Urban and Regional Policy and Its Effects

VOLUME 1

U R BAN AN D R E G I O NAL PO LI CY AN D ITS E F F E CTS

M A R G E RY AU S T I N T U R N E R , H OWA R D W I A L , A N D H A R O L D WO L M A N EDITORS

URBAN AND REGIONAL POLICY AND ITS EFFECTS

VOLUME ONE

URBAN AND REGIONAL POLICY AND ITS EFFECTS margery austin turner howard wial harold wolman editors

brookings institution press Washington, D.C.

Copyright © 2008

the brookings institution 1775 Massachusetts Avenue, N.W., Washington, D.C. 20036 www.brookings.edu All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means without permission in writing from the Brookings Institution Press.

Library of Congress Cataloging-in-Publication data Urban and regional policy and its effects / Margery Austin Turner, Howard Wial, and Harold Wolman, editors. p. cm. Summary: “Brings policymakers, practitioners, and scholars up to speed on the state of knowledge on urban and regional policy issues. Conceptualizes fresh thinking of different aspects (economic development, education, land use), presenting main themes and implications and identifying gaps to fill for successful formulation and implementation of urban and regional policy”—Provided by publisher. Papers originally presented at a conference held at the Brookings Institution on March 29–30, 2007. Includes bibliographical references and index. ISBN 978-0-8157-8601-6 (pbk. : alk. paper) 1. Urban policy—Congresses. 2. Urban economics—Congresses. 3. Urban renewal— Congresses. 4. Regional planning—Congresses. 5. City planning—Congresses. I. Turner, Margery Austin, 1955– II. Wial, Howard. III. Wolman, Harold. IV. Title. HT151.U65 2008 338.973009173'2—dc22

2008016030 987654321

The paper used in this publication meets minimum requirements of the American National Standard for Information Sciences—Permanence of Paper for Printed Library Materials: ANSI Z39.48-1992. Typeset in Adobe Garamond Composition by Peter Lindeman Arlington, Virginia Printed by R. R. Donnelley Harrisonburg, Virginia

Contents

Preface

ix

1

Introduction Margery Austin Turner, Howard Wial, and Harold Wolman

1

2

“Eds and Meds” and Metropolitan Economic Development Timothy J. Bartik and George Erickcek

21

3

Low-Income Homeownership as an Asset-Building Tool: What Can We Tell Policymakers? George C. Galster and Anna M. Santiago

60

4

Tax and Expenditure Limitations and Their Effects on Local Finances and Urban Areas David Brunori, Michael Bell, Joseph Cordes, and Bing Yuan

109

5

Preschool Education and Human Capital Development in Central Cities Clive Belfield

155

6

Can Economically Integrated Neighborhoods Improve Children’s Educational Outcomes? Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel

181

7

Spatial Development and Energy Consumption Elena Safirova, Sébastien Houde, and Winston Harrington

206

Index

247

v

Urban and Regional Policy and Its Effects, volume 1 is the first in a series of publications that provides scholars, policymakers, and practitioners with accessible summaries of what is known about the effectiveness of selected urban and regional policies. This volume contains edited versions of the papers presented at a conference held at the Brookings Institution on March 29–30, 2007, and arranged by the editors. The conference and this volume are products of collaboration between the Brookings Institution’s Metropolitan Policy Program, the George Washington University’s George Washington Institute of Public Policy and Trachtenberg School of Public Policy and Public Administration, and the Urban Institute. All the papers represent the views of the authors and not necessarily the views of the staff members, officers, or trustees of the Brookings Institution, the George Washington University, or the Urban Institute. Coeditors

Staff

Advisers

Margery Austin Turner, Urban Institute Howard Wial, Brookings Institution Harold Wolman, George Washington University Emily Roessel, senior research verifier Eric Haven, senior research verifier Emily Garr, intern Amy Liu, Brookings Institution Alan Berube, Brookings Institution Nancy Y. Augustine, George Washington University Nancy Pindus, Urban Institute

Contributors

Timothy J. Bartik, W.E. Upjohn Institute for Employment Research Clive Belfield, Queens College, City University of New York Michael Bell, George Washington University David Brunori, George Washington University Joseph Cordes, George Washington University Ingrid Gould Ellen, New York University George Erickcek, W.E. Upjohn Institute for Employment Research George C. Galster, Wayne State University Winston Harrington, Resources for the Future Sébastien Houde, Resources for the Future Elena Safirova, Resources for the Future Anna M. Santiago, Wayne State University Amy Ellen Schwartz, New York University Leanna Stiefel, New York University Bing Yuan, George Washington University

Discussants

William Barnes, National League of Cities Timothy J. Bartik, W.E. Upjohn Institute for Employment Research Alan Berube, Brookings Institution Peter Edelman, Georgetown University Richard Green, George Washington University

Royce Hanson, Montgomery County, Maryland, Planning Board Edward W. Hill, Cleveland State University Daniel Muhammad, Office of the Chief Financial Officer, District of Columbia Nancy Pindus, Urban Institute Robert Puentes, Brookings Institution Amy Ellen Schwartz, New York University Margery Austin Turner, Urban Institute Conference Participants

Nancy Y. Augustine, George Washington University Pamela Blumenthal, George Washington University Stephanie Cellini, George Washington University David Garrison, Brookings Institution Anna Karruz, George Washington University Richard McGahey, Ford Foundation Kathryn Newcomer, George Washington University Erika Poethig, John D. and Catherine T. MacArthur Foundation Andrew Reamer, Brookings Institution Jennifer Vey, Brookings Institution Elaine Weiss, George Washington University Garry Young, George Washington University

Preface

W

ith this volume, the Brookings Institution’s Metropolitan Policy Program, the George Washington University’s George Washington Institute of Public Policy and Trachtenberg School of Public Policy and Public Administration, and the Urban Institute launch a series of publications on Urban and Regional Policy and Its Effects. The series is intended to be distinctive in two ways. First, it focuses on particular public policies and their effects rather than on general policy-relevant background research on urban and regional phenomena. Second, it is designed to be accessible and useful to policymakers and practitioners as well as to scholars. Therefore, it devotes more attention to accessible syntheses of what is known about policies and their effects than to the results of new, previously unpublished scholarly research. This volume and the conference on which it is based came about with the support of several people at the sponsoring institutions. At Brookings, Bruce Katz, director of the Metropolitan Policy Program, has provided the program’s support for this project. Amy Liu helped conceive the conference and provided us with invaluable intellectual and practical assistance in planning it. Alan Berube provided essential intellectual advice and also served as a discussant at the conference. At George Washington University, Joe Cordes, director of the Trachtenberg School of Public Policy and Public Administration, provided useful advice throughout. Nancy Augustine not only made important intellectual contributions but also played a major role in organizing the conference. ix

x

Preface

At the Urban Institute, Robert Reischauer, president, provided institutional support for this project, and Nancy Pindus gave us useful advice on the editors’ summary of this volume. A number of other people were instrumental in making the conference and this volume a reality. Matt Taverna, formerly with Brookings, managed the conference logistics. Jamaine Fletcher and Elena Sheridan of Brookings and Olive Cox and Kim Rycroft of the George Washington Institute of Public Policy provided administrative support for the conference and throughout the publication process. Emily Garr, formerly with Brookings, helped prepare the conference papers for publication. Janet Walker of the Brookings Institution Press expertly and gracefully managed the production of the conference volume. We are grateful to the John D. and Catherine T. MacArthur Foundation, whose generous support of the Brookings Institution’s Metropolitan Economy Initiative made Brookings’ cosponsorship of the conference and editorial work on this volume possible. We also thank the Fannie Mae Foundation, George Gund Foundation, Heinz Endowments, and Rockefeller Foundation for providing general support to the Brookings Metropolitan Policy Program. We thank the George Washington University for providing funding for the conference through its Selective Excellence Program.

URBAN AND REGIONAL POLICY AND ITS EFFECTS

1 Introduction margery austin turner, howard wial, and harold wolman

T

he problems faced by metropolitan regions and their residents are well known and have been the subject of substantial research as well as popular discussion. However, in both urban and regional domains, much more attention is paid to the nature of the problems themselves and of policies to address them than to the effects that such policies might actually have. The consequence is that policy deliberations and decisions are frequently based on the elegance of policy design or (perhaps more frequently) the eloquence of policy advocates rather than on solid empirical evidence. At the extreme, this leads to the creation of urban and regional policy fads that are widely believed to be effective approaches but that lack a solid evidentiary base. To fill this gap and to assist policymakers in addressing important urban and regional problems, the Brookings Institution, the George Washington University Institute of Public Policy and the Trachtenberg School of Public Policy and Public Administration, and the Urban Institute held the first of a series of annual conferences on Urban and Regional Policy and Its Effects at the Brookings Institution in Washington, D.C., on March 29–30, 2007. Chapters were commissioned for the conference from distinguished social scientists. The conference sought to engage authors and discussants in a cross-disciplinary dialogue focused on the central theme—evidence of policy effects. The chapters in this volume are revised versions of those commissioned chapters. Our examination of urban and regional policy and its effects is organized around six key policy challenges that most metropolitan areas and local communities face. Each of the chapters in this volume deals with a specific policy topic under one of these challenges: 1

2

Margery Austin Turner, Howard Wial, and Harold Wolman

—Growing a competitive economy through industry-based strategies, represented in this volume by “‘Eds and Meds’ and Metropolitan Economic Development,” by Timothy J. Bartik and George Erickcek —Growing the middle-class, represented by “Low-Income Homeownership as an Asset-Building Tool: What Can We Tell Policymakers?” by George C. Galster and Anna M. Santiago —Governing effectively, represented by “Tax and Expenditure Limitations and Their Effects on Local Finances and Urban Areas,” by David Brunori, Michael Bell, Joseph Cordes, and Bing Yuan —Building human capital, represented by “Preschool Education and Human Capital Development in Central Cities,” by Clive Belfield —Creating quality neighborhoods for families, represented by “Can Economically Integrated Neighborhoods Improve Children’s Educational Outcomes?” by Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel —Managing the spatial pattern of metropolitan growth and development, represented by “Spatial Development and Energy Consumption,” by Elena Safirova, Sébastien Houde, and Winston Harrington The goal of this volume is to inform scholars, policymakers, and practitioners about the state of knowledge on the effectiveness of selected policy approaches, reforms, or experiments on key social and economic problems facing cities, suburbs, and metropolitan areas. Authors were not required to engage in original research, although some did so. Rather, their task was to take a fresh and unfettered look at the area and conceptualize (or reconceptualize) how the issue should be thought of and what questions should be asked to influence intelligent public debate. Given that conceptualization, the authors were then asked to summarize extant research on the topic and, on the basis of that research and their own knowledge, to set forth what is known about the effects of the urban policy approach under discussion and the policy implications of what is known. They were also asked to identify what is still not known but important to find out.

Summary of Chapters The increasing geographic mobility of many export-base industries, including tradable services, is of growing public policy concern, especially in older, highercost metropolitan areas. In many such areas, economic developers are seeking other industries to supplement or replace manufacturing and offshorable services as mainstays of the local economy. Hospitals and universities seem attractive as potential drivers of metropolitan economic development because they employ large numbers of people (especially in central cities), they are growing, they are at least partially export industries, they are not geographically mobile,

Introduction

3

and they offer or have the potential to offer relatively high-wage jobs to less educated workers. In “‘Eds and Meds’ and Metropolitan Economic Development,” Timothy J. Bartik and George Erickcek examine the impact of the health care and higher education industries (“eds and meds”) on the economic development of metropolitan areas. Bartik and Erickcek begin by stating their view that the goal of economic development should be to increase the earnings of a metropolitan area’s existing residents, not simply to increase overall metropolitan earnings, income, or employment. Given this view, they then outline four possible mechanisms by which health care and higher education could contribute to realizing this goal. First, eds and meds could increase existing residents’ earnings because, as (partial) export industries, they bring income into a metropolitan area by serving people who live elsewhere. Second, both industries generate human capital (improved skills in the case of higher education and improved health in the case of health care), which has been shown to raise earnings. Third, both industries engage in research whose findings can be commercialized (for example, scientific and engineering research in universities and biomedical research in major teaching hospitals). If this commercialization occurs in the same metropolitan area where the university or hospital is located (for example, if faculty members or physicians start firms that commercialize their discoveries or if they license their intellectual property to other entrepreneurs), then the residents of that metropolitan area will benefit from economic activity that would not otherwise have occurred there. Finally, as large employers in many metropolitan areas, hospitals and universities can have a powerful impact on the wages, employee benefits, and working conditions offered by other employers in the same metropolitan area. If they offer relatively high wages and benefits and good working conditions, then they can raise labor standards throughout the region. Bartik and Erickcek describe the extent to which health care and higher education, respectively, are geographically concentrated within metropolitan areas. This is important information for metropolitan policymakers who want to understand how large each of these industries is in their own metropolitan area as compared with the size of these industries in other areas. Even more important, the information about the geographic concentrations of eds and meds sheds light on the extent to which each of these industries is an export industry for a particular metropolitan area. The higher the geographic concentration of an industry in a metropolitan area, the more likely it is that the industry is an export industry that brings in income from outside the region, as opposed to simply serving local residents. Bartik and Erickcek find that a number of metropolitan areas have high concentrations of higher education, while health care is more evenly spread out among metropolitan areas. This indicates that higher

4

Margery Austin Turner, Howard Wial, and Harold Wolman

education is likely to be an export industry for many metropolitan areas, while health care is likely to be an export industry for far fewer. Bartik and Erickcek then examine the evidence on each of the four mechanisms by which health care and higher education could contribute to metropolitan economic development. They find that health care and, to a greater degree, higher education are, in part, export industries. A public policy that expanded the size of either industry would raise metropolitan residents’ earnings. This would be the case even if local residents had to finance the entire cost of the expansion through taxes. Bartik and Erickcek find that the higher education industry also raises metropolitan residents’ earnings by improving their skills. This comes about in two ways. First, having an institution of higher education in a metropolitan area increases the share of local residents who earn college degrees and then remain in the area to work. Second, the colleges and universities in a metropolitan area draw in students from outside the area who earn degrees. Some of these people remain in the area to work just because they were educated in the area, raising the share of college-educated people in the local workforce. The higher share of college-educated people in the local workforce raises the productivity and earnings of existing residents. Because health, like education, raises earnings, health care institutions in a metropolitan area could also raise the area’s wages, but there is no evidence available on this topic. The authors survey the case study evidence on the impact of university research on local economic development. They find that research conducted in a metropolitan area is likely to have an important impact, which comes about both through universities’ technology transfer activities and through the broader engagement of universities with local firms. Workers in higher education, Bartik and Erickcek find, receive lower than average wages, even when their education and demographic characteristics are taken into account. Health care, in contrast, pays higher than average wages. Therefore, according to the authors, a public policy that expanded higher education in a metropolitan area would be likely to lower wages paid by other employers in the area, while a policy that expanded health care would be likely to raise them. Bartik and Erickcek conclude with an overall assessment of the economic development impacts of policies to expand eds and meds. They find that these policies would, on balance, raise the earnings of metropolitan residents. Although the impacts are modest in size, they are large enough to be important for policymakers.

Introduction

5

ever since the New Deal, federal housing policies have actively promoted homeownership. And today, home equity represents the primary source of wealth for many American families. Over the last two decades, federal policies have focused more explicitly on encouraging ever-lower-income households to buy a home, with the goal of enabling these families to build wealth and enhance their economic security. But skeptics have argued that these policies may be counterproductive, trapping low-income families in neighborhoods where appreciation rates of house values (and hence wealth gains) are low, saddling them with costly debt, and putting them at risk of financial distress. The recent meltdown in the subprime mortgage market heightens these concerns. Does owning a home help low-income families build the assets they need to weather short-term setbacks and achieve longer-term financial security, or does it perhaps put them at even greater risk? In “Low-Income Homeownership as an Asset-Building Tool: What Can We Tell Policymakers?” George C. Galster and Anna M. Santiago review the existing empirical evidence on the effectiveness of homeownership as an assetbuilding tool for low-income households and provide new evidence based on surveys of very-low-income families who purchased homes through a targeted program offered by the local public housing authority in Denver, Colorado. They begin by assessing the extent to which homeownership enables lowincome households to build wealth, concluding that owning a home does indeed generate financial returns for most low-income families, especially in comparison with renters with comparable income levels. However, wealth gains are generally lower for low-income owners than for those with higher incomes (and higher house values), lower for minorities than for Whites (even controlling for income), and lower in minority neighborhoods than in predominantly White neighborhoods. In addition, with few exceptions, significant wealth gains from homeownership accrue over the long term but are sensitive to ups and downs in the local housing market. Finally, low-income families typically do not enjoy the additional financial returns offered by the preferential tax treatment of homeownership because they do not itemize deductions and their tax liabilities are low. Although homeownership can—under the right circumstances—enable lowincome families to build wealth, Galster and Santiago’s review of the available evidence concludes that often it does not provide them a buffer to weather financial shocks. Many low-income families have insecure jobs and earnings, and housing costs consume a very large share of their income. These homeowners are particularly vulnerable to financial shocks, including unanticipated home repairs, illness or injury, job disruptions, and rising mortgage payments. Even those who manage to accumulate some home equity often have to spend it

6

Margery Austin Turner, Howard Wial, and Harold Wolman

down to cover short-term expenses. As a consequence, attrition rates among low-income homeowners are high, especially over the first six or seven years following home purchase. Not only does homeownership fail to provide low-income families with an effective financial buffer, but it can actually place them at greater risk of financial distress. In particular, some of the mortgage terms and conditions that have been developed to draw lower-income households into homeownership place these homeowners in a tenuous financial position. Galster and Santiago highlight the potential pitfalls of low or no equity at the time of purchase, higher (and variable) mortgage interest rates, and high fees. As recent developments in the subprime market demonstrate, these aggressive lending products may be manageable during periods of rapid appreciation in house prices, but they result in high rates of delinquency, foreclosure, and bankruptcy when home prices flatten or begin to fall. Again, the evidence suggests that the risks of foreclosure are higher for minority homeowners than for whites and higher in minority neighborhoods than in predominantly White neighborhoods. Although homeownership may not be a reliable strategy for building assets and financial security among low-income families in the short term, Galster and Santiago argue that it does appear to benefit children in ways that yield economic gains for the next generation. A growing body of rigorous statistical studies finds higher levels of positive behaviors, cognitive skills, educational attainment, and (ultimately) employment and earnings among children of homeowners, other things being equal. The existing research does not definitively explain how homeownership generates these benefits, and there are good reasons to suspect that some of the observed benefits in fact stem from residential stability, parental attention, or neighborhood characteristics rather than from homeownership per se. Nonetheless, homeownership itself appears to yield some positive effects, even after controlling for residential stability and family wealth. Finally, Galster and Santiago summarize what is known about the barriers low-income households face when they try to attain (and sustain) homeownership and the effectiveness of specific policy mechanisms for helping low-income households become homeowners. These include down payment constraints, insufficient or unreliable income, and the persistence of discrimination in home sales and mortgage transactions. Unfortunately, more is known about the effects of homeownership on the financial well-being of low-income households than on the relative effectiveness (or cost-effectiveness) of alternative strategies for promoting homeownership. The limited evidence that is available suggests that assistance subsidies for down payments can have substantial impacts on the attainment of homeownership and that some forms of prepurchase counseling can reduce the probability of default.

