1 Rousseau, P., 405 Ruano, S., 517n5 Ruhm, C. J., 447 Ryan, S., 308 Sabourin, D., 590 Sadeghi, A., 136 Sadun, R., 272, 272n2 Sala-i-Martin, X., 16n3 Samuelson, L., 1, 53n35, 99, 125, 252, 252n9, 360n10, 377, 383, 448 Sandin, R., 303 Sandusky, K., 201 Saupe, W., 370

Savage, G., 447 Scarpetta, S., 3, 4, 16n3, 20, 46, 46n30, 64n41 Schank, T., 64n41 Scherer, F. M., 397, 560 Schmidt-Dengler, P., 308 Schmitz, J. A., Jr., 365, 367, 405 Schoar, A., 422 Schoeni, R. F., 447, 451 Schott, P. K., 554, 572, 594 Schuh, S., 2, 125, 136, 139, 145, 356n9, 448 Schumpeter, J. A., 372 Schweiger, H., 20n8 Scott, F., 506n1 Seim, K., 306 Shaw, K., 475 Shimer, R., 84, 84n3, 87n8, 98, 106 Shleifer, A., 398, 402 Siegel, D., 397, 398, 401, 402, 405, 405n2, 413 Sieling, M. B., 242 Skiba, A., 554 Slaughter, M. J., 518 Smith, M. G., 370 Smith, R., 475n1 Spence, E. J., 557 Spletzer, J. R., 96n14, 126, 131, 135, 136, 145 Spulber, D. F., 242 Srinivasan, S., 269 Staiger, D., 506n1 Stamas, G., 85n6 Stanton, B. F., 375 Steinmeier, T. L., 476 Stephens, B. E., 197, 199n26, 200 Stephens, M., Jr., 447 Stevens, D., 98, 107, 126, 155, 479 Stewart, R. T., 447 Stinson, M., 481–82n9 Strauss, J., 373 Sullivan, D., 443n2, 447, 452n3 Summers, L. H., 475 Suner, D., 375 Sutton, J., 35n21, 305 Sweetman, A., 447 Syverson, C., 303, 306, 321, 560, 561 Szeidle, A., 519 Tamer, E., 305 Tang, J., 590 Thisse, J.-F., 562, 563, 563n1 Trefler, D., 274

Author Index Triplett, J. E., 286 Tybout, J., 2, 16n2, 308, 517n5, 561, 593 Valetta, R., 84 Van Biesebroeck, J., 440 Van Reenen, J., 20 Vilhuber, L., 151, 156, 193, 195, 197, 199n26, 200, 201, 231, 443, 443n2, 447, 448, 451, 452n2 Vishny, R. W., 402 Wagner, J., 517n5 Waldfogel, J., 303 Wallsten, S., 20n8 Werking, G., 136

Wohlford, J., 87, 88, 111 Wolf, C., 375 Wright, M., 398 Wright, R., 84n3 Wu, H., 370 Xu, L. C., 20n8 Yeap, C., 263 Yeaple, S., 572n7, 594–95 Yee, J., 374, 375 Yellen, J. L., 106 Zeile, W. J., 518, 519

605

Subject Index

Page references followed by t or f refer to tables and figures, respectively. Agriculture industry: comment on, 391–94; concluding remarks and summary, 387–89; defining, 371–72; empirical evidence for 1978–97 in, 376–78, 376nn4–6, 377f, 378n7, 379t, 380t, 382t, 384t, 385t, 386f, 386n8; introduction to, 369–70nn1–2, 369–71; measurement issues about data source and farm size in, 374–76, 374n3; relevance of theoretical frameworks for, 372–74 BED (Business Employment Dynamics), 2 Beveridge Curve, 89–96, 89f, 90f, 91t, 92nn11–12, 93t, 94t, 95f, 95n13 BLS (Bureau of Labor Statistics): business employment dynamics data and, 131– 37, 132t, 133t, 134f, 135f, 136f; business employment dynamics program at, 126–31; labor market studies and, 83 Business employment dynamics: BLS data and, 131–37, 132t, 133t, 134f, 135f, 136f; BLS program and, 126–31; concluding remarks on, 145; distribution of gross job gains/losses and, 137–45, 140f, 141f, 143f, 144f; introduction to, 125–26 Business Employment Dynamics (BED), 2