Introduction

7

After reviewing their evidence to date, Galster and Santiago argue that lowincome homeownership policies should be implemented in conjunction with programs that help low-income households increase their income, reduce their indebtedness, budget effectively, and accumulate some liquid assets before buying homes. In other words, programs should not simply promote homeownership alone but should make the achievement of homeownership part of a larger strategy to accumulate assets and achieve greater financial security. In addition, because the persistence of discrimination and segregation in urban housing markets may increase risks—and reduce potential gains—for low-income minority home buyers, Galster and Santiago recommend a greater emphasis be placed on the enforcement of fair housing and fair lending. california’s proposition 13, enacted by referendum in 1978, sparked a wave of tax and expenditure limitations in other states and localities during subsequent decades. In 2005 Colorado voters decided to suspend part of that state’s “taxpayer’s bill of rights.” Why do voters choose to limit the ability of their elected representatives to set levels of taxation and government spending? Should they? If such limits are to exist, how should they be designed? David Brunori, Michael Bell, Joseph Cordes, and Bing Yuan in their chapter, “Tax and Expenditure Limitations and Their Effects on Local Finances and Urban Areas,” describe the history of tax and expenditure limitations (TELs) and present a typology of the various kinds of TELs. They then explore the various theoretical justifications for TELs and the tension between TEL adoption and local autonomy, the motivation for adopting TELs, empirical evidence on their effects, and in particular their likely distributional effects on large urban areas. Brunori, Bell, Cordes, and Yuan begin by distinguishing between three different kinds of TELs: limitations on property assessment increases, limitations on property tax rates, and limitations on property tax revenue increases or on overall expenditures. They also distinguish between limits that are binding as opposed to those that are nonbinding and between limitations that have override mechanisms (either by referendum or by legislative vote) and those that do not. Brunori and his coauthors then discuss the commitment of the American public to local democracy and local autonomy, noting the tension between local autonomy and TELs, and ask why, in view of the commitment to local autonomy, there has nonetheless been a proliferation of tax and expenditure limitations that reduce local autonomy. They consider theoretical rationales and studies of voter behavior in state referendums on whether to adopt local TELs. The authors consider several outcome measures to assess the effect of TELs. First, in the most direct terms they ask whether TELs have succeeded in their primary purpose: to limit local government revenue and expenditure. After

8

Margery Austin Turner, Howard Wial, and Harold Wolman

reviewing the empirical literature, they conclude that TELs have constrained growth in property tax revenue, which has long been the main broad-based source of revenue for local governments, and that property tax assessment limits, when coupled with property tax rate limits, are particularly binding. However, local governments have reacted to such constraints by substituting other local though narrower revenue sources, such as fees and charges, and by relying increasingly on intergovernmental grants from state government. They also examine outcomes in terms of their effect on public schools. They report that TELs have constrained local spending on public schools, as measured by a variety of indicators such as student-teacher ratios, teacher salaries, and teacher quality. TELs are not only associated with reduced spending on education inputs but also with lower educational outcomes, as measured by test scores. Furthermore, the authors observe that enacting TELs creates both winners and losers among taxpayers. Potential winners include nonresident taxpayers, who include nonresident workers and absentee property owners. Residents of communities with the ability to export tax burdens to nonresidents are potential losers because those communities that were able to export their taxes before TELs through a property tax on a business with out-of-town owners, for example, are now prevented from doing so by the imposition of the TELs. The authors note that there is very little literature on the distributional effects of TELs. Evidence is mixed on whether TELs systematically favor lowincome taxpayers as opposed to high-income taxpayers. There is virtually no evidence on the distributional effects of TELs by place. Finally Brunori, Bell, Cordes, and Yuan turn to the question of policy design. They note that there is substantial public support for TELs (even among some local public officials) and that TELs are likely to continue to exist. Acknowledging the tension between public support for controls on local spending and the tradition of local autonomy, Brunori, Bell, Cordes, and Yuan suggest that allowing for meaningful local overrides of TELs and requiring provision of better information to taxpayers, through such means as full disclosure laws and benchmarking against best practices, may mitigate some of the adverse effects of overly strict TELs. policymakers increasingly are considering the economic as well as the social and developmental benefits of early childhood education. Economics Nobel laureate James Heckman has documented these economic benefits, which include improved educational attainment and higher earnings in adulthood. It has been shown that preschool education is a better economic development investment than are conventional economic development subsidy pro-

Introduction

9

grams. However, researchers and policymakers have paid little attention to the local and regional benefits and costs of preschool education. It seems intuitive that children in low-income jurisdictions, such as central cities, would receive great benefits over their lifetimes from being enrolled in preschool as children. If these children remain in central cities when they become working adults, then the cities themselves would also be expected to reap economic benefits. These intuitive ideas have never been examined carefully, though. Furthermore, what is true at the national level may not be true at the local or state level because the mobility of people across jurisdictional lines can alter the cost-benefit calculus considerably. In “Preschool Education and Human Capital Development in Central Cities,” Clive Belfield considers the benefits and costs of early childhood education for human capital development in central cities. He begins by describing preschool enrollment in the United States. About two-thirds of four-year-olds attend preschool, but there is substantial variation in enrollment among states and demographic groups, including among states with large urban populations and among low-income demographic groups. Consequently, there is no evidence that can be used to infer whether central city children are enrolled at higher or lower rates compared with the national average enrollment rate. Belfield then summarizes the research evidence on the consequences of preschool for participants and for local communities. These consequences are numerous, but the main effect is on human capital accumulation over the lifetime. Non-spatial research on the long-term economic benefits of preschool to participants concludes that center-based preschool education produces substantial gains in participants’ academic achievement during elementary school. These gains are greater for central city children than for suburban and rural children, and they persist longer for urban children (through at least fifth grade). Although the measured cognitive benefits of preschool may not persist beyond elementary school, research shows that preschool participants are less likely to be retained in grade, less likely to be placed in special education classes, less likely to drop out of high school, and more likely to attend college than are those who did not attend preschool. They also have lower rates of teen pregnancy and crime, and they enjoy better health. Preschool participants earn about $30,000 to $40,000 more over their lifetimes than non-participants. The private gains that preschool participants realize benefit society through greater economic output, but they also benefit central city taxpayers and adult residents directly. For example, they reduce expenditures on special education and grade retention, reduce crime, improve health, and increase the tax base of the central city (to the extent that children remain in the same central city in which they received preschool education when they become working adults).

10

Margery Austin Turner, Howard Wial, and Harold Wolman

These benefits, according to Belfield, are likely to be greater for central cities than for suburbs and rural areas. Research shows that these benefits exceed the costs of preschool on average. Belfield distinguishes between the average costs and benefits of preschool and the marginal costs and benefits of expanding preschool coverage. He concludes that although marginal costs are likely to exceed average costs and marginal benefits are likely to be below average benefits, there is still good reason to believe that expanding preschool coverage would be a good investment. Interjurisdictional mobility complicates the cost-benefit analysis if local or state governments are required to finance the costs of expanding preschool coverage. If central city taxpayers are responsible for financing preschool expansion, then the benefits to them of doing so may not exceed the costs if the recipients of preschool education do not remain in the same central city for at least part of their working lives. (A similar concern holds, although less strongly, if states are responsible for paying for preschool.) Belfield concludes that although many central city children who receive preschool education are likely to live and work elsewhere as adults, the rate of mobility is not high enough to overturn the conclusion that expanding preschool is a good investment from the point of view of the central city as a whole. However, this does not mean that interjurisdictional issues can be ignored. Drawing on a careful analysis of preschool financing in Washington, D.C., Belfield shows that only about half the benefits of expanding preschool accrue to the local school district; many of the benefits flow to the federal government in the form of increased federal tax revenues. In addition, only about half the benefits are realized by the time preschool recipients graduate from high school. Depending on the discount rate used to evaluate preschool investment, expanding preschool may not produce a high enough rate of return to the local school district’s taxpayers to induce those taxpayers to make the investment. Yet, because expanded preschool coverage benefits society as a whole, purely local financing may be inappropriate because it leads to too little preschool coverage. in many low-income neighborhoods, public schools perform poorly and children’s educational achievement suffers. A considerable body of research evidence supports the conclusion that concentrated neighborhood poverty and distress contribute to poor educational outcomes, other things being equal. “Can Economically Integrated Neighborhoods Improve Children’s Educational Outcomes?” by Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel, explores the potential of neighborhood economic integration as a strategy for improving educational outcomes. Their chapter synthesizes the existing research

Introduction

11

literature and presents new evidence on patterns of income composition in New York City neighborhoods and schools. Ellen, Schwartz, and Stiefel begin by reviewing the reasons why neighborhood economic integration may affect educational outcomes for children. One possible explanation is that because most students are assigned to schools on the basis of their residential location economically integrated neighborhoods might produce economically integrated schools, which might produce better educational outcomes. However, the evidence suggests that for a number of reasons an economically diverse neighborhood does not necessarily translate into an equally diverse school population. Higher-income households in a mixed neighborhood may be childless. They may send their children to private schools. Or the major differences in school composition may occur between school districts rather than within them. The authors’ original analysis of data on New York City neighborhoods and schools confirms that while school composition is clearly linked to neighborhood economic composition, neighborhood conditions and trends fail to explain variations in school composition fully. Even when neighborhood economic integration contributes to the economic integration of a school’s student population, questions remain about the potential impacts on children’s educational outcomes. A considerable body of evidence shows that after controlling for family income children attending schools with higher-income peers perform better on tests. However, efforts to establish a causal link between school composition and student performance yield mixed results. Nonetheless, a fairly strong case can be made that low-income students probably benefit when the schools they attend are economically integrated, because schools with higher-income students tend to have more resources and possibly because they offer opportunities for exposure to more economically diverse peer groups. Ellen, Schwartz, and Stiefel also assess the available evidence regarding a possible direct link between neighborhood composition and children’s educational outcomes, independent of the composition of the schools the children attend. Neighborhood safety, peer effects, or social networks could lead to better outcomes directly. However, the evidence on this possible causal mechanism remains ambiguous. Thus policies that promote neighborhood economic integration appear to have the potential to increase school economic integration and possibly educational outcomes, especially if the higher-income households that live in the neighborhood have children and send them to the nearby schools. Unfortunately, little rigorous evidence currently exists on the effectiveness of neighborhood economic integration strategies. Ellen and colleagues argue convincingly that research should focus on the impacts of initiatives, such as HOPE VI, on

12

Margery Austin Turner, Howard Wial, and Harold Wolman

neighborhood composition, school composition, and student performance. In addition, they suggest that policymakers also consider strategies for promoting the economic integration of schools directly, even in the absence of neighborhood integration. Such strategies include larger attendance zones, magnet schools, and other choice policies that loosen the link between a child’s place of residence and his or her school assignment. increased concerns with the effect of energy consumption on global warming and with the dependence of the United States on foreign energy sources have combined to make excessive energy consumption a major policy issue at the national level. In “Spatial Development and Energy Consumption,” Elena Safirova, Sébastien Houde, and Winston Harrington examine whether urban and regional policy can affect energy consumption and, in particular, whether policies affecting the spatial form of metropolitan areas can have important impacts on energy consumption in these areas. The authors begin by noting that the main link between energy use and urban form is vehicle emissions from urban transportation. The secondary link is through heating requirements, with multifamily dwellings requiring less energy to heat or cool per household than single family dwellings. After reviewing existing empirical literature on both of these links and finding the evidence inconclusive, they observe that a better way to test the effects of changes in urban form on energy consumption would be through use of a simulation model. Such a model would be able to examine the effects of different types of urban form or different types of urban policies on an area’s energy consumption. To accomplish this, the authors use an integrated model of land use, economic activity, and transportation that simulates the effect of policy changes on transportation and residential choice behavior of area residents and consequent changes in area energy consumption. The model is used to test scenarios and policies within the Washington, D.C., metropolitan area as an illustration. The three alternative scenarios the authors set forth are the following: living inside the Washington, D.C., Beltway becomes 25 percent more attractive than it is at present, resulting in an increase of nearly 125,000 additional residents within the Beltway area; residential development within the Beltway becomes 20 percent denser, with the increase coming completely from single family attached and multifamily dwelling units; and two simulations of changes in the road capacity of the transportation network inside the Beltway, the first becomes an increase of 25 percent and the second a decline of 25 percent. The three policy changes they examine are a Live Near Your Work program that provides a subsidy to first-time home buyers who purchase a house near their work, an inclusionary zoning program that permits developers to build at a

Introduction

13

density in excess of that allowed by existing zoning laws if they provide affordable housing in the development, and a vehicle miles traveled tax of $0.10 per mile traveled. The simulations show that the first two scenarios induce energy savings and improve resident welfare, while the third—changes in road capacity—does not. They also find that all three policies result in reductions in area energy consumption. In all the scenarios and with all the policies, changes in vehicle energy use are more important than changes in residential energy use in bringing down energy consumption. However, the authors also observe that with the exception of the imposition of the vehicle miles traveled tax the potential for energy savings is low (less than 1 percent). Because of the low level of savings for the two other policies and the scale of the intervention that the imposition of an energy tax would imply, as well as the fact that the benefits of reduced energy consumption from any of these policies would be largely external to the metropolitan area, the authors are skeptical that any of the three policies modeled are practical means of reducing energy consumption at the metropolitan level.1 The authors also discuss a variety of policies that could reduce energy consumption in metropolitan areas but that they were not able to model, including congestion pricing, an increase in Corporate Average Fuel Economy (CAFE) standards, elimination or reduction of the mortgage interest tax deduction, increased transit use through development of mass transit, and transit-oriented development. The authors conclude that, rather than attempt to impose metropolitan-level policies to reduce energy consumption, metropolitan policymakers should leave such policies largely to the federal government. Instead, local jurisdictions should focus on reducing external costs, from which they will capture most of the benefit, such as area-wide congestion, air pollution, and protection of public open space.

Cross-Cutting Themes In selecting the topics for this volume and for the conference, we sought to acknowledge the broad scope of urban and regional policy issues, foster interdisciplinary discussion, and focus attention on policy effects and challenges. The topics and the backgrounds of authors and conference participants were diverse. Yet, several issues cut across many of the chapters and pose shared challenges for both researchers and policymakers. 1. Reducing global climate change and securing greater energy independence are national and even global benefits and thus are not susceptible to capture within a metropolitan area.

14

Margery Austin Turner, Howard Wial, and Harold Wolman

Mobility People and households move a lot—within and between the jurisdictions of a metropolitan region and between regions. Mobility may undermine the effects of some policies, or it may mean that the benefits of a policy intervention are ultimately accrued in a different jurisdiction. For example, families typically have to remain in their homes for several years to accrue meaningful benefits from homeownership. If they are forced to move more frequently—either because of employment instability or because they cannot afford to stay—the potential wealth gains are likely to be lost. Similarly, if families move frequently, their children may not experience any potential benefits of local investments in early childhood education or the benefits of living in an economically integrated neighborhood and attending economically diverse schools. In addition, the mobility of students between their precollege residences and the locations of their colleges affects the impact of a college or university on the region in which it is located. Tax and expenditure limitations may change patterns of interjurisdictional mobility of residents within a metropolitan area. The impact of preschool education on a city that provides it depends on whether the children receiving the education remain in that city when they become working adults. And, as “Spatial Development and Energy Consumption” points out, the success of metropolitan-scale policies to reduce energy consumption depends on how those policies induce households to change their commuting and housing choices; such changes occur in large part because of household mobility within a metropolitan area. These issues lead to several important questions: —Should policies try to reduce mobility (that is, promote residential stability) so that families and children have more time to accrue the benefits of placebased policies? —How can jurisdictions be persuaded to invest in policies that generate a substantial portion of their benefits outside of their boundaries? —What does mobility imply about the appropriate level of government for different types of interventions? —Does mobility provide a possible rationale for regional collaboration or cost-sharing in some areas? Or does the difficulty of such collaboration cast doubt on the effectiveness of some policies?

Outcome Measures The research evidence presented by the authors suggests that it is not always clear how success should be measured—partly because the desired goal is not always clear. Explicit attention to this question can help frame (or reframe) policymakers’ thinking about an issue. For example, policies promoting low-income home-

Introduction

15

ownership have typically focused on the attainment of homeownership, while Galster and Santiago suggest that homeownership retention and gains in net wealth constitute more meaningful policy goals. Similarly, Ellen, Schwartz, and Stiefel show that increased economic integration at the neighborhood level does not necessarily translate into increased diversity of schools and that school economic integration may not always lead to improved academic performance, especially for low-income children. Belfield’s chapter shows that the value of expanded preschool enrollment should be evaluated differently depending on whether the goal is improved human capital for individuals or better economic and fiscal outcomes for local jurisdictions. Safirova, Houde, and Harrington evaluate a variety of metropolitan-level policies on the assumption that their goal is to reduce energy consumption, but policymakers may adopt some of these policies because of their impact on other outcomes even if the policies have little effect on energy consumption. Bartik and Erickcek assume throughout their chapter that the goal of regional economic development policies should be to raise the incomes of existing residents. Yet, economic developers and policymakers often have other, sometimes conflicting goals. Bartik and Erickcek’s work opens up the questions: What is the goal of regional economic development? Is it higher incomes for existing residents? More jobs for existing residents? Higher average income or more jobs for the region, even if the current residents do not benefit? A higher tax base? Higher real estate values? The chapter by Brunori, Bell, Cordes, and Yuan likewise raises the questions: What is the goal of tax and expenditure limitations? Is it to reduce taxes while keeping public expenditures constant (by reducing wasteful government spending)? To keep government small, or to align the size of government more closely with voters’ preferences? For all the policies considered in this volume, there may be multiple desired outcomes as well as conflicts about what the right outcomes should be. If there are multiple desired outcomes, what are the potential trade-offs among those outcomes? Can policies be designed to improve the terms of trade-offs? Some of the chapters in this volume hint at ways to ameliorate the trade-offs among policy objectives. Brunori and colleagues recommend ways to structure TELs to ameliorate some of their potentially harmful effects. Safirova, Houde, and Harrington suggest that policies to reduce energy consumption are best left to the federal government, while local governments should make land use policy decisions based on other criteria. Similarly, Belfield considers increased federal funding of preschool education to deal with the conflict that can arise between individual economic well-being and local government fiscal well-being as a result of interjurisdictional mobility by preschool recipients.

16

Margery Austin Turner, Howard Wial, and Harold Wolman

Cost-Benefit Calculations In principle, cost-benefit analysis offers a convenient summary measure of a policy’s merits. But there are several different ways to operationalize cost-benefit analysis that are important to distinguish. Total social costs and benefits compared with public sector (taxpayer) costs and benefits. Both are appropriate measures in different contexts, but they are different and need to be clearly distinguished. This issue arises most strongly in the chapters on health care and higher education, preschool education, and tax and expenditure limitations. In these contexts, the interests of taxpayers in a particular political jurisdiction may diverge from those of society as a whole, so the results of cost-benefit analyses may differ depending on whose interests the costbenefit analysis takes into account. Incremental or marginal costs and benefits compared with total costs and benefits. Cost-benefit analyses are often presented for a typical program, for example, a typical preschool program. However, where the program is already in place and what is being contemplated is an expansion of the program’s coverage (for example, an increase in the number of children enrolled in preschool), the costs and benefits of such an expansion of coverage may differ from the total costs and benefits of a typical program. A similar issue arises for low-income homeownership programs. In both cases, the people who are already in the program are presumably the ones who receive the greatest benefit from it and are the ones for whom the costs of supplying the program are lowest. The program is likely to provide smaller incremental benefit and to require larger incremental cost when it is expanded to cover others. Therefore, the fact that a program’s total costs exceed its total benefits may not tell policymakers much about whether the costs of expanding the program’s coverage exceed the benefits. Identifying the appropriate counterfactual. Cost-benefit analysis implies comparison, and identifying the appropriate counterfactual is critical to evaluating results. With respect to the interventions addressed in this volume, is it no action (for example, no organized activity for four-year-olds or no intentional public sector economic development policy) or some alternative action (for example, child care without educational content, low-quality preschool, or home schooling, or recruitment of industries other than health care or higher education, public support for local small business, or efforts to improve the quality of primary and secondary education)?

Differential Effects The effects of the policies discussed in all the chapters in this volume vary in important ways across different population subgroups, different parts of a met-

Introduction

17

ropolitan area, different types and sizes of regions, and different market conditions. But these effects are rarely addressed in the existing literature, leading to vastly oversimplified conclusions about what works (or does not work) and what steps policymakers can and should take to achieve desired outcomes given local circumstances. For example, homeownership may enable a low-income family to build substantial wealth if the house is located in a market that is experiencing significant price appreciation. But in regions with soft housing markets or in undesirable neighborhoods or in both (where house values may be growing slowly or declining), homeownership may not be a wealth-building tool. Moreover, because segregation and discrimination in urban housing markets discourage minorities from buying homes in predominantly White neighborhoods (where values are most likely to be rising) and encourage them to buy in neighborhoods where demand is weaker, they may be less likely to benefit from homeownership initiatives. Similarly, as Ellen, Schwartz, and Stiefel discuss in their chapter on neighborhood economic integration and educational outcomes, in some market settings efforts to achieve greater neighborhood economic integration may accelerate the flight or withdrawal of middle- and upper-income children from public schools, while in others such efforts might attract childless households to a previously low-income neighborhood, diversifying the neighborhood but not the neighborhood schools. As Brunori, Bell, Cordes, and Yuan observe in their chapter on tax and expenditure limitations, research is mixed on the distributional effects of TELs on general purpose municipal governments. California’s Proposition 13 may have benefited low-income homeowners, but there is evidence that TELs imposed on school districts result in larger reductions in student performance for economically disadvantaged communities. The authors reason that the educational effect might be more severe in cities than in suburbs. They explicitly call for more research on the distributional effects of TELs both by income class and by type of place. Whether the expansion of preschool education in a central city increases the educational attainment of the city’s residents depends in part on whether or not the city can attract highly educated residents from elsewhere. Cities such as San Francisco, which have been successful in doing so, may find that expanded preschool education does little to increase the overall educational attainment of their residents, while cities that have been less able to attract the college educated may reap greater human capital benefits from expanded preschool investment. The impacts of health care and higher education on the incomes of a metropolitan area’s residents may depend on the kinds of services provided by the health care and higher education industries in a particular region. Major teach-

18

Margery Austin Turner, Howard Wial, and Harold Wolman

ing hospitals are more likely to attract patients from outside the region than are other hospitals. Similarly, some colleges and universities attract national student bodies, while others serve mostly students from a particular metropolitan area. Although the data on which Bartik and Erickcek rely in their chapter on eds and meds are not fine grained enough to measure these differences, the extent to which a metropolitan area’s health care and higher education industries are export industries will influence the way in which those industries affect residents’ incomes. Expanding a university or a hospital that primarily serves local residents will probably have a smaller impact on income than expanding one that serves many students or patients from elsewhere. Ongoing research on urban and regional policy should more thoroughly investigate the extent to which outcomes differ across markets, neighborhoods, and demographic subgroups. It is a truism that one size does not fit all. But without more empirical evidence about the differential effects of promising policy approaches, local and regional policymakers are at a loss to know whether these approaches make sense for their communities and to know how to ensure that the outcomes are equitable, particularly for racial and ethnic minorities.