Canadian Economic Council, 558 CC (Competition Commission), 282n14, 294, 294n21, 295 CEW (Covered Employment and Wages) program, 153, 154 CPR (Composite Person Record), 154 CRIW (Conference on Research in Income and Wealth), 2 Cross-country differences: comment on, 77–79; concluding remarks on, 72–74; creative destruction process assessment and, 32–58, 33–35nn17–20, 35n22, 36t, 37t, 38n23, 39t, 40t, 41t, 42t, 42– 43nn24–27, 44f, 45f, 46nn28–30, 47f, 48n31, 49t, 50t, 51t, 52f, 53nn33–35, 54t, 55t, 56t, 57f, 57n36; creative destruction process effects on productivity and, 58–72, 59nn37–38, 61nn39–40, 62f, 63f, 64n41, 66t, 67f, 68f, 69n42, 69t, 72f; data description and, 22–27, 23nn11–12, 25t, 26t; distributed microdata analysis and, 19–22, 20–21nn6– 10; introduction to, 15–19, 16–17nn2– 5, 18f; measurement problems and, 27–30nn13–16, 27–32 ECF (Employer Characteristics File), 166–67, 167–68nn14–15, 171n16, 201– 2, 203

607

608

Subject Index

EHF (Employment History File), 159–60, 159nn6–8, 203 ES-202 program, 153, 154 EUKLEMS, 271, 299 EUROSTAT (EU Statistical Office), 19, 20, 20n7, 22, 23n11, 53n33 Firm deaths: analysis, 455–67, 455t, 456t, 457t, 458t, 460t, 461t, 463t, 464t, 465t, 467t; concluding remarks on, 467–68; data, 453–55; definitions, 450–53, 450n1, 452nn2–3; introduction to, 447– 49; multiple displacement firms and, 468–70, 469t, 470t Firm recruitment: concluding remarks on, 122; introduction to, 109–10; JOLTS, and matching function study for, 118– 22, 120t, 121f; JOLTS job openings measurements and, 115–18, 115nn2–3, 117f, 119f; JOLTS turnover data consistency and, 110–15, 112f, 113n1 Fringe benefits: appendix of tables, 503t, 504t; background to, 475–78, 475nn1– 2; comment on, 505–8, 506n1, 507t; data, 478–82nn3–10, 478–85, 482t, 483t, 484n11, 485nn13–15; introduction to, 473–75; results, 485–502, 486t, 487n16, 488t, 490t, 491t, 492n17, 493t, 494nn18–20, 495t, 496n21, 497t, 498t, 499t, 500t; summary, 502 GAL (Geocoded Address List), 163–66, 163–64nn12–13, 164t, 166t, 202 Health services industries: concluding remarks on, 325–26; empirical evidence of entry, exit, and market structure for, 317–25, 318f, 319f, 320t, 322t, 324t; empirical model of entry, exit, and number of firms for, 316–17; entry and exit measuring for dentists and chiropractors in, 309–16, 310–11nn6–7, 311f, 312f, 313–14nn8–9, 314f, 315f; introduction to, 303–5, 304n1; market structure models for entry and exit in, 305–9, 306–9nn2–5 Human capital: comment on, 443–45; 443– 44nn1–3; concluding remarks on, 439– 41, 439–40nn19–20; data on, 406–11, 406n3, 407t, 408t, 409t, 410nn4–5; econometric models and, 411–15, 411n6, 414–15nn7–9; empirical results