Policy Design and Implementation By policy design we refer to the features of a public policy beyond its broad objectives, including the tools used, eligible recipients, funding instruments, and so on. The literature generally emphasizes the effects of stylized or illustrative policies on selected outcomes (for example, expand preschool education, reduce energy consumption), but it pays little attention to the actual design of policies. There may be different ways of carrying out a particular policy (for example, expanding preschool education by making it mandatory and universal or expanding it through programs or subsidies targeted at kids who do not already receive it). This concern leads to some important questions: —What are the effects of different policy designs on policy outputs and outcomes? —Are there alternative ways of combining policies to achieve a desired outcome (for example, a gasoline tax plus various land use policies to reduce energy consumption)? By implementation mechanisms, we mean the processes through which the policy is expected to intervene to bring about change, the theory of change underlying the policy, and the design features of the policy. An elegant policy design will likely be useless if it reflects a theory of change or an intervention that is inappropriate. Most of the empirical evidence focuses on links between

Introduction

19

outputs of policy and outcomes (for example, becoming a homeowner leads to the building of assets for low-income homeowners; attending preschool leads to better student performance), but very little is known about how to produce those outputs or about the mechanisms that undergird the associations between outputs and outcomes. This suggests the need for: —more research that focuses on the relative merits of the various means for producing desired policy outputs (for example, what is the best way of achieving higher levels of low-income homeownership, expanding the health care or higher education industry in a metropolitan area, expanding access to or improving the quality of preschool education, or making a neighborhood more economically integrated?), —opening up the black box that connects the output to desired outcomes, both to determine whether the connection exists and is effective and to understand what the processes are that link the policy output (for example, preschool attendance, residence in an economically integrated neighborhood) to the desired outcome (improved educational attainment or student performance). The cross-cutting themes that we have identified suggest two broad ways in which future research on urban and regional policy should be reoriented to be of greater use to policymakers. First, policymakers would benefit from more research on how urban and regional policy actually works. Such research should take into account alternative normative and factual assumptions about the economic and political environments in which policies are intended to operate and the changes they are intended to produce. Of course, research of this type may be considerably more difficult, costly, and time consuming than statistical analysis of existing data sources. However, the incremental benefit of the latter type of research may be low enough and that of the former high enough that a rebalancing of research priorities could be in order. Second, our cross-cutting themes suggest that empirical research on urban and regional policy should take more explicit account of interregional and interjurisdictional mobility and pay more attention to variations in policy outcomes across different kinds of places. Theoretical work in urban and regional economics, of course, pays a great deal of attention to mobility (though in highly stylized ways), but empirical policy-oriented research typically does so only on some issues, often because of data limitations. Limitations of existing data also make it difficult to examine differential effects of policies across different kinds of places, as do urban and regional theories that posit that a given policy will always produce the same outcomes everywhere. Of course, the role of research on urban and regional policy is to generalize across diverse places; if researchers treated every neighborhood, local jurisdiction, and metropolitan area as unique,

20

Margery Austin Turner, Howard Wial, and Harold Wolman

they would have little to say to policymakers. However, dialogue between researchers and policymakers on the policy-relevant distinctions that could be made between different types of places could benefit both. It is our hope that the clear emergence of these themes across the topics of the following chapters will move the research and policy field forward in ways that promote collaborative solutions to the many challenges of urban and regional policy.

2 “Eds and Meds” and Metropolitan Economic Development timothy j. bartik and george erickcek

T

his chapter analyzes the effects on the economic development of a metropolitan area of investments by state and local governments in higher education and health care services—”eds and meds.” We conclude that eds and meds can affect a metropolitan area’s economic development; we quantify the size of such effects and discuss the mechanisms by which such effects occur. The effects of eds and meds on economic development occur through mechanisms that differ greatly from those of more traditional economic development policies, such as providing manufacturing plants with tax breaks. The analysis presented here is based on a view that state and local economic development policies are an integral part of regional labor market policy. This view is more extensively justified in other publications.1 A more traditional view, held by many practicing state and local economic developers, is that growth is assumed to be good in and of itself. Under this traditional view, whatever maximizes local growth for a given amount of resources devoted to economic development programs is the best economic development policy. We appreciate the assistance of Wei-Jang Huang, Claire Black, Linda Richer, Babette Schmitt, and Julie Kurtz. We also thank Lara Shore-Sheppard, Diane Pfundstein, and David Jaeger for help in matching metropolitan areas from the various censuses. We thank Nancy Pindus, Ned Hill, Hal Wolman, and other participants in the Conference on Urban and Regional Policy and Its Effects for helpful comments. Finally, we thank Howard Wial for extensive and detailed editorial comments. This chapter should not be construed as representing the official views of the W.E. Upjohn Institute for Employment Research. 1. Bartik (1991, 2004a).

21

22

Timothy J. Bartik and George Erickcek

However, more local growth is only good insofar as it provides benefits to particular persons. Empirical studies suggest that by far the greatest benefit of more local growth is the resulting increase in earnings of the original local residents.2 Other possible benefits of local growth, such as increased property values, increased profits of local businesses, fiscal benefits to various governmental units, and labor market gains to in-migrants, are much less important quantitatively, as measured by their present value in dollars. Growth’s effects on the earnings of local residents occur because growth affects the demand for the labor of local residents. This increase in labor demand affects the likelihood of local residents finding work, the occupations they attain, and the wages they are paid in such occupations. More generally, economic development policies may also affect the composition of local labor demand across different types of industries or firms, or across different types of workers, not just the overall number of local workers demanded. Therefore, the analysis of metropolitan economic development policy is viewed in this chapter as the analysis of how the earnings of the original residents of the metropolitan area are affected by any policy that alters the number or types of jobs open to these residents. Such policies may be traditional economic development policies that provide manufacturing branch plants with tax breaks. But, under this definition, metropolitan economic development policies may include other policies that create jobs through public spending or improved local labor quality, increase local productivity and local wages through greater local business knowledge, or increase wages by altering local standards for “typical wages.” All of these other mechanisms by which policies may affect the amount and type of demand for local labor are included as metropolitan economic development policies. These mechanisms will be shown to be important ways in which eds and meds affect a metropolitan area’s economic development. This chapter chooses to focus on the impact of eds and meds on the earnings of the original local residents who stay in the metropolitan area. Thus the chapter takes a metropolitan area perspective rather than a national perspective that would include effects on earnings throughout the United States. How the analysis might vary from a national perspective is briefly considered in the conclusion. The metropolitan perspective is adopted for several reasons: benefits for metropolitan residents alone are easier to analyze, there is more empirical evidence available on such benefits, and the metropolitan perspective is the most relevant perspective for state and local government policymakers. This chapter also chooses to focus on eds and meds defined as industries that provide services to local residents. Therefore, the definition of eds and meds used in this chapter does not include the biotechnology or pharmaceutical 2. Bartik (1991, 2005).

Metropolitan Economic Development 23 industries. This focus is adopted in part because an analysis of the effects of service industries on economic development raises more surprising issues. It is not surprising at all that increasing a metropolitan area’s ability to export biotechnology research products or pharmaceutical products to the nation or the world would increase the metropolitan area’s economic development. What is surprising is that increasing a metropolitan area’s services to local residents might increase the metropolitan area’s economic development. Therefore, the analysis in this chapter contributes to a broader and more flexible analysis of economic development policy. This focus on eds and meds services is also motivated by recent trends in regional economic development and the thinking of local economic developers. With the declining importance of manufacturing as a source of jobs, economic developers have increasingly been looking for other growth levers. As a result, some services have been moved up from being considered as unimportant, secondary, non-export-base activities, which are dependent on other activities to bring new monies into the area, to export-base industry status as potential generators of new monies into the area by serving national and international customers. Education and medical institutions can attract new monies to an area by attracting students, patients, and visitors from outside the region or encouraging students and patients to stay in the region. Eds and meds attract particular attention because of the size of higher education and health care institutions. For example, a 1999 Brookings Institution paper pointed out that in the twenty largest U.S. cities eds and meds typically are among the largest private employers: more than one-third of those employed by the ten largest private employers in these cities work for higher education or health care institutions, and an average of 6 percent of total city jobs are accounted for by these large eds and meds.3 In addition, lower transportation and communication costs have caused many businesses to become more footloose. Many footloose businesses are increasingly selecting locations based on the availability and quality of the local labor supply, because one business location factor that remains difficult to move is the local labor force. This confirms the continuing relevance of Adam Smith’s contention in The Wealth of Nations that “a man is of all sorts of luggage the most difficult to be transported.”4 Furthermore, some of the outstanding regional economic success stories of the last thirty years, such as Silicon Valley in California, Route 128 in Boston, and the Research Triangle in North Carolina, are high technology developments 3. Harkavy and Zuckerman (1999). 4. Smith (1776, book I, chapter 8).

24

Timothy J. Bartik and George Erickcek

that are believed to be in part attributable to the quality of the local labor force and the quality of local universities. Finally, a widespread perception is that a key future growth area will be biotechnology, as the health care sector expands and develops. It might reasonably be hypothesized that biotechnology might be attracted by larger or better quality local higher education and health care service sectors. For all these reasons, therefore, there is good reason for policymakers to be interested in exploring whether investments in higher education and health care institutions may boost a metropolitan area’s economic development. Whatever has been the historical impact of eds and meds on Silicon Valley, Route 128, and the Research Triangle, the relevant question is whether investments in eds and meds will pay off economically today and for most metropolitan areas. Some policymakers are clearly hoping that such investments will pay off. For example, various institutions and civic leaders in Grand Rapids, Michigan, provided tens of millions of dollars of subsidies to convince Michigan State University to agree in January 2007 to move its medical school to Grand Rapids.5 This chapter aims to provide the conceptual framework and quantitative evidence needed for analyzing whether investments in eds and meds, such as the Grand Rapids example, pay off in boosting metropolitan area economic development.

Types of Impacts In this section, we briefly describe several possible types of impacts that eds and meds might have on a metropolitan area’s economic development. In a later section, we provide empirical evidence on the magnitudes of these possible impacts. In this chapter, we focus on the following possible types of impacts of eds and meds on metropolitan economic development: stimulation of local demand by increasing the local export base, the effects of job creation and earnings creation due to eds and meds increasing local human capital, the spillover effects of research and development (R&D) on the local economy, and the effects of eds and meds expansion on local standards of what constitutes a fair wage.

The Demand Stimulus Effects of Expanding the Export Base If any industry, including eds and meds, brings in dollars to purchase local goods, and these dollars would otherwise be spent outside the local economy, then this increase in demand will increase local economic output and thereby increase local earnings. An industry can bring in new dollars by selling its goods or services to persons or businesses located outside the local economy (export5. Miller (2007).

Metropolitan Economic Development 25 base production). An industry also can bring in new dollars by selling its goods or services to local persons or businesses that otherwise would have purchased these goods or services from sources outside the local economy (import substitution). In regional economics, either expanding export-base production or import substitution is frequently referred to as stimulating the local economy’s export base. These new dollars for local industries in turn are respent in part by the local industries on local suppliers and in part by the local industries’ workers on local retailers, resulting in multiplier effects of the initial infusion of new dollars on the local economy. The resulting expansion in local labor demand will enable the original local residents to get jobs, and better jobs, more readily. In turn, this more extensive and higher-quality job experience of the original local residents will make local residents more productive and more employable in the long run, therefore increasing their long-run earnings potential.6 Intuitively, such demand stimulus seems more plausible for eds than it does for meds. If a particular higher education institution closed, many of its students and research dollars quite possibly would go to higher education institutions outside its local economy. For health care institutions, demand for services tends to be more local. If a particular health care institution closed, other health care institutions in the region would arise or expand to meet local demand for health services. Of course, the extent to which a particular higher education or health care institution brings in new dollars will vary quite a bit across different institutions. For example, a community college, compared with a nationally known selective liberal arts college or research university, will tend to serve local residents and businesses. Although some local students of a community college may have otherwise left the metropolitan area, other local residents would have stayed and either not have gone to college or gone to other local educational institutions that would arise or expand to meet this local demand. We must also be careful not to confuse the rapid expansion of an institution or an entire economic sector with whether that sector brings new dollars into the regional economy. For example, even if local hospitals are rapidly expanding and hiring many additional workers, this expansion could simply reflect expanding local demand for hospital services. The net demand effects of a state or local area investing in eds and meds must include the negative economic effects on the local economy of any taxes required. These increased taxes might be needed to subsidize the expansion of eds or meds directly, to provide public services to the eds or meds expansion, or to make up for foregone property taxes if the expansion of these tax-exempt institutions removes property from the tax rolls. In general, such taxes will be 6. Bartik (1991).

26

Timothy J. Bartik and George Erickcek

considerably less than the expenditures associated with these industries, because these industries are substantially funded by fees charged to students or patients. In addition, even if state or local public spending on a good or service is 100 percent financed by taxes, an equal expansion in both taxes and spending will still have a net positive effect on the local economy. Only a portion of the increased taxes reduces consumer spending on local goods, whereas the first impact of spending is to increase demand for local goods.7

Job Creation and Earnings Effects of Eds and Meds on Local Human Capital Increases in the local supply of more productive workers will tend to encourage business development that will use these skilled workers to expand local output and local earnings. Increased local supply of a particular labor skill may increase productivity more if that skill has spillover effects—that is, if an increased supply of workers with that type of labor skill increases the productivity of other types of labor. In addition, long-run productivity effects of an increased local supply of some types of labor skill increase if those skill types allow workers not only to be more productive but also to adapt more rapidly to new technology that increases productivity. Development of greater skills and knowledge in their graduates is still a primary goal of colleges and universities. The greater skills and knowledge of graduates of local colleges and universities will stimulate the local economy to the extent that these graduates increase the average prevailing skills and knowledge in the local economy. A greater availability of local higher education institutions may encourage greater educational attainment by local residents by making higher education more accessible.8 In addition, graduates of a college or university, both those who originally were residents of the metropolitan area and those who moved to the metropolitan area to attend the college, may be more likely to locate in that metropolitan area than they would have been had they attended college elsewhere. Of course, different types of colleges and universities produce different types of skills. Many liberal arts colleges and universities focus more on increasing their graduates’ general skills. Community colleges may focus more on producing occupation-specific or even job-specific skills. For example, many states, most notably North Carolina, give funds to community colleges to provide free customized job training for new workers or incumbent workers in new or expanding manufacturing companies. Both general skills and occupation7. This is the version at the state and local levels of the balanced budget multiplier from any standard macroeconomics course and has been explored by Orszag and Stiglitz (2001). 8. Card (1995).

Metropolitan Economic Development 27 specific skills can increase the productivity of the local workforce, though perhaps different companies will value general skills differently from occupationspecific skills. The productivity of the local labor force is potentially also improved by increases in health, such as those brought about by the health care industry. However, it is unclear whether some marginal expansion of local health care institutions will improve the health of the local labor force sufficiently to improve labor productivity significantly.

R&D Spillover Effects of Eds and Meds on the Local Economy Increased research knowledge of local businesses may raise local output by directly raising local productivity. This increased local productivity may lead to further rises in local output by allowing local businesses to gain a greater share of the national market. Increased research knowledge of local businesses or potential local entrepreneurs also may allow production of new products. These increases in local productivity and wages, in turn, may increase wages and employment opportunities for local residents. Local businesses’ research knowledge may be increased in several ways by various spillover effects of the R&D knowledge and activities of professors at colleges and universities and of doctors at hospitals. The most direct and obvious economic spillover of the research of eds and meds is by some professor or doctor deciding to use his or her research knowledge to start up his or her own business in the local area. In addition, the R&D of researchers at universities and hospitals may be licensed to local businesses. University and hospital researchers may convey a wide variety of research knowledge to local businesses, either through formal consulting contracts or more informally through meetings and casual conversations.

Effects of Eds and Meds on Local Standards of Wage Fairness Empirical evidence suggests that wages persistently vary across industries and firms for workers with the same characteristics.9 Economic theories suggest such wage variations may be due to differences across industries and firms in what is considered to be a fair wage, employee turnover costs, business profitability, and the ease of monitoring worker productivity. Some economic theories imply that prevailing wages in a local labor market may depend on notions of what wage policy of employers (or other labor market practice by employers) is considered fair. There is some empirical evidence to support this notion. For example, studies of local living wage laws, which typi9. See, for example, Krueger and Summers (1988); Dickens and Katz (1987); Groshen (1991).

28

Timothy J. Bartik and George Erickcek

cally regulate the wages paid to government contractors or firms receiving economic development subsidies, suggest far larger effects on local wages and poverty than would be expected by their direct effects on firms whose wages are covered by these laws.10 The labor market practices of local higher education and health care institutions, as large local employers, may influence beliefs in local labor markets about what wages and other employer practices are considered to be fair. If these large employers choose high-road labor market practices, with higher wages, more internal promotion, and lower employee turnover, these practices may influence other local employers to adopt such practices.

Other Possible Impacts of Eds and Meds on Metropolitan Economic Development A longer draft version of this chapter considers several other possible mechanisms by which eds and meds might affect metropolitan economic development.11 These other possible mechanisms include eds and meds’ effects on increasing urban amenities and attracting in-migration and the effects of higher education on increasing entrepreneurship. If higher education and health care investments are focused in cities to a greater extent than in suburbs, this reduction in intra-metropolitan disparities could positively affect the metropolitan area. Finally, higher education and health care institutions may provide better local leadership for economic development. These other possible impacts are omitted in part for reasons of space. In addition, in our judgment we were able to come to more surprising conclusions about the issues focused on in this chapter than for these other issues.

Variation across Metropolitan Areas in Eds and Meds Before discussing the empirical evidence on how a metropolitan area’s specialization in eds or meds affects its economy, we first summarize some patterns in variations across metropolitan areas in specialization in eds and meds. Our definition of eds includes colleges and universities, both four-year and community colleges (NAICS codes 6112 and 6113; SIC codes 8221 and 8222). Our definition of meds includes doctors’ offices and other ambulatory medical facilities, hospitals, and nursing homes and other residential care facilities (NAICS codes 621, 622, and 623; SIC code 80).12 We strive to include both 10. Bartik (2004b); Neumark and Adams (2003). 11. Bartik and Erickcek (2007). 12. See North American Industry Classification System (NAICS) codes at the U.S. Census Bureau website (www.census.gov/epcd/www/naics.html) and Standard Industrial Classification system codes at SEC’s website (www.sec.gov/info/edgar/siccodes.htm).

Metropolitan Economic Development 29 government-owned and privately owned enterprises. (Note that this is not always possible with many establishment-based databases on industry. Most of our empirical data, however, come from census-based data collected from households, in which the industry definitions used in the employment questions typically do include both government-owned and privately owned enterprises.) Our meds definition does not include pharmaceutical companies, biotech companies, or similar enterprises that are primarily concerned with inventing and selling new products and services to be used in health care. Similarly, our eds definition does not include activities in non–higher education industries that might produce the new knowledge that is “sold” in the higher education industry. Our focus is on service industries whose original primary goal is to enhance human capital, either through education or through health care. Of course, some of the possible positive economic effects of these service industries may come from positive effects on other industries; for example, the health care industry might have positive effects on the biotech industry. Our definition of a metropolitan area’s specialization in eds and meds is based on the metropolitan area’s employment location quotient in each of these industries, a standard concept in regional economics. The employment location quotient for eds will be simply equal to the share of higher education employment in the metropolitan area’s total employment divided by the share of higher education employment in total national employment. A similar definition holds for the meds employment location quotient. We use data from the Public Use Microdata Sample (PUMS) of the 2000 census, which are derived from household surveys. From these data, employment in higher education in 2000 was 2.3 percent of national employment, while employment in health care industries was 8.9 percent of national employment. Both eds and meds are large economic sectors. However, meds is about four times as large as eds. As pointed out above, it is quite plausible that meds is much more locally oriented than is the eds industry or, to look at the other side, that a much lower share of meds than of eds is export base, that is, devoted to bringing in dollars to the metropolitan economy that would otherwise go elsewhere. However, because meds is so much larger than eds, even a small share of the meds industry that is export base may be of economic importance to the metropolitan economy. Across the 283 metropolitan areas in our sample, there is far greater variation in the eds location quotient than in the meds location quotient. For example, thirty metropolitan areas have eds location quotients equal to or exceeding 2.49, which means that their share of employment in higher education is at least twoand-a-half times the national average. Thirty metropolitan areas have eds location quotients of less than or equal to 0.44, meaning that their share of employment

30

Timothy J. Bartik and George Erickcek

in higher education is less than half the national average. In contrast, only one metropolitan area has a med location quotient exceeding 2.0, and the thirtiethranked metropolitan area has a med location quotient of 1.31. No metropolitan area has a med location quotient of less than 0.58, and the thirtieth-lowest ranked metropolitan area has a med location quotient of about 0.8, only onefifth lower than the average share of employment in the meds sector.13 The greater variation across metropolitan areas in eds activity compared with that in meds activity strongly suggests that a much larger share of eds activity than of meds activity is export oriented in the typical metropolitan area. Apparently meds activity is largely determined by total metropolitan area economic activity, whereas eds activity varies much more independently of the size of the metropolitan economy. This is not surprising, as the term “college town” is better known and more widely heard than “hospital or medical town.” Correlation analysis shows that there is little relationship between a metropolitan area’s specialization in eds and its specialization in meds. The calculated correlation across the 283 metropolitan areas in the eds and meds location quotients is –0.04. Table 2-1 reports the employment location quotient for eds by metropolitan area for the thirty metropolitan areas with the highest eds location quotients. As one would expect, this eds location quotient is highest in what we generally think of as college towns—large state universities located in relatively modestsized communities. Table 2-2 provides a similar list of the thirty metropolitan areas with the highest meds location quotients. Rochester, Minnesota, stands out as a specialized city in medical care. Further analysis, more extensively reported in the working paper version of this chapter, shows that the larger metropolitan areas tend to have considerably less variation in location quotients for eds, and to a lesser extent for meds. Larger metropolitan areas tend to have strengths in a wider variety of industries, which means they are less likely to have unusually high concentrations of one industry. In addition, the analysis suggests that across the four census regions, the Midwest region appears to have more college towns than do the other regions. Finally, analysis of the PUMS for 1970, 1980, 1990, and 2000 suggests that the variation across metropolitan areas in the eds location quotient has declined in each decade. This may reflect an expansion of higher education to more metropolitan areas, as a higher percentage of the population decides to get a college education. 13. A descriptive table is in Bartik and Erickcek (2007). The standard deviation of the eds location quotient across these 283 metro areas is 1.24, whereas the standard deviation of the meds location quotient is 0.24.

Metropolitan Economic Development 31 Table 2-1. Top Thirty Metropolitan Areas, Ranked by Location Quotient for Higher Educationa Ranking

Metropolitan area

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

State College, PA Champaign–Urbana–Rantoul, IL Bryan–College Station, TX Bloomington, IN Iowa City, IA Gainesville, FL Lafayette–W. Lafayette, IN Columbia, MO Athens, GA Yolo, CA Charlottesville, VA Auburn–Opelika, AL Ann Arbor, MI Muncie, IN Flagstaff, AZ–UT Madison, WI Tuscaloosa, AL Provo–Orem, UT Lansing–E. Lansing, MI Tallahassee, FL Las Cruces, NM Greenville, NC Lincoln, NE Lexington–Fayette, KY Lubbock, TX Fort Collins–Loveland, CO Bloomington–Normal, IL Springfield–Holyoke–Chicopee, MA Hattiesburg, MS Santa Barbara–Santa Maria–Lompoc, CA

Employment location quotient for eds 8.173 7.734 7.160 7.072 6.664 5.936 5.391 5.385 4.934 4.724 4.444 4.439 3.789 3.615 3.568 3.559 3.472 3.317 3.280 3.043 2.973 2.941 2.865 2.837 2.801 2.788 2.770 2.689 2.530 2.496

Source: Authors’ calculations using the Public Use Microdata Sample of the 2000 census. a. All 283 metropolitan areas are ranked in Bartik and Erickcek (2007).

Empirical Evidence on Different Types of Impacts This section considers the empirical evidence on the possible different types of impacts of eds and meds on metropolitan economic development, which have been conceptually described above.

32

Timothy J. Bartik and George Erickcek

Table 2-2. Top 30 Metropolitan Areas, Ranked by Location Quotient for Medical Carea Ranking

Metropolitan area

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Rochester, MN Alexandria, LA Iowa City, IA Columbia, MO Waterbury, CT Duluth–Superior, MN–WI Punta Gorda, FL Gainesville, FL Sharon, PA Asheville, NC Redding, CA Shreveport, LA Johnstown, PA New Haven–Meriden, CT New Bedford, MA Worcester, MA Augusta–Aiken, GA–SC Lexington–Fayette, KY Tyler, TX La Crosse, WI Galveston–Texas City, TX Barnstable–Yarmouth, MA Dutchess County, NY Pueblo, CO Greenville, NC Pittsburgh, PA Utica–Rome, NY Scranton–Wilkes-Barre, PA Hartford–Bristol–Middleton–New Britain, CT Vineland–Millville–Bridgeton, NJ

Employment location quotient for meds 3.081 1.905 1.782 1.561 1.507 1.491 1.485 1.469 1.405 1.402 1.399 1.390 1.387 1.384 1.382 1.377 1.364 1.356 1.355 1.344 1.344 1.344 1.344 1.342 1.337 1.335 1.334 1.326 1.319 1.314

Source: Authors’ calculations using the Public Use Microdata Sample of the 2000 census. a. All 283 metropolitan areas are ranked in Bartik and Erickcek (2007).

Eds and Meds’ Effects as Export-Base Industries that Stimulate Demand for Local Goods and Services Eds and meds will stimulate demand for local goods and services in three distinct ways: direct purchases from their regional suppliers, consumption expenditures of their staffs, and consumption expenditures of students and visitors.

Metropolitan Economic Development 33 There is a huge literature on the demand effects of eds and meds. Most of this literature is so-called fugitive literature published in reports or working papers and not in scholarly journals. In the case of demand effects of higher education, an article by Larry Leslie and Sheila Slaughter reviewed sixty reports on demand impacts, and a paper by John Siegfried, Allen Sanderson, and Peter McHenry reviewed 138 studies since 1992.14 The literature on demand-induced economic impacts of hospitals and other health care facilities has not generated as many literature reviews, but there are easily several dozen studies, mostly of the economic impacts of rural hospitals, in addition to numerous local economic studies by the Kentucky Rural Health Works Program at the University of Kentucky.15 In addition, there are some comprehensive national summaries generated by advocacy groups of local economic impacts stemming from demand effects of eds and meds: for example, a study by the National Association of State Universities and Land-Grant Colleges (NASULGC) of the economic impacts of public universities and a study by the Association of American Medical Colleges (AAMC) of the economic impacts of medical schools and teaching hospitals.16 How to estimate demand impacts of eds and meds is conceptually clear, but the details are tricky. Among the issues that must be addressed are the following four concerns: What proportion of the induced increase in eds and meds, in fact, will bring new dollars into the community—that is, what proportion will enhance the export base, including substituting for imports? What are plausible multipliers? How can one avoid double-counting? What are the economic costs of inducing increased local activity in eds and meds?

The Export-Base Percentage The policy-relevant issue in estimating the demand effects of some policy to induce eds or meds investment is the extent to which this investment will induce new dollars to enter the local economy, as opposed to substituting dollars that are already being spent locally. For example, if we add a new local college or university, a key question is to what extent does this add new students to the community? We do not want to count as additional local demand any expenditures associated with students in the new college who would have otherwise attended another local college or those who would not have attended college but would have been living and spending money in the local community. 14. Leslie and Slaughter (1992); Siegfried, Sanderson, and McHenry (2006). 15. Kentucky Rural Health Works Program (2003–05). For an example of a study of the economic impacts of rural hospitals, see McDermott, Cornia, and Parsons (1991). Local economic studies by the Kentucky Rural Works Program can be found at its website (www.ca.uky.edu/ krhw/impact.html [accessed March 2007]). 16. NASULGC (2001); AAMC (2007).

34

Timothy J. Bartik and George Erickcek

As another example, if we add a new local hospital, a relevant issue is the extent to which this adds new patients and health care spending to the community. We would not want to count as additional local demand any patients or health care dollars that otherwise would have gone to some already existing facility in the community. In the case of meds, this may be more difficult to see because of the natural growth of the industry. With growing income and an aging population, the demand for medical services will continue to grow. In such a growth environment, the displacement impact may be difficult to see as the expansion at one hospital will not affect the current demand for services at the other existing hospitals but instead will curtail their growth plans. Many studies ignore this issue by implicitly assuming that 100 percent of the activity at any eds or meds facility in the local community is truly from new dollars brought to the community and would not have occurred in the community if this institution, or this group of institutions, had disappeared. This assumption can only be rationalized as an answer to a quite different and much less relevant policy question: what would happen if we closed down a particular ed or med institution, or even all such institutions in a local community, and made it illegal to open up or expand any ed or med facilities to replace the closed institution or institutions? This is a less relevant policy question because this is not a feasible experiment that we can imagine carrying out in anything resembling today’s society. (In addition, the effect on local demand would not be the main social impact of such insane and infeasible policies as forbidding hospitals or colleges to operate in a particular metropolitan area.) In the real world, even if we somehow imagined that all the colleges and hospitals in a metropolitan area were closed, there would be some alternative institutions that would arise to replace in part the closed institutions.17 In determining the effect of an addition to local capacity in eds and meds, we must not only look at new students or new patients attracted to the metropolitan area but also at whether any local residents who would have otherwise gone to a college or health care facility outside the metropolitan area are induced to stay in the metropolitan area. In the jargon of regional economics, we need to look not only at the effect of an induced capacity expansion in eds and meds on the metropolitan area’s exports of services to nonmetropolitan area residents but also at the extent to which the induced capacity expansion substitutes for 17. This discussion oversimplifies things a bit. Actually, the usual approach has to assume that after closing existing eds or meds institutions, not only would there not be any substitution of increased spending on newly arising eds or meds institutions but there would not be any increased spending on other local goods or services. Throughout this paper, we focus on the substitution of spending for eds (or meds) institutions for one another, because we think that it is the most important demand-side effect to consider, but there are also other types of expenditure substitution.

Metropolitan Economic Development 35 imports by reducing the consumption by metropolitan area residents of services outside the metropolitan area. Intuition suggests that investment in expansion in eds is much more likely to increase the metropolitan area’s exports or substitute for imports than is investment in expanded meds. College students are far more geographically mobile than are health care consumers. This intuition is backed up by the earlier results that show there is far more variation in location quotients for eds than there is for meds. Meds activity is very tightly tied to local demand, whereas eds activity is far more independent. A greater percentage of eds and meds activity would be part of the export base in smaller metropolitan areas. In a smaller metropolitan area, for example, the disappearance of a particular ed or med institution is less likely to result in its replacement by the expansion of another institution. In addition, the extent to which a particular ed or med institution expands exports, or substitutes for imports, will be greater if that institution is offering an ed or med service that is more specialized. If the ed or med service is more specialized, there are less likely to be good local substitutes for that service if this particular institution goes away. The entry of a more specialized ed or med service into the area is not likely to cause any displacement of existing activities since this specialized service was not offered locally before.

The Multiplier A second tricky issue is the likely multiplier effect of any expansion in metropolitan area exports or substitution for imports induced by expansion in eds and meds capacity. By multiplier effect, we mean the ratio of the total increase in local economic activity to the direct increase in economic activity due to expanded exports or reduced imports in eds and meds. This ratio will be greater than 1 because of the induced effects on local suppliers to eds and meds and also because of the effects on local industries that sell goods or services to the employees of eds and meds. We would expect these multiplier effects for eds and meds to be relatively modest compared with the metropolitan-area multipliers that we see for many manufacturing industries. Eds, and to a lesser extent meds, do not depend as much as do many manufacturers on specialized local suppliers who must be located close by for manufacturers to be able to communicate new technologies. For example, there is no equivalent in the eds and meds industries to the networks of various tiers of nearby suppliers that characterize the auto industry. In addition, as we will review below, eds (and to a lesser extent meds) is not a particularly high-paying industry, which reduces the local demand effects that are due to local spending by employees.

36

Timothy J. Bartik and George Erickcek

Considering weaker local supplier links and the modest wages of eds and meds, it would be surprising if multiplier effects for eds and meds exceeded 2. Most studies do seem to get multipliers of less than 2. For example, Siegfried, Sanderson, and McHenry found that median multipliers for expenditure in college impact studies were 1.7, and for employment, 1.8. Leslie and Slaughter found mean multipliers of 1.6 for two-year colleges and 1.8 for four-year colleges.18 However, some studies seem to get multipliers considerably higher than 2, which is questionable.19 Multipliers for eds and meds will generally be higher in larger metropolitan areas. A larger metropolitan area will have more specialized local suppliers and will have a wider variety of goods and services to capture the spending of the employees of eds and meds.

Double-Counting One subtle problem that occurs in some studies, as pointed out by Siegfried, Sanderson, and McHenry, is double-counting certain expenditures.20 For example, it would be incorrect to count a college’s total spending and student spending on tuition or to count a college’s total spending and its employees’ spending.

Opportunity Cost of Inducing Eds and Meds Expansion A difficult issue is the economic impact of the costs of inducing an eds or meds expansion. In general, expansion of publicly owned colleges and universities would be heavily influenced by state government policy, as state government supplies such a significant share of the revenue of public colleges and universities. The state and local governments’ share of public college and university revenues is about 40 percent, of which about 36 percent is state and 4 percent local.21 However, local political lobbying for an expansion of a publicly owned higher education facility may require local “spending” of considerable political capital and giving up other projects or local benefits. At the extreme, one might imagine that the local area essentially has to incur political costs equivalent to paying for the entire state and local shares of the increased revenue needed for expanding the eds facility. For public hospitals, state and local subsidies amount to about 15 percent of revenue (this excludes Medicare and Medicaid, which amount to another 19 percent and 37 percent of revenue, respectively).22 Of the $103 billion in annual 18. Siegfried, Sanderson, and McHenry (2006); Leslie and Slaughter (1992). 19. Siegfried, Sanderson, and McHenry (2006, p. 18) cite employment multipliers as high as 4.75. 20. Siegfried, Sanderson, and McHenry (2006). 21. See IES (2006, table 329). 22. National Association of Public Hospitals and Health Systems (2004).

Metropolitan Economic Development 37 state and local government spending on hospitals, 59 percent comes from local governments.23 For private colleges or universities and private hospitals or other medical care facilities, the location and expansion decisions are made by the private entity that controls that institution. It is unclear what the typical costs would be for inducing such an entity to expand its local capacity. If inducing the expansion of eds or meds has local costs, which will require higher local taxes, the costs of such taxes should be considered in the analysis. Other things being equal, higher local taxes will lower local demand and local economic activity. In general, this local economic effect will be less than the same-sized local spending, as only a portion of any increase in local taxes will come at the expense of less spending on local goods. This is because local residents would have spent a considerable amount on out-of-metropolitan-area purchases from Internet sales, tourism, and travel-related sales, or they would have saved some of these funds. Any tax costs of financing an expansion of eds or meds are likely to be only a small fraction of the increased expenditures associated with the eds or meds expansion. As mentioned, total state and local spending on public higher education is only about 40 percent of such an institution’s revenues, and state and local spending on health care is only 13 percent of total health care spending.24

Some Illustrative Calculations We provide reference calculations of some plausible impacts of eds and meds using the W.E. Upjohn Institute for Employment Research’s version of the REMI model for the Grand Rapids–Wyoming and Kalamazoo–Portage metropolitan areas in Michigan.25 For each metropolitan area, we consider the economic effects of inducing an expansion of a higher education institution or a hospital that is associated with a $100 million increase in expenditures of the institution. Such calculations allow us to consider the issues outlined above in a consistent way for both eds and meds. As shown in table 2-3, the REMI model has default estimates for what proportion of such expansions will come from substituting for other local economic activity and what proportion will come from either expanding local exports or substituting for local imports. As shown in the table, this export-base percentage 23. U.S. Census Bureau (2007). 24. See NCHS (2006, table 120). 25. Regional Economic Models Inc. (REMI) constructed the model for the Grand Rapids and Kalamazoo metro areas at the request of the Upjohn Institute. REMI models combine the standard regional input-output model with a general equilibrium model having a forecast component. REMI is well regarded by regional economists and has been used in hundreds of studies.

38

Timothy J. Bartik and George Erickcek

Table 2-3. Estimated Local Economic Development Impacts of Eds and Meds Expansion in Two Typical Metropolitan Areasa Grand Rapids

Kalamazoo

Higher Health Higher Health education care education care

Category of effect

Source of calculation

Export-base percentage

From REMI model

75

16

87

31

Multiplier

From REMI model

1.33

1.69

1.16

1.47

Gross economic impact (in millions of dollars)

= 100 million expenditure 100 expansion of eds and meds ⫻ export-base percentage ⫻ multiplier

27

101

45

Gross effects on local resident earnings (in millions of dollars)

= gross impact ⫻ 70 percent of labor share ⫻ 0.4 local earnings effect

28

8

28

13

Rescaled effects: gross effects as a percentage of local earnings for initially induced change of 0.5 location quotient (= 50 percent of average metropolitan area)

= 0.5 ⫻ national earnings share of industry ⫻ ratio of gross impact to initial expenditure ⫻ 0.4 local earnings share

0.44

0.54

0.45

0.89

Source: Authors’ calculations using the W.E. Upjohn Institute for Employment Research’s version of the REMI model a. These figures are largely derived from simulating the effects of inducing a $100 million expansion in expenditure on higher education or health care institutions in Grand Rapids or Kalamazoo, Michigan. All effects are in millions of dollars except for the export-base percentage, the multiplier effects, and the rescaled effects. The multiplier effects are unit free and represent the ratio of the impact on total local economic activity to the increase in export-base expenditure. The rescaled effects are percent effects on local earnings due to an attempt to induce a 0.5 location quotient expansion in eds or meds in each metropolitan area. These rescaled numbers are derived from these simulations by rescaling the size of the expansion and rescaling the impact to a percentage of local earnings. In making these estimates, the national earning shares of eds and meds are 2.20 percent and 9.87 percent, respectively, calculated for the PUMS database of the 2000 census.

is much greater for higher education than for health care. In addition, the export-base percentage is higher for Kalamazoo than for Grand Rapids, which reflects the larger size of Grand Rapids and the greater substitution response that comes from an expansion in ed and med capacity in Grand Rapids. The multiplier estimates from the REMI model, as expected, are modest in size. Multipliers are lower for higher ed than for health care. The lower multipli-

Metropolitan Economic Development 39 ers for eds may be due to fewer local suppliers and to lower wages in higher education than in health care. As expected, multipliers are also lower in the smaller metropolitan area of Kalamazoo. The export-base percentages and multipliers yield an estimate that the gross local economic impact of higher education institutions is about equal to the total expenditure in the higher education institution. This finding is consistent with Melanie Blackwell, Steven Cobb, and David Weinberg’s statement that in “most of the university impact studies [that] we reviewed . . . a university’s annual impact approximately equals its annual budget.”26 This finding occurs because the export-base percentage for the typical university is moderately less than 100 percent and the multiplier is moderately higher than 1, so the resulting impact is close to the university’s budget. In contrast, the economic impact of meds is only one-quarter to one-half of the institution’s budget. This occurs largely because of a much lower export base–import substitution role for the typical health care facility, which is only partly offset by a larger multiplier. As mentioned before, most of the ultimate benefits of local economic development are due to increased earnings of the original local residents. These increased local earnings will be some lesser percentage of the total increase in earnings of the metropolitan area as a result of expanded capacity in eds and meds. We assume that about 70 percent of increased local economic activity goes to labor. Using findings from previous regional studies, we assume also that increased earnings of the original local residents, because of a demand shock to local earnings, are about 40 percent of the total increase in metropolitan area earnings.27 The result is an increase in local earnings of about one-quarter of the institution’s expenditure for eds and about one-tenth of the institution’s expenditure for meds. We might want to standardize such impacts not in terms of a given dollar increase in induced institutional capacity, but rather in terms of the size of that sector. This standardization would allow us, for example, to consider such issues as the relative impact of inducing an x percent expansion in eds compared with an x percent expansion in meds. We consider, therefore, the effect on local residents’ earnings of a policy that attempts to induce a 50 percent increase in ed or med capacity in the average metropolitan area. By definition, a 50 percent increase in ed or med capacity in the average metropolitan area is an increase of 0.5 in the location quotient. Given the national averages presented previously, this policy would be an attempt to increase eds employment by 1.16 percent of total local employment (that is, 50 percent of the share of average eds employ26. Blackwell, Cobb, and Weinberg (2002, p. 94). 27. Bartik (1991, 2001).