and, 415–39, 415n10, 416t, 417t, 418t, 418–19nn11–13, 420t, 421n14, 423f, 424f, 425t, 426f, 427f, 428f, 429f, 430– 32nn15–18, 431t, 433t, 434t, 436t, 437t, 438t; impact of ownership change theories on economic performance and, 404–6, 405n2; introduction to, 397–99; plant-level studies and, 399–404, 400t, 403t ICF (Individual Characteristics File), 160– 63, 160–63nn9–12, 202–3 JOLTS (Job Openings and Labor Turnover Survey): concluding remarks on, 104– 7, 105f, 122; introduction to, 2–3; job openings measurements for firm recruitment and, 115–18, 115nn2–3, 117f, 119f; labor market studies and, 87–89, 88f; matching function study for firm recruitment and, 118–22, 120t, 121f; turnover data consistency for firm recruitment and, 110–15, 112f, 113n1, 114f, 115f Labor market studies: concluding remarks on, 104–7, 105f; data and measurement for, 85–89, 85nn4–5, 86–88nn7–10, 88f; introduction to, 83–85, 84nn1–3; vacancies and the Beveridge Curve and, 89–96, 89f, 90f, 91t, 92nn11–12, 92t, 93t, 94t, 95f, 95n13; worker flows and, 96–104, 96n14, 97f, 100f, 101n15, 101f, 102f, 102t, 103nn16–18, 103t, 104f LBD (Longitudinal Business Database), 238, 243–50, 244nn2–3, 246–47nn4–7, 248t, 249n8, 548–49 LED (Local Employment Dynamics), 150 LEHD (Longitudinal Employer-Household Dynamics): fundamental concepts of, 205–15; introduction to, 2; job flow, worker flow, and earnings statistics definitions and, 215–29 LEHD infrastructure files: comment on, 230–34; concluding remarks on, 202–4; creation of, 158–72, 158–64nn5–13, 159n6, 164t, 165t, 166t, 167–68n14, 171n16; input files and, 152–58, 152n2, 154nn3–4, 158n5; introduction to, 149–52, 150n1; job-level missing data completion and, 172–79, 173n17, 175n18, 177–78nn19–23; public-use

Subject Index files and, 200, 200n27; QWI, and disclosure avoidance procedures for, 183– 87, 184t, 185f; QWI, and file analysis for, 188f, 189f, 190f, 191f, 192f; QWI, and forming aggregated estimates for, 179–82, 179–80nn24–25, 181f; QWI file analysis and, 187–200, 194t, 195t, 198t, 199n26, 199t; restricted-access files and, 200–202 Longitudinal Business Database (LBD), 2, 238, 243–50, 244nn2–3, 246–47nn4–7, 248t, 249n8, 548–49 Longitudinal Employer-Household Dynamics. See LEHD (Longitudinal Employer-Household Dynamics) Multinationals: comment on, 552–55; concluding remarks on, 546; data, 519–22, 520nn9–10, 521t; data appendix, 547– 51, 547–48nn19–20, 550n21, 550t; importer and exporter activity and, 522–33, 522nn11–12, 523t, 524t, 525t, 526t, 526–27nn13–15, 528t, 530t, 531t, 532nn16–17, 533t; importer and exporter dynamics and, 542–46, 543t, 544n18, 545t, 546t; introduction to, 513–16, 513n1, 514t, 515–16nn3–4; research on, 517–19, 517–18nn5–7; role of, 533–42, 534t, 535t, 536f, 537f, 538f, 539t, 540t, 541t OECD (Organization for Economic Cooperation and Development), 3, 18, 19, 21 ONS (Office for National Statistics), 285, 285n15, 292, 296 PCF (Person Characteristics File), 154 Producer dynamics: concluding remarks on, 10–11; cross-country comparison of, 3–4; empirical study of, 1–2; employer-employee dynamics and, 8–9; employment dynamics and, 4–6; in international markets, 9–10; sector studies of producer turnover and, 6–8 QCEW (Quarterly Census of Employment and Wages), 150, 153–54 QWI (Quarterly Workforce Indicators): aggregated estimates and, 179–80nn24– 25, 179–82, 181f; comment on, 230–34;