40

Timothy J. Bartik and George Erickcek

ment of total employment of 2.32 percent) and increase meds employment by 4.45 percent of total local employment (which is 50 percent of the share of average meds employment of total employment of 8.90 percent). It should be emphasized that we allow for substitution effects that occur because this induced 50 percent expansion in eds or meds will not be all export base but will instead displace existing activity in other eds or meds facilities.28 In particular with meds, an attempt to increase meds activity by 50 percent (or 4.45 percent of total employment) ends up displacing so much existing meds activity that the net increase in meds activity is only one-sixth to one-third of the initial increased activity, or only an 8 to 16 percent increase in meds. For eds, because of displacement, the net increase in eds activity is three-quarters to seven-eighths of the initial increase in eds, or a 38 to 44 percent net expansion in eds. As shown in table 2-3, because meds is a much larger industry than eds, we end up getting similar-sized increases in local earnings from a 50 percent expansion in each of these industrial sectors. For both industries, an attempt to change eds or meds capacity by 50 percent would affect local earnings by about one-half of 1 percent (with a little higher figure for meds in Kalamazoo of about 1 percent). Even though the meds sector is far less an export-base sector than is the eds sector, its much larger size means that its demand-side impact on local residents’ earnings is equal to, or even somewhat greater than, the impact of the eds sector. Although this earnings impact may seem modest, we should recall that a 50 percent increase in eds capacity is only 1 percent of total local employment and a 50 percent increase in meds capacity is only 4 percent of total employment. We also did some speculative calculations in which we adjust for some hypothetical local tax costs of inducing these eds and meds expansions. These assumed costs are based on the extreme assumption that the costs of inducing an expansion will be equal to the state and local share of revenues for the industry, which is 40 percent for public higher education and 13 percent for the overall health care sector. We then run these costs into the REMI model to see the economic impact of these increased local taxes and the extent to which these tax costs might offset the gross impact of the increased capacity in eds and meds. We find that these tax costs, at most, might offset one-third to one-half of the previously calculated impacts, which assumed zero local tax costs. (The detailed figures are not presented in table 2-3 but are available in the working paper version of this chapter.29) The economic impact is not fully offset because of the local costs for two reasons: it does not cost $1 to induce a $1 increase in local 28. It may also displace spending on other local goods and services. 29. See Bartik and Erickcek (2007).

Metropolitan Economic Development 41 capacity, and in general, a $1 increase in local taxes, even with multiplier effects of those taxes, reduces local demand by less than $1.

Eds and Meds’ Effects on Metropolitan Economic Development by Increasing Local Human Capital An increase in local eds and meds capacity likely will increase the quality of local human capital, which will affect local economic development and local earnings. We focus here on empirical evidence that bears on how changes in eds capacity affect human capital and subsequently local economic development and earnings. Empirical evidence of such effects of increases in meds capacity is lacking. The logical chain of causation between an expansion in eds capacity and an increase in local college graduates is as follows. Some induced increase in eds capacity will result in a somewhat lesser increase in the size of the eds sector, after allowing for some substitution for otherwise existing eds capacity. The net increase in eds capacity will increase local production of college graduates. Only some proportion of this increase in the local production of college graduates will result in a net increase in the proportion of college grads in the local labor force, for two reasons. First, some of the locally produced college grads will move out, even if nothing has changed in local labor market conditions. Second, with more local college graduates, wages and employment rates for college graduates will be depressed. While this will attract some employers seeking college graduate workers, it also will encourage out-migration and discourage in-migration of college graduates. The ultimate increase in the local college grads percentage—the percentage of college graduates that make up the local labor force—has effects on the earnings of the original local residents in two ways. First, some of the increase in local college graduates may reflect local residents who were induced by the ed capacity expansion to get a college degree and ended up getting higher earnings and staying in the local area. This reflects a private return to more educational attainment of local residents. Second, there is significant evidence that there are spillover effects on the local labor market of educational attainment. The productivity of an individual worker will not only depend on that individual’s educational attainment but also on the average educational attainment in the local labor market. This average educational attainment may allow employers to use more advanced technologies or introduce new technologies more readily. The productivity increases stemming from spillover effects of a greater local college grad percentage will increase local earnings by attracting business growth, and this growth in labor demand will raise local earnings by increasing local employment-topopulation ratios and by increasing local occupational attainment.

42

Timothy J. Bartik and George Erickcek

From a local economic development perspective, we do not want to count the gains from this area’s expansion of eds capacity that will accrue outside the area’s original residents who remain in the area. For example, we do not want to count the gain in earnings of original local residents whose educational attainment is increased by the expanded capacity but who end up obtaining higher earnings in some other local economy. Furthermore, we do not count the extra local earnings of outsiders who come in for a college degree and stay, as these outsiders otherwise probably would have been just as well-off in some other area. However, we do count the education spillover effects of these outsiders on the original local residents. Because persons are induced to come to the area and get a college degree and stay, the productivity of the local economy is enhanced, which will help stimulate business growth and thereby provide spillover benefits to the original residents.30 All of these effects take some time to occur. It takes a while for an increase in local educational capacity to have its full effects upon the college grad percentage in the local labor force. It also takes a while for an increase in the local college grad percentage to have its full effects in attracting additional business activity and increasing local earnings. We will try in a rough way to assign numbers to the long-run effects on local earnings due to an ed capacity expansion and then scale back these effects for shorter-run analyses. We scale these effects for an initially induced increase in eds capacity of 50 percent for an average metropolitan area, or an increase of 0.5 in the location quotient for eds. To get long-run numbers, we make various assumptions that are based on the empirical literature. A technical appendix details the resulting calculations. These calculations suggest that an increase of 50 percent in the eds capacity in the average metropolitan area will in the long run increase the share of college graduates in the area by 1.63 percent (for example, from 29 to 30.63 percent of the population). Of that 1.63 percent, a conservative estimate is that one-quarter is due to local residents being more likely to attend and complete college. In the long run, the increase in earnings of the local residents who complete more education as a result of the extra local eds capacity is estimated to be equal to 0.27 percent of overall local earnings of all local residents. The spillover effects on the productivity of the local economy due to the higher college grads percentage are estimated as an increase in total local earnings by 0.28 percent. Therefore, the overall effect of a 50 percent increase in local eds capacity is to increase overall local earnings by 0.55 percent. However, this is a long-run effect that only occurs 30. Of course, from a national perspective, we would want to consider all benefits and costs, regardless of the location of who receives them. The national perspective will be briefly considered later in this chapter.

Metropolitan Economic Development 43 after about forty years, time enough for one generation to attend and complete their education in the local eds sector. If these long-run effects take place evenly over time, only one-fourth of this 0.55 percent effect would have occurred after 10 years, or an increase in local earnings after ten years of 0.14 percent. Would these same effects occur regardless of whether the increase in local eds capacity was in a university, a four-year college, or a community college? We simply do not know. Existing estimates of the spillover effects of the college grad percentage on local wages focus solely on completion of a four-year degree, and we do not know whether associates’ degrees from community colleges have the same spillover effects. Nevertheless, it could be argued that local community college capacity has more effects on the educational attainment of local residents than does local university or four-year college capacity, as the students considering community college attendance may be less mobile. But empirical evidence on the differential effects of different types of local eds capacity is lacking. It could be argued that human capital effects might also be produced by improvement in local health care quality, which might occur because of expansions in the local health care industry. In theory, better local health care quality should reduce absenteeism and worker turnover and improve workers’ mental health and therefore productivity at work. However, there is insufficient evidence of what the magnitude of these effects might be. Will improvements in local meds capacity actually improve health care quality and hence the health of local workers? If so, what is the magnitude of these effects? What is the magnitude of the link between local health care quality and local productivity? We do not know the answers to these questions.

R&D Spillovers The research activities of higher education institutions, and perhaps some research-oriented hospitals, may have significant spillover effects on local economic development. It has been argued that it is this role of universities that has “the greatest potential to affect economic development.”31 The policy community’s interest in this role of higher education institutions is probably largely inspired by the successes of Silicon Valley, Route 128 around Boston, and the Research Triangle in North Carolina; these successes are usually believed to be in part attributable to the research strengths of nearby universities. What is the research evidence on the magnitude of these effects of universities on economic development due to research spillovers? To summarize at the outset, the case study evidence, from case studies of particular universities in particular local economies, suggests that the impact of higher education research 31. Paytas, Gradeck, and Andrews (2004, p. 4).

44

Timothy J. Bartik and George Erickcek

activities on local economic development is not a mechanical function of the size of the research or the size of the university, but rather it depends upon many idiosyncratic features of the university and the local economy. The impacts on local economic development of higher education research activities do not occur solely or in many cases primarily because of technology transfer to new local business start-ups but rather occur because of a wide variety of ways in which the research knowledge and expertise at the university can help local businesses address productivity problems and other business problems. Empirical estimations of effects of university research activities on local economic development tend to be fragile—that is, results vary widely depending upon what measures of university research activities and local economic development are used and what metropolitan areas and time periods are considered. The fragility of results may reflect the difficulty of measuring what activities of universities matter most to different types of local economic development and how these effects vary according to the local economy as well as our inability in most cases to find natural experiments in which government policy or some other exogenous change has caused large changes in university research activities. Some major recent reports that have included multiple case studies of university research–local economic development interaction are a report by the Local Innovation Systems Project, based at the MIT Industrial Performance Center, and a recent report for the Economic Development Administration done by the Center for Economic Development at Carnegie Mellon University.32 As these case studies point out, the typical U.S. university cannot be expected to have the same economic influence on the local economy as would an MIT or a Stanford. For example, the University of Akron sought to focus on helping the local economy transition from tire production to innovation in polymer production. But, according to the case study evidence, Akron-area polymer firms “didn’t see much of value emerging from the university’s laboratories, and some had already developed sophisticated strategies for interacting with universities nationally.”33 In fact, the evidence suggests that the overall influence of a university on local economic development through technology transfer is quite limited. As pointed out by Richard Lester, “new business formation around university science and technology is a very small fraction—probably no more than 2–3 percent—of the total rate of new business starts in the U.S.” Furthermore, patenting by universities in the United States “is only a minor contributor to the overall stock of patented knowledge. About 3,700 patents were granted to U.S. 32. The Local Innovation Systems Project case studies can be found in Lester (2005), while the report by the Center for Economic Development can be found in Paytas, Gradeck, and Andrews (2004). 33. Lester (2005, p. 19), based on research by Safford (2004).

Metropolitan Economic Development 45 universities in 2001, out of a total of about 150,000 U.S. patents issued in that year.”34 A review paper by Maryann Feldman and Pierre Desrochers made the judgment that “since the 1980s, despite the establishment of university technology transfer offices, incentives from federal and state governments, and new industrial outreach efforts, most research universities have not been particularly successful at technology transfer and have not yet generated significant local economic development.”35 However, for the typical university, university research activities can have a much broader impact on local economic development than is captured by just looking at technology transfer through new business start-ups. This broader impact is through a wide variety of formal and informal interactions in which professors, researchers, and students at the university interact with nearby businesses, through either formal contracts or more informal interaction to help local businesses solve a wide variety of problems. Jerry Paytas, Robert Gradeck, and Lena Andrews concluded from their various case studies that “the structure of the [university’s] technology transfer office does not determine a university’s performance in generating economic impact.”36 According to them, one factor that really distinguishes a university that is effective in local economic development is the “breadth of involvement” of the university: “Universities need to address business and legal issues, workforce education, infrastructure, and industry relationships, as well as technology and R&D capacity, in order to yield regional benefits. The most engaged universities demonstrate these kinds of diverse, integrated commitments across administrative and academic units, including the schools of business, engineering, law, medicine, and public policy.”37 They also argue that impacts depend on the alignment of the university’s research activities with the characteristics of local industries. Given the diversity of interactions between university research activities and local economic development, it is perhaps not surprising that quantitative research on these interactions has yielded diverse results. For example, Luc Anselin, Attila Varga, and Zoltan Acs found that more university research in a metropolitan area is positively associated with private research in the metropolitan area and with the number of business innovations in the metropolitan area.38 Varga later went on to find that the impact of university research on business 34. Lester (2005, p. 10). 35. Feldman and Desrochers (2003, p.5). 36. Paytas, Gradeck, and Andrews (2004, p. 7). 37. Paytas, Gradeck, and Andrews (2004, p. 9). 38. Anselin, Varga, and Acs (1997).

46

Timothy J. Bartik and George Erickcek

innovations is greater in metropolitan areas that are larger and have more existing high technology activity.39 Edward Hill and Iryna Lendel found that higherrated science and technology doctoral research programs at a metropolitan area’s universities are associated with significantly higher metropolitan employment and per capita income growth. They reported that this result is fragile when a control for metropolitan area size is included.40 Neil Bania, Randall Eberts, and Michael Fogarty found that university research in a metropolitan area affected business start-ups in the metropolitan area positively in only one industry— electrical and electronic equipment—out of the six industries they studied. They also pointed out that even if university research leads to innovation, “any resulting new products or processes will frequently be developed in other locations.”41 An important recent empirical paper on the connection between university research and local economic development is a working paper by Roland Andersson, John Quigley, and Mats Wilhelmsson.42 The main strength of this paper is that it focuses on the results of an explicit policy decision by the Swedish government to establish new universities with a strong research component in a number of regions from the late 1970s on. Their data are pooled cross-section data on output per worker and number of researchers in different Swedish communities and years. Controlling for fixed community effects and year effects, they found that having more university researchers is associated with higher output per worker. That is, a community’s relative productivity, compared with its past relative productivity, is positively influenced by its relative share of university researchers compared with its past share. These effects are particularly strong for the newer universities set up by the Swedish government. Because of these stronger effects for the newer universities, their simulations found that the decentralization of university research in Sweden raised the average productivity of the Swedish economy, compared with what would have happened if the research had been kept in the old universities. This finding is in obvious contrast with the empirical estimates of Varga.43 Andersson, Quigley, and Wilhelmsson’s results support decentralization of higher education research activities as the best way to promote overall national economic development, whereas Varga’s results support more spatial concentration of higher education research activities as the best way to promote overall national economic development. On the basis of these case studies and quantitative research studies, it seems likely that spillover effects of university research activities on local economic 39. Varga (2000). 40. Hill and Lendel (2007). 41. Bania, Eberts, and Fogarty (1993, p. 765). 42. Andersson, Quigley, and Wilhelmsson (2005). 43. Varga (2000).

Metropolitan Economic Development 47 development are important. However, it seems impossible, and also misleading, to come up with a summary estimate of how much a given expansion of a local university will affect local employment growth and hence local earnings. The research findings are too diverse to come up with a believable summary, and the literature suggests that the impact will depend upon many features of the local university and local economy. Finally, there does not appear to be much evidence on the potential role of teaching hospitals or other health care institutions in creating research spillover effects.

Effects of Eds and Meds on Local Standards of Fair Wages and Fair Labor Market Practices Increases in eds or meds capacity may also affect local labor market earnings because of these industries’ wage practices and other labor market practices (internal career ladders, training opportunities, and so forth). Different industries persistently tend to pay different wages for workers, holding worker characteristics constant. These wage differentials may reflect a wide variety of industry characteristics, including unionization, wage norms, the costs of worker turnover, and the ease of monitoring worker productivity. Obviously, a redistribution of a metropolitan area’s workers toward industries with positive wage differentials, and away from industries with negative wage differentials, will directly increase wages for the workers who move to higherpaying industries. In addition, however, there is evidence that a metropolitan area that moves toward higher-paying industries will find its overall earnings per capita increasing by more than would be predicted, given national industry wage differentials.44 If a metropolitan area’s industry mix changes so that, within the context of national industry wage patterns, we would expect wages to be higher by 1 percent, average real earnings per working-aged adult would be estimated to increase by 2.2 percent. These overall wage effects of the mix of different-paying industries in the metropolitan area appear to reflect in part an influence of average metropolitan wage practices on the wages paid in each industry. In addition, higher metropolitan area wages will tend to increase labor force participation and will also increase local demand and job creation. To our knowledge, no prior analysis of industry wage differentials, while controlling for worker characteristics, has included all workers, both public and private, in these two industries. We also wanted to explore how these industry wage differentials differed by education level. Therefore, we decided to do our own 44. Bartik (1993).

48

Timothy J. Bartik and George Erickcek

Table 2-4. Effects of Eds or Meds Industry on Log Wages, Controlling for Worker Characteristicsa Eds

Meds

Regression sample

Coefficient

t statistic

All workers High school dropout High school graduate Some college Associates degree College degree Postgraduate degree

–0.145 0.008 –0.038 –0.141 –0.166 –0.200 –0.129

–89.60 0.70 –7.22 –46.35 –28.08 –55.24 –46.21

Coefficient

t statistic

Number of observations

0.048 –0.024 –0.059 –0.014 0.155 0.050 0.132

57.59 –7.20 –33.25 –8.76 84.64 25.36 58.44

4,410,914 583,070 1,147,998 1,124,519 345,489 787,622 422,216

Source: Authors’ calculations using the Public Use Microdata Sample of the 2000 census. a. The observations are on the natural logarithm of wages. Wages are calculated as annual earnings for the person divided by the product of annual weeks worked and usual weekly hours. Observations are excluded if earnings, weeks worked, or usual weekly hours were imputed rather than actually observed. Observations were excluded if calculated wages were outliers; that is, wages were less than $2 or more than $200. The sample is all workers who were sixteen- to sixty-four-years-old in metropolitan areas. The regressions controlled for gender; marital status; marital status interacted with gender; mutually exclusive race categories of Hispanic, White non-Hispanic, Black non-Hispanic, and other non-Hispanic; a quartic in age; 283 metropolitan area dummies, and dummy variables for 254 industries. The industry dummies were all included; the weighted sum of the industry variables, where the weights are the estimated proportion of each regression sample in each industry, was constrained to equal zero. This means that the industry wage coefficient measures the wage paid in that industry compared with the all-industry average wage for that education group. Each row corresponds to a different regression. The education groups are defined so that they are mutually exclusive. For example, “some college” means some years of college attendance without any degree.

wage analysis that would include estimations of how wages differ from other industries for both eds and meds, controlling for worker characteristics. We analyzed wages using a sample from the 2000 PUMS. The wage regressions controlled for gender, marital status, race, age, 283 metropolitan areas, education (in the overall regression), and 254 industries. Table 2-4 reports the results. Overall, eds pays about 14.5 percent less than the average industry for a given set of worker characteristics. Meds pays about 4.8 percent more than the average industry. The separate regressions for different education groups make it clear that these results are not simply being driven by high wages for doctors and low wages for professors, relative to their education. On the other hand, for some of the lowest education groups, the wage differentials are quite different. We can apply these results to estimate the effects on local earnings of a policy that tries to induce an expansion in each industry of 50 percent in an average metropolitan area (or 0.5 location quotient). As previously discussed, because of

Metropolitan Economic Development 49 displacement effects, the proportion of metropolitan area employment in the industries will, on net, expand by less than 50 percent. For eds we assume, as was done previously, that a policy that adds 50 percent to eds activity will lead to a net expansion of 37.5 percent (50 percent adjusted downwards by a 25 percent displacement of existing eds activity). For meds, displacement is higher. We assume, as was done previously, that a policy that adds 50 percent of meds activity will lead to a net expansion of 8 percent to meds activity, after allowing for five-sixths of the initial capacity increase to be offset by displacement of existing meds activity. For eds, the mean percentage of national employment in this industry is 2.32 percent as of 2000. Therefore, an attempted expansion in a metropolitan area of this industry by 50 percent will result in a net increase in the percentage of local employment in this industry of 0.87 percent (37.5 percent of 2.32 percent). We assume that we are switching employment to eds from industries of average pay. Therefore, a switch of employment of 0.87 percent from an average-paying industry to an eds industry that pays 14.5 percent below average will directly lower average wages in the metropolitan area by –0.13 percent (0.87 percent ⫻ –14.5 percent). But, as mentioned above, estimates suggest that a 1 percent change in the average predicted wage in a metropolitan area based on the metropolitan area’s industry mix will change overall metropolitan area earnings per capita by 2.2 percent, because of changes in other industries’ wage practices and changes in employment rates. Therefore, we predict that this investment in attempting to expand local eds capacity by 0.5 location quotient will, through its effect on local efficiency wages, lower average local earnings per capita by –0.29 percent (–0.13 percent ⫻ 2.2). For meds, the mean percentage of national employment in this industry is 8.90 percent as of 2000. Therefore, an attempted expansion in a metropolitan area of this industry by 50 percent will result in a net increase in the percentage of local employment in this industry of 0.71 percent (0.16 ⫻ 50 percent ⫻ 8.90 percent). A switch of employment of 0.71 percent from an average-paying industry to a meds industry that pays 4.8 percent above average will directly increase average wages in the metropolitan area by 0.03 percent (0.71 percent ⫻ 4.8 percent). With a multiplier on local earnings of 2.2, which is due to effects on local wage practices and employment rates, the estimated effect on metropolitan area earnings per capita is an increase of 0.07 percent (0.03 percent ⫻ 2.2). The above analysis simply considers the labor market effects of expanding eds or meds, conditional on current labor market practices in these industries. There also is the policy option of altering employer practices in these industries in terms of pay, worker training, and promotion practices. Sectoral employment programs are one option for the public sector to seek to intervene to alter train-

50

Timothy J. Bartik and George Erickcek

ing and upgrade opportunities in a given industry. Many such programs involve public sector training organizations working with different firms and organizations in a particular industry to set up training programs that are suited to the industry’s needs, while also improving worker advancement options. Sectoral employment programs frequently target health care because it is a major industry that employs many low-income persons, is expanding, and frequently has concerns about worker shortages. An evaluation of one group of initiatives in sectoral programs is provided by Pindus and others.45 While such programs are promising, as of now there is still considerable uncertainty about sectoral programs’ effects on worker outcomes.

Conclusion To sum up, eds and meds have a variety of positive effects on a metropolitan area’s economic development. These effects are large enough to be important for public policy. However, there is considerable uncertainty about the magnitude of many of these effects. The magnitude of effects likely depends a great deal on the nature of the metropolitan area economy and the specific characteristics of the induced expansion of eds or meds. Taking all the previous estimates together, a policy that attempted to expand a metropolitan area’s eds capacity by 50 percent, or about 1.2 percent of total local employment, is estimated to increase average earnings of the original residents of the metropolitan area after ten years by 0.3 percent (0.44 percent that is due to the demand stimulus from enhancing the export base + 0.14 percent that is due to effects on human capital development of the eds expansion – 0.29 percent due to the eds industry’s relatively low wages). To this should be added the positive effects on local earnings of the R&D spillovers that are due to eds expansion. A policy that attempted to expand a metropolitan area’s meds capacity by 50 percent, or about 4.5 percent of local employment, is estimated to increase local earnings by 0.6 percent (0.54 percent that is due to the demand stimulus from enhancing the export base plus 0.07 percent that is the result of the meds industry’s relatively high wages). Therefore, for policies that attempt to expand eds or meds by 50 percent, the effects of the meds expansion on the earnings of local residents are about twice as great. This is largely due to the greater size of the meds sector. If we considered instead eds or meds expansion equal to the same percentage of overall local employment—say an eds or meds expansion of 1 percent of local employment—the effects of eds expansion on local residents’ earnings is about twice the size of the effects of meds expansion. A policy attempting 45. Pindus and others (2004).

Metropolitan Economic Development 51 to expand eds by 1 percent of total local employment in all industries is estimated to increase average earnings of the original residents of the metropolitan area by 0.2 percent, whereas a policy attempting to expand meds by 1 percent of total local employment in all industries is estimated to increase average earnings of the original residents by 0.1 percent. (These figures are derived by rescaling the estimated effects of 50 percent expansions in eds or meds.) These estimated effects of eds or meds expansion on local earnings are an important part of considering the benefits and costs of any public policy that attempts to induce such an expansion. For eds, such policies could include state or local investments in new higher educational institutions, proposals to expand higher educational institutions or encourage expansion by subsidizing tuition, and proposals to either consolidate or decentralize the provision of higher educational services across a state. For meds, such policies could include state and local efforts to subsidize or constrain the expansion of health care institutions, either directly or indirectly through the details of health care programs such as Medicaid. We emphasize again that all these estimates are for effects of eds or meds expansion on a local metropolitan economy. The economic development effects of eds or meds expansion at the national level might be quite different. On the one hand, from a national perspective we would have to consider whether an expansion of eds or meds capacity in one metropolitan area might displace some other eds or meds capacity in other metropolitan areas. This displacement at the national scale would diminish the effects of eds and meds expansion. On the other hand, from the national perspective we would consider the effects of eds or meds expansion on increasing earnings of individuals who migrate outside the metropolitan area. This would enhance the effects of investments in eds and meds. What about research gaps? Three research gaps appear obvious. First, we know less than we would like about the effects of particular types of eds or meds expansion in particular types of metropolitan areas. What, for example, are the relative effects of community college expansion as opposed to those of university expansion, and how do these effects differ in metropolitan areas of different sizes with different industrial specializations? Second, we know much less about the local economic development effects of meds than those of eds. For example, we know relatively little about how a higher-quality local meds sector might enhance the health and productivity of local workers and thereby affect the local economy. Third, our knowledge of the research spillover effects of eds is incomplete. We do not know as much as we would like about the quantitative magnitude of different types of research spillover effects of different activities of higher education institutions. Which activities of higher education institutions pay off the most in terms of their research spillover effects on the local economy, and

52

Timothy J. Bartik and George Erickcek

what are the magnitudes of those payoffs? The general belief is that the research activities of higher education institutions are likely to have large long-run effects on a metropolitan area’s economic development. Although this is useful for lobbying to expand overall university budgets, this belief lacks sufficient, specific quantitative backing to provide definitive advice to policymakers that might guide more selective investments in universities. In addition, we need to know more about whether it makes more sense for a state government to focus research investments on the state’s flagship university or universities, or whether state governments should instead spread such research investments more broadly among many educational institutions.

Technical Appendix This technical appendix provides more details about how we calculate the magnitude of the metropolitan economic development effects of eds that occur due to eds’ effects on local human capital. This appendix focuses on the effects on local earnings in the long run. (In the main text of the chapter, we then scale these back to shorter-run effects). We scale these effects for an initially induced increase in eds capacity of 50 percent in the average metropolitan area, or 0.5 location quotient (LQ). Our long-run numbers are derived from various assumptions based on the empirical literature.

Substitution Effects Per our discussion above, we assume that the net effects on ed capacity are 75 percent of the initially induced effects. (This uses the export-base percentage for higher education in the Grand Rapids MSA from the REMI model.) To put it another way, if some college increased its capacity in an amount equal to onehalf of the entire current local educational sector, we assume that 25 percent of this increase would be offset by reductions in capacity in other local educational institutions. Therefore, the net change in ed capacity from a 0.5 location quotient–induced change is a change of 0.375 LQ.

Long-Run Effects on the Stock of College Grads, Allowing for Their Direct and Induced Migration Bound and others estimated that fifteen years after some increase in a state’s production flow of college graduates, the increase in the stock of college graduates is about 30 percent of the stock increase.1 This presumably reflects some normal out-migration of college graduates, plus the effects on migration patterns, due to changes in state labor market conditions caused by a shock to college grad 1. Bound and others (2004).

Metropolitan Economic Development 53 labor supply. As metropolitan labor markets are, in general, smaller than state labor markets, we assume that the net effects on the local stock of college graduates of an increase in the flow are about 15 percent of the flow. Therefore, the college grad stock will go up by about 5.625 percent (15 percent of 0.375LQ) of whatever stock of college graduates would occur nationally given a location quotient of 1.0. About 29 percent of the current labor force (aged twenty-five to twenty-nine) has a college degree.2 By definition, the typical area in the United States has a location quotient of 1.0. This amount of higher education activity must be sufficient at the national level to result in 29 percent of the labor force having a college degree. Therefore, this implies that an initial induced increase in the local ed location quotient of one-half will lead to an increase in the local percentage of college grads of 1.63 percent (5.625 percent ⫻ 29 percent) of the local labor force.3

Impact of the College Grad Percentage on Long-Run Local Labor Demand Owing to Productivity Effects A number of studies suggest that an increase in the percentage of college graduates in the local labor force will increase local employment growth.4 The empirical estimates from Jesse Shapiro are typical in their magnitude. Shapiro’s estimates imply that an increase in the local percentage of college graduates of 1.63 percent (from above) will increase a metropolitan area’s employment after ten years by 0.72 percent.5 2. See IES (2006, table 8). 3. Groen (2004) has sometimes been cited (for example, by Siegfried, Sanderson, and McHenry 2006) as showing a weak relationship between college location and the eventual choice of college-educated workers of a place to live and work. Groen concluded that at the state level, attending a college in state X increases the proportion of students who eventually choose to work in that state by only 10 percent. However, Groen obtained these estimates by focusing on students who apply to colleges in more than one state, and these students may be more footloose than the students who apply to college in only one state. In addition, Groen’s estimates do not allow for any effect of college expansion in increasing the proportion of the local population that attends college. All of Groen’s estimates are conditional on students deciding to attend college. Furthermore, Groen’s estimates do not allow for any aggregate effects of more students graduating on the attractiveness of an area to college graduates. Finally, Groen’s estimates are contradicted by the aggregate estimates of Bound and others (2004), which seem more directly pertinent to the issue at hand. 4. Shapiro (2006); Glaeser and Saiz (2004); Gottlieb and Fogarty (2003). 5. Shapiro’s (2006) empirical estimate in table 3 is that the coefficient on the ln (college grad share) in an equation explaining ten-year growth is 0.0786. But Shapiro’s estimates are for the 1940–90 period, with lower average educational attainment. Using 1980 as a typical year, the average percentage of college graduates in that year was 17 percent, according to the 2005 edition of Digest of Education Statistics (IES 2006). Thus ln (17 + 1.63) – ln (17) = 0.09156. Multiplied by 0.0786, this gives a ten-year increase in ln (employment) of 0.0072.

54

Timothy J. Bartik and George Erickcek

But the ten-year increase in employment will be an underestimate of the long-run increase in employment. We know that the employment level in regions only gradually adjusts to its long-run equilibrium level in response to changes in regional economic conditions. In his well-known article, Jay Helms estimated that regional employment adjusts annually by 8.9 percent of the difference between current regional employment and the region’s long-run equilibrium employment level. This implies that the long-run increase in employment will be 1.64 times the ten-year increase in employment.6 Therefore, the implied long-run increase in employment, due to a 1.63 percent increase in the local percentage of college graduates, will be 1.18 percent (0.72 percent ⫻ 1.64).7 Shapiro’s estimates suggest that about 60 percent of the effects on employment growth of the local college grad percentage are due to the college grad percentage’s effects on the local labor force’s productivity. (The remaining 40 percent of the growth effects of the college grad percentage are estimated to be due to effects of the local college grad percentage on the metropolitan area’s amenities; we consider these amenity effects further in the working paper version of this chapter.) This implies that the productivity effects of an increase of 1.63 percentage points in the local college grad percentage will only explain 60 percent of the total long-run increase in employment of 1.18 percent. Therefore, an increase of 1.63 percentage points in the local college grad percentage will increase labor demand, resulting from higher labor force productivity, by 0.71 percent (60 percent of 1.18 percent).

Effects of Shocks to Local Labor Demand on Local Earnings The regional economics literature suggests that a 0.71 percent shock to local labor demand will increase local earnings by two-fifths, or 40 percent, of the shock to labor demand, or 0.28 percent (0.4 ⫻ 0.71 percent).8 About half of this increase in local earnings is due to increases in the local employment-topopulation ratio, and the other half is due to local residents moving up to higher-paying occupations.9 6. Helms (1985). Helms’s adjustment estimates imply that the long-run effect on regional business activity is equal to the effect after t years divided by [1 – (0.9104 taken to the Tth power)]. For T = 10, this equation means that long-run employment will increase by 1.64 ⫻ the ten-year effects. 7. This is derived by using a national average college grad percentage of 17 percent for the Shapiro (2006) results in table 3 and then taking ten-year growth and multiplying by 1.64 (on the basis of Helms 1985). 8. Bartik (1991). 9. This is a conservative estimate of the effects on the earnings of the local college grad percentage compared with more direct estimates of the effects of the local college grad percentage on wages. However, we believe this conservative estimate is more reliable. For example, Moretti (2004) estimated that a 1 percentage point increase in the local college grad per-

Metropolitan Economic Development 55

Private Earnings Returns to Local Residents from Increased Local Educational Attainment Card estimated a quite large effect of the availability of local four-year colleges on educational attainment.10 The presence of such a college in the local community when a youth is fourteen is estimated to raise the average number of years of education by at least 0.32 year. An increase of a 0.32 year is roughly equivalent to an increase in the percentage of youths graduating from college of 8 percentage points (0.32 / 4).11 Suppose we assume that the effects of college availability are roughly linear, that is, a change in the college LQ by one unit always has the same effect on college graduation rates in local youth (from 0.0 to 1.0 or from 1.0 to 2.0), and a change in the college location quotient by 0.50 location quotient will have half of that effect on college attainment, and so on. Card’s estimates can be interpreted as switching from a location quotient for local colleges of zero to a location quotient of 1.0. The implied effect of this is to increase the college graduation rate for local youth by 8 percentage points. As described above, we assume that if we induce an initial increase in the ed LQ by 0.5, the resulting net increase in the local ed LQ will be 0.375, due to displacement of existing ed activity. Under these assumptions, this investment centage will increase local wages (in addition to the effects on those educated) by 0.6 to 1.2 percentage points. He argued that this increase is an increase in the real wage. These estimates imply that a 3.26 percentage points increase in local college graduates will increase local wages, and thereby local earnings, by 2 to 4 percent. This estimate is 4 to 8 times the earnings effects we use in the text. However, we regard our smaller estimate as more reliable than Moretti’s, for several reasons. First, most of his estimates do not control for any local prices. The one estimate that does control for local prices only controls for a local rent index. But we would expect overall local prices, other than local rents, to go up, owing to the share of local land and labor costs in the production and distribution costs of many local goods and services. Therefore, it is unclear whether Moretti’s estimates really represent real wage and real earnings effects. Second, the local college grad percentage can affect local wages in a number of ways, not just as a result of the effects on productivity. Local wages can also be affected by the amenity effects of the local college grad percentage and by the effects of higher education institutions on local wage norms. Moretti’s estimates do not allow us to separate these effects out, unlike Shapiro’s estimates that do. 10. Card (1995). 11. For example, if the presence of the local college only has one effect, that of causing 8 percent of local youth to increase their education by four years and graduate from college, 0.08 ⫻ 4 will equal a 0.32 increase in average years of education. Of course, in the real world, the presence of a local college may cause some local youth to increase their education from zero years of college to one year, which will not affect the college graduation rate. But it may also cause some local youth to increase their years of college from three years to four years, which will increase the college graduation rate. If we assume these effects are roughly offsetting, then a 0.32 increase in average years of education is equivalent to an 8 percentage points increase in college graduation.

56

Timothy J. Bartik and George Erickcek

in ed capacity will increase the college graduation rate for local youth by 3 percentage points. However, not all these youth will stick around the metropolitan area. Estimates suggest that in the long run, a little more than 50 percent of college graduates stay in the same state as their state of birth.12 Suppose we assume that this percentage is half as much for the typical metropolitan area, or that about 25 percent of local area residents who get a college education in the area stay in the area. Then the implication is that the local residents who get a college education because of expanded local college options, and then stay in the local area, will make up about 0.75 percent of the local labor force. This would be about half of the 1.63 percent increase in the local percentage of college graduates that we previously calculated. To be conservative in our calculations, we assume that only one-quarter of the 1.63 percent increase in the local college grad percentage is due to local residents who were induced to get a college degree, or 0.41 percent of the total local labor force. Earnings differentials due to a college degree are now more than 60 percent; we use a figure of 66.5 percent, taken from the College Board.13 The percentage increase in overall local earnings due to the increase in college graduation of the original residents who stay in the local area will be 0.27 percent (0.41 percent times ⫻ 66.5 percent). Therefore, the total estimated increase in local earnings due to an inducement of a 50 percent, or a 0.5 LQ, increase in ed capacity is 0.55 percent: 0.28 percent (social spillover effects of education on earnings) + 0.27 percent (effects on local residents induced to increase their college graduation rate). Half is due to extra educational attainment for the area’s youth, and the other half is due to the social spillover effects of having more college grads.

12. Bartik (2006). 13. College Board (2006).

Metropolitan Economic Development 57

References Andersson, Roland, John M. Quigley, and Mats Wilhelmsson. 2005. “Urbanization, Productivity, and Innovation: Evidence from Investment in Higher Education.” Working Paper W05-001. University of California–Berkeley, Institute of Business and Economic Research, Program on Housing and Urban Policy. Anselin, Luc, Attila Varga, and Zoltan Acs. 1997. “Local Geographic Spillovers between University Research and High Technology Innovations.” Journal of Urban Economics 42, no. 3 (November): 422–48 . [AAMC] Association of American Medical Colleges. 2007. The Economic Impact of AAMCMember Medical Schools and Teaching Hospitals. Washington (January). Bania, Neil, Randall W. Eberts, and Michael S. Fogarty. 1993. “Universities and the Startup of New Companies: Can We Generalize from Route 128 and Silicon Valley?” The Review of Economics and Statistics 75, no. 4 (November): 761–66. Bartik, Timothy J. 1991. Who Benefits from State and Local Economic Development Policies? Kalamazoo, Mich.: W.E. Upjohn Institute for Employment Research (January). ———. 1993. Economic Development and Black Economic Success. Report prepared for the U.S. Department of Commerce, Economic Development Administration. Upjohn Institute Technical Report 93-001. Kalamazoo, Mich.: W.E. Upjohn Institute for Employment Research. ———. 2001. Jobs for the Poor: Can Labor Demand Policies Help? New York: Russell Sage Foundation and Kalamazoo, Mich.: W.E. Upjohn Institute for Employment Research. ———. 2004a. “Economic Development.” In Management Policies in Local Government Finance, 5th ed., edited by J. Richard Aronson and Eli Schwartz, pp. 355–90. Washington: International City/County Management Association. ———. 2004b. “Thinking about Local Living Wage Requirements.” Urban Affairs Review 40, no. 2 (November): 269–99. ———. 2005. “Solving the Problems of Economic Development Incentives.” Growth and Change 36, no. 2: 139–66. ———. 2006. Taking Preschool Education Seriously as an Economic Development Program: Effects on Jobs and Earnings of State Residents Compared to Traditional Economic Development Programs. Report prepared for the Committee for Economic Development and the Pew Charitable Trusts. Kalamazoo, Mich.: W.E. Upjohn Institute for Employment Research. Bartik, Timothy J., and George Erickcek. 2007. “Higher Education, the Health Care Industry, and Metropolitan Regional Economic Development: What Can ‘Eds & Meds’ Do for the Economic Fortunes of a Metro Area’s Residents?” Paper prepared for the Conference on Urban and Regional Policy and Its Effects, Washington, March 29–30, 2007. Available as an Upjohn Institute Working Paper. Blackwell, Melanie, Steven Cobb, and David Weinberg. 2002. “The Economic Impact of Educational Institutions: Issues and Methodology.” Economic Development Quarterly 16, no.1: 88–95. Bound, John, and others. 2004. “Trade in University Training: Cross-State Variation in the Production and Stock of College-Educated Labor.” Journal of Econometrics 121, nos.1–2: 143–73. Card, David. 1995. “Using Geographic Variation in College Proximity to Estimate the Return to Schooling.” In Aspects of Labour Market Behaviour: Essays in Honour of John Van-