609

disclosure avoidance procedures and, 183–87, 184t, 185f; file analysis, 187– 200, 188f, 189f, 190f, 191f, 192f, 194t, 195t, 198t, 199n26, 199t; introduction to, 3, 149–50, 151, 179–80, 179n24 Retail chains: comment on, 262–69; concluding remarks on, 259–61, 260t; data and measurement issues and, 243–50, 244nn2–3, 246–47nn4–7, 248t, 249n8; introduction to, 237–38; results and, 250–59, 251f, 252n9, 253t, 254t, 255t, 256t, 257n10, 258t; U.S. retail sector trends and, 238–43, 239n1, 239f, 240f, 241f Small business dynamics. See Young and small business dynamics SSEL (Standard Statistical Establishment List), 548–49 Staff of the LEHD Program, 480n5 Trade: closed economy model and, 560–67, 563–65nn1–4; comment on, 593–95, 593nn1–2, 595n3; concluding remarks on, 588–91, 589t; data, 572–74, 572– 74nn7–9; empirical results and, 574– 88, 574–75nn9–13, 576f, 577f, 578f, 579f, 580f, 583t, 584n14, 585t, 587t, 588t; introduction to, 557–60; open economy model and, 567–72, 567n5, 570n6 U.K. retailing: concluding remarks on, 300–301; data for, 271–74nn3–6, 272– 82, 273t, 275f, 276–78nn7–13, 279t, 280t, 281t; entry and exit in, 282–84, 282n14, 283t, 284t; introduction to, 271–72, 271n1; productivity and, 285–89, 285n15, 286t, 288t; sources of productivity growth in, 289–93, 289t, 290nn16–18, 291t, 292–93nn19– 20; store size planning in, 294–300, 294n21, 296n22, 297t, 298t, 299t UNIDO (United Nations Industrial Development organization), 19 World Bank, 18, 21 Young and small business dynamics: comment on, 366–68; concluding remarks on, 365–66; constructing integrated

610

Subject Index

Young and small business dynamics (cont.) longitudinal business database and, 331–38, 332nn1–2, 333t, 334t, 335n3, 335t, 336t; employers and nonemployers basic facts and, 338–48, 338n4, 339t, 340n5, 341t, 342t, 343t, 344t, 345f, 347f, 348f; introduction to, 329–

31; issues in universes of, 362–65, 365n12; ownership links and transition dynamics for, 348–60, 349f, 350t, 352nn6–7, 353f, 354n8, 354f, 355f, 356n9, 357t, 358t, 359t, 359f; revenue growth and dispersion by age and size for, 360–62, 360nn10–11, 361f, 362f

The Economics of New Goods (National Bureau of Economic Research Studies in Income and Wealth)

International and Interarea Comparisons of Income, Output, and Prices (National Bureau of Economic Research Studies in Income and Wealth)

Modeling the Distribution and Intergenerational Transmission of Wealth (National Bureau of Economic Research - Studies in Income and Wealth, Volume 46)

Producer Dynamics: New Evidence from Micro Data (National Bureau of Economic Research Studies in Income and Wealth, Vol 68)

The Economics of New Goods (National Bureau of Economic Research Studies in Income and Wealth)

Measuring Capital in the New Economy (National Bureau of Economic Research Studies in Income and Wealth)

New Developments in Productivity Analysis (National Bureau of Economic Research Studies in Income and Wealth)

Business Cycles, Indicators, and Forecasting (National Bureau of Economic Research Studies in Income and Wealth)

Price Measurements and Their Uses (National Bureau of Economic Research Studies in Income and Wealth)

The Measurement of Labor Cost (National Bureau of Economic Research Studies in Income and Wealth)

Reducing Inflation: Motivation and Strategy (National Bureau of Economic Research Studies in Income and Wealth)

Monetary Policy Rules (National Bureau of Economic Research Studies in Income and Wealth)

Monetary Policy Rules (National Bureau of Economic Research Studies in Income and Wealth)

Labor Statistics Measurement Issues (National Bureau of Economic Research Studies in Income and Wealth)

A New Architecture for the U.S. National Accounts (National Bureau of Economic Research Studies in Income and Wealth)

International and Interarea Comparisons of Income, Output, and Prices (National Bureau of Economic Research Studies in Income and Wealth)

Modeling the Distribution and Intergenerational Transmission of Wealth (National Bureau of Economic Research - Studies in Income and Wealth, Volume 46)

Productivity Growth in Japan and the United States (National Bureau of Economic Research Studies in Income and Wealth)