58

Timothy J. Bartik and George Erickcek

derkamp, edited by Louis N. Christofides, E. Kenneth Grant, and Robert Swidinsky, pp. 201–24. University of Toronto Press. College Board. 2006. Education Pays, Second Update. A Supplement to Education Pays 2004: The Benefits of Higher Education for Individuals and Society. Trends in Higher Education Series. Washington. Dickens, William T., and Lawrence F. Katz. 1987. “Inter-Industry Wage Differences and Industry Characteristics.” In Unemployment and the Structure of Labor Markets, edited by Kevin Lang and Jonathan S. Leonard, pp. 48–89. New York: Basil Blackwell. Feldman, Maryann, and Pierre Desrochers. 2003. “Research Universities and Local Economic Development: Lessons from the History of Johns Hopkins University.” Industry and Innovation 10, no. 1 (March): 5–24. Glaeser, Edward L., and Albert Saiz. 2004. “The Rise of the Skilled City.” In BrookingsWharton Papers on Urban Affairs: 2004, edited by William G. Gale and Janet Rothenberg Pack, pp. 47–94. Brookings. Gottlieb, Paul D., and Michael Fogarty. 2003. “Educational Attainment and Metropolitan Growth.” Economic Development Quarterly 17, no. 4 (November): 325–36. Groen, Jeffrey A. 2004. “The Effect of College Location on Migration of College-Educated Labor.” Journal of Econometrics 121, nos. 1–2: 125–42. Groshen, Erica L. 1991. “Five Reasons Why Wages Vary among Employers.” Industrial Relations 30, no. 3 (Fall): 350–81. Harkavy, Ira, and Harmon Zuckerman. 1999. “Eds and Meds: Cities’ Hidden Assets.” Brookings Institution, Center on Urban and Metropolitan Policy, Survey Series (August). Helms, L. Jay. 1985. “The Effect of State and Local Taxes on Economic Growth: A Time Series–Cross Section Approach.” The Review of Economics and Statistics 67, no. 4 (November): 574–82. Hill, Edward W., and Iryna Lendel. 2007. “The Impact of the Reputation of Bio-Life Science and Engineering Doctoral Programs on Regional Economic Development.” Economic Development Quarterly 21, no. 3 (August): 223–43. IES (Institute of Education Sciences). 2006. Digest of Education Statistics 2005. Washington: U.S. Department of Education, IES. Kentucky Rural Health Works Program. 2003–05. Economic Impact Studies. Louisville, Ky.: University of Kentucky, College of Agriculture, Kentucky Cooperative Extension Service. Krueger, Alan B., and Lawrence H. Summers. 1988. “Efficiency Wages and the InterIndustry Wage Structure.” Econometrica 56, no. 2 (March): 259–93. Leslie, Larry L., and Sheila A. Slaughter. 1992. “Higher Education and Regional Development.” In The Economics of American Higher Education, edited by William E. Becker and Darrell R. Lewis, pp. 223–52. Boston: Kluwer Academic Publishers. Lester, Richard K. 2005. “Universities, Innovation, and the Competitiveness of Local Economies: A Summary Report from the Local Innovation Systems Project–Phase I.” Working Paper 05-010. Cambridge, Mass.: MIT Industrial Performance Center (December). McDermott, R. E., G. C. Cornia, and R. J. Parsons. 1991. “The Economic Impact of Hospitals in Rural Communities.” Journal of Rural Health 7, no. 2: 117–33. Miller, Matthew. 2007. “Moving Forward: MSU Trustees OK Purchase of Site for Grand Rapids Med School.” Lansing State Journal, January 19.

Metropolitan Economic Development 59 Moretti, Enrico. 2004. “Estimating the Social Return to Higher Education: Evidence from Longitudinal and Repeated Cross-Sectional Data.” Journal of Econometrics 121, no. 1: 175–212. National Association of Public Hospitals and Health Systems. 2004. “How Are Safety Net Hospitals Financed? Who Pays for ‘Free Care’?” NAPH Issue Brief. Washington (September). NASULGC (National Association of State Universities and Land-Grant Colleges). 2001. Shaping the Future: The Economic Impact of Public Universities. Washington: NASULGC, Office of Public Affairs (August). NCHS (National Center for Health Statistics). 2006. Health, United States, 2006. Hyattsville, Md.: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention. Neumark, David, and Scott Adams. 2003. “Detecting Effects of Living Wage Laws.” Industrial Relations 42, no. 4 (October): 531–64. Orszag, Peter, and Joseph Stiglitz. 2001. “Budget Cuts vs. Tax Increases at the State Level: Is One More Counter-Productive than the Other during a Recession?” Washington: Center on Budget and Policy Priorities (November). Paytas, Jerry, Robert Gradeck, and Lena Andrews. 2004. “Universities and the Development of Industry Clusters.” Prepared for the Economic Development Administration, U.S. Department of Commerce. Pittsburgh: Center for Economic Development, Carnegie Mellon University. Pindus, Nancy M., and others. 2004. Evaluation of the Sectoral Employment Demonstration Program. Washington: Urban Institute (June). Safford, Sean. 2004. “Searching for Silicon Valley in the Rust Belt: The Evolution of Knowledge Networks in Akron and Rochester.” Working Paper 04-002. Cambridge, Mass.: MIT Industrial Performance Center, Local Innovation Systems project. Shapiro, Jesse M. 2006. “Smart Cities: Quality of Life, Productivity, and the Growth Effects of Human Capital.” Review of Economics and Statistics 88, no. 2: 324–35. Siegfried, John J., Allen R. Sanderson, and Peter McHenry. 2006. “The Economic Impact of Colleges and Universities.” Working Paper 06-W12. Vanderbilt University Department of Economics (May). Smith, Adam. 1776. An Inquiry into the Nature and Causes of the Wealth of Nations. University of Chicago Press, facsimile ed. (1977). U.S. Census Bureau. 2007. “Table 1. State and Local Government Finances by Level of Government and by State: 2004–05.” Governments Division (May 14) (www.census.gov/ govs/estimate/0500ussl_1.html). Varga, Attila. 2000. “Local Academic Knowledge Transfers and the Concentration of Economic Activity.” Journal of Regional Science 40, no. 2 (May): 289–309.

3 Low-Income Homeownership as an Asset-Building Tool: What Can We Tell Policymakers? george c. galster and anna m. santiago

F

ederal housing policy has for many generations encouraged owner occupancy over rental tenure.1 Since the 1980s, however, federal policies have explicitly extended this encouragement to households of ever-lower incomes. During the Reagan administration, the U.S. Department of Housing and Urban Development (HUD) piloted a program for selling public housing units to their occupants.2 In 1991 Congress established goals for Fannie Mae and Freddie Mac for the purchasing of mortgages originated for low-income borrowers and for homes located in traditionally underserved (that is, low-income and minorityoccupied) urban neighborhoods, in an attempt to ease potential liquidity constraints in these market segments. Substantial results have followed, often spawned by innovations in low down payment mortgage instruments.3 The Clinton administration set a national homeownership goal, and the Bush administration has placed the expansion of low-income and minority homeownership at the core of HUD’s mission.4 In the last decade, Congress has enacted pilot programs The research on Denver reported in this chapter is supported by grants from the Ford Foundation and the MacArthur Foundation. The views expressed in this chapter are the authors’ and do not necessarily reflect those of these foundations or the Board of Governors of Wayne State University. We also express our gratitude to the Denver Housing Authority for their support. Nina Butler, Caitlin Malloy, and Ana H. Santiago provided excellent research assistance. Debt also is owed to Richard Green, Margery Turner, and Hal Wolman. 1. Stegman and others (1991); Galster and Danielle (1996); Shlay (2006); Rohe (2007). 2. Rohe and Stegman (1992). 3. Listokin and others (2002). 4. Retsinas and Belsky (2002); HUD (2002).

60

Low-Income Homeownership

61

for homeownership education and counseling and housing choice vouchers for the purchase of homes.5 HUD’s 602 Nonprofit Property Disposition program sells HUD-held single family homes at deep discounts to local governments and nonprofits, which rehabilitate and resell the homes to low-income buyers. It is clear that this consistent political and institutional emphasis, coupled with policy innovations and a generally favorable macroeconomic climate, produced remarkable increases in homeownership rates among low-income households beginning in the mid-1990s. About 800,000 low-income households bought their first home in the 1990s, raising the homeownership rates of nonelderly households in the lowest income quintile from 22 percent in 1989 to 28 percent in 2001.6 Mortgage loans to low-income borrowers almost doubled between 1993 and 1999, raising their share of all home purchase loans from 14 to 19 percent during this period.7 Average real value of home equity, held by households with incomes below 80 percent of their metropolitan area’s median, rose rapidly: from $67,683 in 1985 to $76,505 in 1995 and $96,011 by 2001.8 In the last few years, the heralded explosion of home foreclosures has signaled retrenchment on many of these indicators, though its full magnitude has yet to become clear.9 Nevertheless, attaining and sustaining homeownership has remained an issue of high salience for the public and policymakers alike.10

Why Should Policymakers Care About Expanding Low-Income Homeownership? There are two basic, not mutually exclusive, categories of potential rationales for public polices aimed at expanding homeownership opportunities for lowincome households. The first is that homeownership conveys external benefits, “positive externalities,” upon the greater society that are over and above the benefits accrued by the homeowners themselves. The externalities typically cited include enhanced home maintenance, social and political participation, and attachment to community.11 These will not be considered in this chapter, although they may provide strong justification for public sector intervention. 5. Locke and others (2006); Rohe and Quercia (2003). 6. See Retsinas and Belsky (2002) regarding absolute number of first-time homebuyers and Nothaft and Chang (2005) for homeownership rates from 1989 to 2001. 7. See Pitcoff (2003) for the number of mortgage loans and Wiranowski (2003) for percent share of all home purchase loans. 8. Nothaft and Chang (2005). 9. Eduardo Porter and Vikas Bajaj, “Rising Trouble with Mortgages Clouds Dream of Owning Home,” New York Times, March 17, 2007, sec. A1, B4; Immergluck (2008). 10. Rohe and Watson (2007). 11. See literature reviews by McCarthy, Van Zandt, and Rohe (2001); Rohe, Van Zandt, and McCarthy (2000).

62

George C. Galster and Anna M. Santiago

The second category is that homeownership is essentially a merit good, one that has certain intrinsic, private benefits that all members of society deserve a chance to consume on the grounds of distributive justice. This rationale for expanding homeownership has been multifaceted.12 Some studies have described the benefits accruing to low-income homeowners as increased wealth, social status, security of tenure, control over dwelling, pride, and life satisfaction.13 Others have described benefits accruing to children who are able to live in homes owned by their parents.14 A variant of the merit good rationale for homeownership focuses on its potential antipoverty effects. This will be the focus of this chapter. Beginning in the early 1990s, some advocates and analysts began reframing the antipoverty strategic debate as one of asset acquisition instead of merely income support.15 They argue that building up assets offers two advantages over building income flows. First, assets provide a financial cushion when income flows are interrupted, such as during a layoff, injury, or illness. Second, assets may be drawn upon to enhance the earning power of the household, such as purchasing an automobile to aid access to employment or obtaining more training and education (for the adults and the children in the household). In this vein, homeownership may provide a superior method for accumulating assets compared with alternative ones available to low-income households. Given that a down payment is typically highly leveraged by mortgage money, even a modest appreciation of the home can produce startling rates of return on the initial investment by the buyer. Add to this the favorable federal tax treatment of housing capital gains, potential deductibility of mortgage interest and local property taxes, and payments of principal within the amortized mortgage and the superior wealth-building potential of homeownership compared with that of other assets appears even more attractive.16 Finally, since low-income households typically have little discretionary income that they can use to invest in other forms of savings, homeownership may be seen as “the only game in town.” Alternative investment vehicles, such as stocks and bonds, simply are moot as a basis for comparison of relative financial performance.17 12. Shlay (2006). 13. Rohe and Stegman (1994); Rohe and Basolo (1997); Rohe, Van Zandt, and McCarthy (2000); McCarthy, Van Zandt, and Rohe (2001); Rohe and Quercia (2003). 14. Green and White (1997); Boehm and Schlottmann (1999); Aaronson (2000); Harkness and Newman (2002, 2003); Haurin, Parcel, and Haurin (2002a, 2002b); Haurin, Dietz, and Weinberg (2002); Galster and others (forthcoming). 15. Sherraden (1991); Blank (2002). 16. Retsinas and Belsky (2005). 17. Orzechowski and Sepielli (2003); Collins (2004b); Boehm and Schlottmann (2004c); Belsky and Retsinas (2005); Herbert and Belsky (2006).

Low-Income Homeownership

63

Statistics leave no doubt that home equity is indeed the most important (and often only) component of wealth for most low-income households. The 2001 Survey of Consumer Finances shows, for example, that the median net worth of homeowners earning less than $20,000 was $72,750, whereas that of renters in the same income range was only $900.18 During the 1990s low-income, minority homeowners saw their average home equity rise by $1,712 annually, but nonhousing wealth of low-income, minority owners and renters alike did not change on average.19 Owning a home is not necessarily an unmitigated financial benefit for lowincome households, however. Five major caveats may be identified. First, lowincome home buyers may invest more in housing than is optimal compared with their investing in other assets, given the risk and return characteristics of various assets and the fact that the household’s effective marginal federal tax rate is likely zero.20 Second, low-income (especially minority) homeowners are much less likely to prepay their mortgages (typically with refinancing) when interest rate declines make such options profitable, thereby foregoing potential gains in wealth.21 Third, unexpected major home repairs or loss of income through illness, injury, or layoff raise the specter of unsustainable financial stress that potentially could culminate in mortgage delinquency and default, with concomitant psychological damages, loss of home equity, and destruction of consumer credit ratings. Fourth, the homes purchased by low-income buyers may not appreciate and may even decline in value over some period (especially if they are located in distressed neighborhoods), and low-income homeowners may have less flexibility in choosing when to sell. This risk is magnified because home equity represents a less-diversified portfolio than many alternative bundles of financial instruments. Fifth, low-income (especially minority) homeowners may be more susceptible to victimization by predatory lenders, leading to erosion of equity through excessive refinancing fees or, in the worst case, induced default and foreclosure.22 The latter three issues have been of preeminent concern, because they challenge the premise that homeownership for low-income households typically does lead to increased stability of tenure and the potential wealth acquisition associated with it.23 Thus the policy discussion has increasingly evolved from only con18. Belsky, Retsinas, and Duda (2007). 19. Boehm and Schlottmann (2004c). 20. Ambrose and Goetzmann (1998); Goetzmann and Spiegel (2002). 21. Van Order and Zorn (2002); Nothaft and Chang (2005). 22. Renuart (2004); Willis (2006); Goldstein (2006); Zimmerman (2006). 23. These concerns are expressed in Meyer, Yeager, and Burayidi (1994); Pitcoff (2003); Immergluck (2008).

64

George C. Galster and Anna M. Santiago

sidering how more low-income households can attain homeownership to also considering how more low-income households can sustain homeownership.24 These issues related to low-income homeownership as an asset-building strategy contribute vitally to broader discussions of urban and regional economic development policy. Should expanding the pool of lower-income homeowners be seen as a key component of local economic development strategy if, by doing so, it would help “grow the middle class?” If so, what particular strategies should be pursued and what pitfalls should be avoided? The chapter aims to respond to such concerns related to the desirability and sustainability of low-income homeownership in urban areas. In particular, we address six questions, each in a separate section of the chapter: —Does homeownership typically enable low-income households to build wealth? —Does homeownership typically provide low-income households with a financial buffer to weather financial setbacks? —Does homeownership for low-income households typically involve terms and conditions that, when coupled with their other instabilities, render a high probability of financial distress? —Does homeownership provide benefits to children in low-income households that might lead to superior economic outcomes in the next generation? —What are the primary barriers to attaining and sustaining homeownership for low-income households? —What mechanisms have we tried to expand and sustain homeownership among low-income households, and what has worked? To answer these questions, we review primarily extant literature, noting several important reviews of literature related to many of the aforementioned questions.25 The substance of our chapter differs from these works by our focus on the policy-relevant aspects of the literature and gaps therein and our use of previously unreported evidence from very-low-income households who recently purchased their homes after an extensive program that focused on asset building and homeownership education and counseling offered by the Housing Authority of the City and County of Denver. The following section provides a brief summary of this program. Before proceeding, however, we think it appropriate to address explicitly three overarching issues related to low-income homeownership policy that will be interwoven throughout the rest of the chapter: appropriate outcome meas24. For examples of attainability, see Galster, Aron, and Reeder (1999); Listokin and others (2002); for sustainability, see Wiranowski (2003); Shlay (2006). 25. Examples include Rohe, Van Zandt, and McCarthy (2000); McCarthy, Van Zandt, and Rohe (2001); Dietz and Haurin (2003); Herbert and others (2005); Herbert and Belsky (2006); Cortes and others (2006).

Low-Income Homeownership

65

ures, cross-group differences in outcomes, and geographic differences in outcomes. In most of the public discourse on low-income homeownership the desired outcomes have not been explicitly stated, although helping more households to buy a home seems to be an implicit one. We have suggested above that the ability to sustain homeownership over several years at minimum is a superior indicator of success. However, for the particular focus of this chapter—lowincome homeownership as an asset-building strategy related to urban and regional development—sustaining homeownership is only an intermediate outcome. It is of interest here only insofar as typically it is a necessary, but not sufficient, condition for housing equity growth. In this chapter, the primary outcome of interest will be the net wealth position of the low-income household. A secondary outcome of interest will be the human capital acquired by children of low-income parents. A focus on both outcomes is consistent with the view that fighting poverty and expanding the middle class can be accomplished by encouraging the creation of financial and human capital. The second overarching issue is the extent to which conclusions about the desirability, ability, and sustainability of low-income homeownership apply across the board to all racial and ethnic groups. Our analysis of the literature is that, in general, they do not. We therefore will take care in noting the extent to which our answers to the six research questions of this chapter require nuances to make them applicable to minority households. The third is the extent to which the desirability and sustainability of homeownership varies geographically. It is obvious that long-term intermetropolitan variations in home appreciation rates can make substantial differences in the growth of home equity. Some evidence of differential appreciation rates within metropolitan areas exists as well. These geographic considerations interact with the aforementioned racial and ethnic ones because persistent discrimination in housing and mortgage markets serves to limit the feasible residential choices (and, thereby, prospective home appreciation rates) of minorities.

The Denver Housing Authority’s Foundations for Home Ownership Program The Housing Authority of the City and County of Denver (DHA) has operated its voluntary Foundations for Home Ownership (FFHO) program since 1995, as a component of its Family Self-Sufficiency (FSS) and, subsequently, Resident Opportunities for Self-Sufficiency (ROSS) programs. The goal of FFHO is to assist DHA tenants to enhance their human, financial, and social capital assets, with the ultimate target of buying their own homes. Program participants are eligible for homeownership assessments, free credit reports, credit repair and

66 George C. Galster and Anna M. Santiago money management counseling, classes on a wide variety of topics (for example, housing finance, home repairs, and shopping for real estate and mortgages), individual development accounts with dollar-for-dollar matches up to $1,000, and (for FSS graduates) rent escrow accounts (where increments in DHA rents associated with increasing tenant income are placed into escrow for use as a down payment or other asset-building activities). Participants, working with program case management staff, develop individual training and services plans outlining their goals. During the final stage of the FFHO program, eligible participants (that is, those who are within a year of being able to purchase a home, have at least $500 in savings, and have stable employment) are invited to join the Home Buyer’s Club. The Home Buyer’s Club provides intensive real estate and finance training; presentations by housing industry representatives; peer support; and special benefits such as low interest rates, discount fees, and assistance with down payments and closing costs and second mortgages. Since the implementation of the FFHO program in 1995, 135 participants have purchased homes. According to DHA administrative records, most were single parents, with Latinos being the largest ethnic group represented. Their median housing price at the time of purchase was $136,523, and average monthly mortgage payments were $851. The vast majority of the participants obtained mortgages through public or nonprofit agencies and received supplemental mortgage assistance from DHA or other nonprofit housing or neighborhood development organizations. Slightly less than two-thirds of all FFHO homebuyers purchased homes in the city and county of Denver; the remainder bought homes in the greater Denver metropolitan area. Since 2001 the authors have been engaged in a panel study of a sample of these homeowners, the results from which we will report in this chapter.

Does Homeownership Typically Enable Low-Income Households to Build Wealth? For more than a century, homeownership has been the primary vehicle for building wealth among low- and moderate-income families in the United States. Indeed, accruing housing wealth has been or is the only wealth-building strategy employed by many low-income households.26 For households with wealth constraints, homeownership is one of the few leveraged investments available that enables owners with little or no equity to reap financial benefits primarily through appreciation in overall home value and through the forced savings associated with paying down the outstanding mortgage principal.27 The financial 26. Belsky and Retsinas (2005); Collins (2004b); Herbert and Belsky (2006). 27. Herbert and Belsky (2006).