Globalization and Poverty (National Bureau of Economic Research Conference Report)

Mergers and Acquisitions (National Bureau of Economic Research Project Report)

Population and Economic Change in Developing Countries (National Bureau of Economic Research Universities-National Bureau Conference Ser)

Hard-to-Measure Goods and Services: Essays in Honor of Zvi Griliches (National Bureau of Economic Research Studies in Income and Wealth)

Studies of Supply and Demand in Higher Education (National Bureau of Economic Research Project Report)

Taxing Multinational Corporations (National Bureau of Economic Research Project Report)

Trade and Employment in Developing Countries, Volume 1: Individual Studies (National Bureau of Economic Research Monograph)

Foreign Direct Investment (National Bureau of Economic Research Project Report)

Studies in State and Local Public Finance (National Bureau of Economic Research Project Report)

Social Experimentation (National Bureau of Economic Research Conference Report)

Agglomeration Economics (National Bureau of Economic Research Conference Report)

Social Experimentation (National Bureau of Economic Research Conference Report)

Income and Wealth

Studies in State and Local Public Finance (National Bureau of Economic Research Project Report)

International Capital Flows (National Bureau of Economic Research Conference Report)

Japanese Monetary Policy (National Bureau of Economic Research Project Report)

Producer Dynamics: New Evidence from Micro Data (National Bureau of Economic Research Studies in Income and Wealth, Vol 68)

The Economics of New Goods (National Bureau of Economic Research Studies in Income and Wealth)

Measuring Capital in the New Economy (National Bureau of Economic Research Studies in Income and Wealth)

New Developments in Productivity Analysis (National Bureau of Economic Research Studies in Income and Wealth)

Business Cycles, Indicators, and Forecasting (National Bureau of Economic Research Studies in Income and Wealth)

Price Measurements and Their Uses (National Bureau of Economic Research Studies in Income and Wealth)

The Measurement of Labor Cost (National Bureau of Economic Research Studies in Income and Wealth)

Reducing Inflation: Motivation and Strategy (National Bureau of Economic Research Studies in Income and Wealth)

Monetary Policy Rules (National Bureau of Economic Research Studies in Income and Wealth)

Monetary Policy Rules (National Bureau of Economic Research Studies in Income and Wealth)

Labor Statistics Measurement Issues (National Bureau of Economic Research Studies in Income and Wealth)

A New Architecture for the U.S. National Accounts (National Bureau of Economic Research Studies in Income and Wealth)

International and Interarea Comparisons of Income, Output, and Prices (National Bureau of Economic Research Studies in Income and Wealth)

Modeling the Distribution and Intergenerational Transmission of Wealth (National Bureau of Economic Research - Studies in Income and Wealth, Volume 46)

Productivity Growth in Japan and the United States (National Bureau of Economic Research Studies in Income and Wealth)

Globalization and Poverty (National Bureau of Economic Research Conference Report)

Mergers and Acquisitions (National Bureau of Economic Research Project Report)

Population and Economic Change in Developing Countries (National Bureau of Economic Research Universities-National Bureau Conference Ser)

Hard-to-Measure Goods and Services: Essays in Honor of Zvi Griliches (National Bureau of Economic Research Studies in Income and Wealth)

Studies of Supply and Demand in Higher Education (National Bureau of Economic Research Project Report)

Taxing Multinational Corporations (National Bureau of Economic Research Project Report)

Trade and Employment in Developing Countries, Volume 1: Individual Studies (National Bureau of Economic Research Monograph)

Foreign Direct Investment (National Bureau of Economic Research Project Report)

Studies in State and Local Public Finance (National Bureau of Economic Research Project Report)

Social Experimentation (National Bureau of Economic Research Conference Report)

Agglomeration Economics (National Bureau of Economic Research Conference Report)

Social Experimentation (National Bureau of Economic Research Conference Report)

Income and Wealth

Studies in State and Local Public Finance (National Bureau of Economic Research Project Report)

International Capital Flows (National Bureau of Economic Research Conference Report)

Japanese Monetary Policy (National Bureau of Economic Research Project Report)

Recommend Documents