Low-Income Homeownership

67

benefits of homeownership may be enhanced further by provisions in the tax law that allow for deductions of mortgage interest payments, property taxes, and housing capital gains. Christopher Herbert and Eric Belsky proposed that homeownership also contributes to the financial well-being of households by insulating them from rapidly escalating housing costs, especially when households can avail themselves of fixed-rate mortgage financing.28 All of these benefits, in turn, facilitate a greater potential for increased savings or other wealthgenerating activities by “owner households” rather than by “renter households.” Nonetheless, the advent of mortgage products that target ever-lower-income households has triggered growing skepticism among scholars and policymakers about the wealth-generating potential of homeownership. As William Apgar so clearly articulated: Many low-wealth and low-income families are being pushed into homeownership, not necessarily because they fully appreciate the implications of their choices, but because they perceive (or rather hope) that homeownership in and of itself will help them achieve a better life.29 Apgar draws attention to the deleterious consequences associated with lowincome homeownership when families assume mortgage debts that they are unable to repay. Under such circumstances, homeownership does not build wealth; rather, it diverts scarce financial resources away from meeting other pressing household needs. In this section, we summarize the literature assessing the extent to which homeownership serves to generate wealth among lowincome households. As of 2000, housing equity accounted for nearly one-fifth of total aggregate wealth in the United States, a proportion similar to the aggregate wealth held in stocks, bonds, and mutual funds combined.30 Nevertheless, housing equity was the primary form of aggregate wealth held by low-income and minority households. According to Shawna Orzechowski and Peter Sepielli, housing equity accounted for 56.2 percent of aggregate wealth in 2000 held by households in the lowest income quintile; in contrast, stocks and mutual funds accounted only for 7.7 percent of aggregate wealth of these households. Moreover, Orzechowski and Sepielli underscore the importance of homeownership as a source of wealth within minority households, which accounted for 61.8 percent of aggregate Black wealth and 50.8 percent of aggregate Latino wealth in 2000.31

28. Herbert and Belsky (2006). 29. Apgar (2004, p. 5). 30. Orzechowski and Sepielli (2003). 31. Orzechowski and Sepielli (2003).

68

George C. Galster and Anna M. Santiago

Despite widespread support for homeownership initiatives, the existing literature is sketchy and findings are mixed regarding the financial returns to homeownership for low-income homeowners. Moreover, given the lack of detailed longitudinal data about the short- and long-term rewards of homeownership, few studies have been conducted to date that examine housing wealth accumulation over time. Nor is much known about the relationship between the timing of a purchase of a house and wealth accumulation.32 Previous studies generally have employed simulations in an attempt to estimate these returns relative to housing tenure, race, income, structure, location and condition of units, and, more recently, duration of homeownership. In light of growing concerns about the financial returns of homeownership among ever-lower-income households, a number of recent studies have attempted to measure these returns.33 Although the findings are not consensual, these studies suggest that homeownership pays off for most families. The financial benefits attributed to homeownership are particularly striking when lowincome owner households are compared with similar renter households. Michael Collins estimated that the median wealth of a low-income homeowner under the age of 65 was twelve times that of a similar renter.34 Zhu Di, Yi Lang, and Xiadong Liu reported that the average net wealth of owners was 2.2 times that of renters.35 They also found that low-income households owning a home for eighteen years increased their average wealth by $49,700 compared with $5,700 for those who rented during the entire period. Using data from the Panel Study of Income Dynamics (PSID), Carolina Reid found that during the period between 1983 and 1994, low-income homeowners had accumulated roughly $25,000 to $30,000 in wealth compared with $0 for renters.36 Although recent studies suggest that the financial rewards to homeownership are generally positive across all homeowners, the literature also reports marked differences in these rewards between low-income and high-income market segments as well as by race.37 Using PSID data for period (1984–1993), Thomas Boehm and Alan Schlottmann estimated housing values, savings through amortization, rates of housing appreciation, nonhousing wealth, and savings for four racial and income groups. They reported that the average home values were $97,030 for high-income Whites, $70,094 for high-income minorities, $64,291 32. Boehm and Schlottmann (2004c). 33. For examples, see Boehm (2004); Boehm and Schlottmann (2002, 2004a, 2004c); Collins (2004b); Belsky, Retsinas, and Duda (2007); Herbert and Belsky (2006); Reid (2004, 2007); Stegman, Quercia, and Davis (2007). 34. Collins (2002). 35. Di, Yang, and Liu (2003). 36. Reid (2004). 37. See Boehm and Schlottmann (2004a, 2004c).

Low-Income Homeownership

69

for low-income Whites, and $42,454 for low-income minorities. Median savings realized through amortization (forced savings) were estimated to be $5,704 for high-income Whites, $4,120 for high-income minorities, $3,989 for lowincome Whites, and $2,634 for low-income minorities.38 Further, Boehm and Schlottmann found that median annual rates of appreciation ranged from 4.2 percent for low-income Whites and 4.3 percent for lowincome minorities to 4.6 percent for high-income Whites and 4.8 percent for high-income minorities. These rates translated to median annual housing wealth appreciation of $4,460 for high-income Whites, $3,359 for high-income minorities, $2,729 for low-income Whites, and $1,712 for low-income minorities.39 Nonetheless, their findings underscore the difficulties faced by lowincome and minority households in building nonhousing wealth over time. They reported that in 1984 median net nonhousing wealth was $20,700 for high-income Whites; for high-income minorities this net nonhousing wealth was $6,650. Low-income Whites had $5,000 in median net wealth; low-income minorities, $150. Median savings accrued between 1984 and 1993 was $2,650 for high-income Whites; $300 for high-income minorities, $300 for lowincome Whites, and $0 for low-income minorities.40 The bottom quartile of low-income households experienced an average annual change in nonhousing wealth that was negative, leading Boehm and Schlottmann to conclude that housing wealth and total wealth were synonymous for low-income households. Chenoa Flippen observed dramatic geographic differences in home appreciation rates across neighborhoods delineated by racial and ethnic composition. From 1970 to 1990, homes in neighborhoods with less than a 2 percent population of Black residents appreciated more than 23 percent, whereas those in neighborhoods with 2 to 30 percent Black residents appreciated 10 percent, and those in neighborhoods with a Black population of more than 30 percent appreciated less than 8 percent. Over the same period, the patterns were less clear for Latino composition. Homes in neighborhoods with less than a 2 percent Latino population appreciated more than 14 percent, those in neighborhoods with 2 to 5 percent appreciated 27 percent, with 5 to 10 percent appreciated 23 percent; appreciation was 15 percent in neighborhoods composed of more than 10 percent Latinos.41 In a recent evaluation of 21,497 Community Advantage Secondary Market Demonstration Program (CAP) loans originating between 1998 and 2002, Michael Stegman, Roberto Quercia, and Walter Davis reported that the median 38. Boehm and Schlottmann (2004c, p. 30). 39. Boehm and Schlottmann (2004c, p. 13). 40. Boehm and Schlottmann (2004c, p. 13). 41. Flippen (2004).

70

George C. Galster and Anna M. Santiago

net wealth of owners in the lowest income quintile was $68,000 in 2001, with home equity accounting for nearly 80 percent of the net wealth of low-income owners. They also found that CAP families enjoyed an annual average appreciation in home values of 5.4 percent and a median gain in housing value of $16,433. However, racial differences in equity appreciation were noted: Blacks experienced lower annual equity appreciation—approximately 10 per cent less per year—than did Whites.42 Data from our assessment of 135 loans made to participants in the Denver FFHO program between 1995 and 2006 reveal that the median net wealth of all owners was $36,100, with home equity accounting for 55.5 percent of the net wealth accrued to date by these owners. This varied from 71.3 percent of net White wealth, 51.6 percent of net Black wealth, to 55.5 percent of net Latino wealth in 2006. FFHO home buyers realized an annual average home value appreciation of 6.3 percent and a median gain in housing value of $37,188. Annual average appreciation rates again varied by ethnicity: homes owned by Blacks and Latinos appreciated at annual rates of 5.8 and 5.4 percent, respectively, less than half the rate for Whites (12.3 percent). Previous studies also noted important differences in appreciation by structure type, age, location, and physical condition of unit.43 Compared with moderateand high-income buyers, low-income households are less likely to purchase single family detached homes and more likely to purchase manufactured homes. Indeed, nearly one-quarter of first-time, low-income households purchased manufactured homes. Since large numbers of these homes are sited on leased land, appreciation of such units is limited.44 Nearly half of low-income buyers purchased homes that were built prior to 1970—units that often require significant updating and repair.45 Low-income homeowners also were more likely to purchase housing in low-income communities with lower housing values.46 One in twelve low-income home buyers lived in moderately or severely inadequate housing.47 All of these circumstances place constraints on current and future housing wealth accumulation for low-income homeowners. As a result of the high transaction costs associated with buying and maintaining a home, many of the financial benefits of homeownership accumulate only over a long period of time. Edwin Mills estimated that households need to own a home for five to seven years for homeownership to be a better financial choice 42. Stegman, Quercia, and Davis (2007). 43. Tong and Glascock (2000). 44. Boehm and Schlottmann (2004b). 45. Herbert and Belsky (2006). 46. Stuart (2000); Flippen (2004). 47. Herbert and Belsky (2006).

Low-Income Homeownership

71

than renting.48 In their analysis of changes in household wealth between 1984 and 1999 Di, Yang, and Liu found that owning a home for one to five years was associated with an increase in wealth of more than $50,000. Moreover, each additional year of homeownership up to seventeen years added about $6,700 a year to wealth, on average.49 Various studies by the authors Eric Belsky, Nicolas Retsinas, and Mark Duda stress the importance of timing of housing purchases and sales in estimating gains or losses experienced by low-income and other homeowners.50 Belsky, Retsinas, and Duda emphasize that returns to homeownership are heavily dependent upon the direction of home prices and rents after purchase.51 In housing markets experiencing significant upswings in home values, homeowners who purchased early in the boom period are likely to reap significant financial returns to their investments; those purchasing at the peak of a housing boom may be vulnerable to significant losses when the market cools. Conversely, homeowners who purchased in declining or slumping housing markets may experience significant losses to their investments, particularly if they are unable to ride out financial downturns. Duda and Belsky underscore the significant risk of negative returns associated with the purchase of low-cost homes. They reported nominal losses of 7 to 20 percent among low-income homeowners who may be unable to sustain homeownership in times of economic distress.52 Finally, investments in homeownership generate additional financial returns to owners when tax considerations, imputed rent, and financial leverage are included in estimates of housing wealth.53 Yet, as Christopher Herbert and Eric Belsky and others have emphasized, low-income home buyers are less likely to benefit from these additional financial returns because many pay little income tax or do not itemize deductions.

Does Homeownership Typically Provide Low-Income Households With a Financial Buffer to Weather Financial Setbacks? Not coincidentally, the tremendous growth in the number of low-income homeowners during the 1990s has been associated with the introduction of an array of flexible and riskier mortgage products that have made it possible to buy a home with higher levels of debt, lower levels of savings, and worse credit histo48. Mills (1990). 49. Di, Yang, and Liu (2003). 50. See Duda and Belsky (2000); Belsky and Duda (2002); Belsky, Retsinas, and Duda (2007). 51. Belsky, Retsinas, and Duda (2007). 52. Duda and Belsky (2000). 53. Herbert and Belsky (2006).

72

George C. Galster and Anna M. Santiago

ries than was possible in the past.54 Further exacerbating these risks among lowincome homeowners are the increased likelihood of involuntary employment and the use of multiple incomes to service a loan.55 As a result, the boom in lowincome homeownership during the past decade may have brought many households into homeownership that already have lower average income and wealth, carry high debt burdens, are less capable of meeting their mortgage obligations under the best of circumstances, and, therefore, are more susceptible to losing their homes during times of financial crisis. Recent studies have stressed the tenuous nature of homeownership for lowincome owners.56 Contributing to this heightened vulnerability are several demographic characteristics that distinguish low-income from high-income owner households. Low-income owner households are more likely to have elderly and disabled members or be headed by sole wage earners. Josephine Louie, Eric Belsky, and Nancy McArdle reported that about half of lower-income owner households had elderly members and 17 percent had at least one disabled member. They also found that nearly one in ten low-income owner households was headed by a single parent with children.57 In their recent analysis of American Housing Survey data for the period from 1989 to 2003, Herbert and Belsky determined that nearly half of low-income home buyers were in single-earner households; in contrast, only 10 percent of moderate- to high-income buyers had only one wage earner in the household.58 Among FFHO program participants, six out of ten home buyers were in single-earner households. With each of these different household configurations, serious concerns are raised about the ability of these owners to respond to financial crises, such as job loss or illness in the absence or loss of multiple wage earners, and support the financial costs of homeownership. Given these concerns about the short- and long-term financial viability of homeownership for low-income households, do these households accrue sufficient income and wealth postpurchase to sustain homeownership? Although prior research is limited to examining the postpurchase earnings of low-income households, recent work by Donald Haurin and Stuart Rosenthal has provided some estimates of income growth among new homeowners. They reported that real postpurchase income of low-income homeowners grew at an annual rate of 12.9 percent—a rate about three times higher than the rate for moderate- to 54. Herbert and Belsky (2006). 55. McCarthy, Van Zandt, and Rohe (2001). 56. Recent studies include Louie, Belsky, and McArdle (1998); Orr and Peach (1999); Ford and Quilgars (2001); Boehm and Schlottmann (2004a, 2004c); Herbert and Belsky (2006). 57. Louie, Belsky, and McArdle (1998). 58. Herbert and Belsky (2006).

Low-Income Homeownership

73

high-income homeowners (3.6 percent). In addition, they found that the share of the household budget devoted to mortgage payments declined rapidly in the early years of homeownership for the typical low-income homeowner. With increasing postpurchase real income and concomitant declining housing cost burdens, Haurin and Rosenthal speculated that low-income households would be able to sustain their homeownership even when faced with unexpected maintenance costs or other expenses.59 In their study of 239 home buyers from the Neighborhood Reinvestment Homeownership Pilot program, William Rohe and Roberto Quercia found that buyers experienced a 4 percent loss in assets between the baseline and second interviews (generally eighteen months) but a 27 percent increase in average income (from $30,300 to $38,586).60 FFHO homeowners in Denver experienced a $1,612 increase in household income for each year of homeownership.61 Less optimistically, Stegman and others reported that families with very low and low expected incomes were more likely to remain persistently poor.62 As a result, housing assistance programs targeted to persistently poor families are more likely to require deeper and longer-term homeownership subsidies than are those that target low-income families who are only temporarily poor. Several additional factors also contribute to the fragility of homeownership for low-income owners: higher monthly housing cost burdens associated with mortgage payments, taxes, insurance, and utilities; the costs of regular home maintenance; and the costs for unanticipated maintenance and repairs. For many low-income homeowners, the housing cost burden associated with homeownership may be as high as 40 to 60 percent of total monthly household income.63 During the past decade, increases in low-income homeownership have been linked to increases in severe housing cost burdens, defined as payments in excess of 50 percent or more of monthly household income for housing costs. Analyzing American Housing Survey data, Herbert and Belsky reported that one in five low-income, first-time home buyers was paying more than 50 percent of of his or her income for housing in the period since 1997—an increase of 5 percentage points from the period between 1989 and 1995. Herbert and Belsky also reported that another third of all first-time home buyers experienced moderate housing cost burdens from 1989 to 2003. They found that nearly one in ten first-time home buyers had a severe payment burden at time of purchase.64 In their recent study, Rohe and Quercia found that 60 percent of these 59. Haurin and Rosenthal (2005a). 60. Rohe and Quercia (2003). 61. Santiago and Galster (2006). 62. Stegman and others (1991, p. 72). 63. Orr and Peach (1999). 64. Herbert and Belsky (2006, p. 55).

74

George C. Galster and Anna M. Santiago

low-income home buyers experienced increases in housing costs within the first few years of homeownership.65 FFHO home buyers also experienced significant housing cost burdens at the time of purchase: the typical household was spending approximately 52 percent of monthly household income for housing costs. While these housing costs have increased over time, the housing cost burden among FFHO participants has actually declined: by 2006 the typical FFHO household was spending 34 percent of its monthly earnings on housing costs. As Stegman and others underscored, unanticipated housing cost increases play havoc with low-income homeowners’ budgets.66 Nonetheless, these types of costs were very common. Nearly half of the participants in the Neighborhood Reinvestment Homeownership Pilot program experienced unexpected costs, with the most common problem being a repair to one of the home’s major systems. These participants also reported major unexpected increases in utility payments (36 percent), property taxes (27 percent), and homeowners insurance (16 percent).67 In addition, previous studies suggest that many low-income homeowners cannot afford the upkeep on their homes, let alone invest in home improvements.68 From 1984 to 1993, nearly 1 million low-income homeowners spent less than $100 per year on maintenance.69 Given the heightened inability of many low-income homeowners to weather any one of these aforementioned economic shocks, it is not surprising that previous studies have reported high attrition rates among low-income homeowners, further reducing the likelihood that low-income households can attain the financial rewards associated with long-term homeownership.70 In their analysis of first-time home buyers using National Longitudinal Survey of Youth (NLSY) data, Haurin and Rosenthal found that the likelihood of attrition from homeownership status varies over the duration of homeownership, with terminations of homeownership peaking in the third year at 7 percent, decreasing to 5 percent in year 5, and leveling off at 2 percent by year 10. They concluded that the risk of ending homeownership remains significant through the seventh year. Attrition among FFHO home buyers was low, however: only 6 percent had sold their homes within the first five years of homeownership.71 Previous studies underscore that the low-income and minority home buyers face a higher risk of being unable to sustain homeownership over time.72 Estimat65. Rohe and Quercia (2003). 66. Stegman and others (1991). 67. Rohe and Quercia (2003). 68. Herbert and Belsky (2006). 69. Louie, Belsky, and McArdle (1998). 70. Examples of these studies include Boehm and Schlottmann (2004a); Reid (2004); Haurin and Rosenthal (2005a). 71. Haurin and Rosenthal (2005a). 72. See Reid (2004); Haurin and Rosenthal (2005a).

Low-Income Homeownership

75

ing from NLSY data, Haurin and Rosenthal reported that 43 percent of lowincome homeowners do not sustain homeownership for more than five years.73 Reid found that 53 percent of low-income PSID homebuyers left homeownership within five years of buying their first home compared with 23 percent of high-income buyers.74 Haurin and Rosenthal estimated that minority households have a 40 percent greater hazard rate of terminating a period of homeownerhsip compared with the rate for Whites.75 Similarly, Boehm and Schlottmann reported that low-income households were more likely to slip back to renting after attaining homeownership: nearly one in eight White families and nearly one in four minority families became renters again. Further, Boehm and Schlottmann observed that minority households were less likely to return to homeownership if they became renters again.76 As Herbert and Belsky noted, the “high rates of exit for low-income and minority first-time buyers are a cause for concern as the benefits of homeownership will generally be much greater for those who continue as owners for longer periods.”77 It is interesting that the extant literature emphasizes that attrition among low-income homeowners generally is not linked to their higher propensity to default on their mortgages.78 Rather, trigger events, which are unanticipated changes in a household’s circumstances, are prominent reasons for ending periods of homeownership.79 Major trigger events associated with reducing the ability to maintain homeownership include a reduction in earnings as a result of job loss, the breakup of a household due to divorce or separation, or an increase in expenses or a reduction in earnings owing to a health crisis. Reid observed that job loss was more common among low-income households, with nearly one in ten low-income households experiencing a period of unemployment between 1984 and 1993.80 Boehm and Schlottmann reported that the loss of income and assets was the major determinant of low-income households slipping from homeownership to renting.81 Haurin and Rosenthal also found that a significant decline in earnings was associated with termination of homeownership: the average decrease in earnings in the year of a termination was $13,629, or about 37 percent of the average low-income earnings.82 Reasons cited for the 73. Haurin and Rosenthal (2005a). 74. Reid (2004). 75. Haurin and Rosenthal (2005a). 76. Boehm and Schlottmann (2004a). 77. Herbert and Belsky (2006, p. 51). 78. Herbert and Belsky (2006). 79. See Vandell (1995); Cutts and Green (2004); Herbert and Belsky (2006). 80. Reid (2004). 81. Boehm and Schlottmann (2004a). 82. Haurin and Rosenthal (2005a).

76

George C. Galster and Anna M. Santiago

reduction in earnings included decreased hours worked, decreased wages, change in the number of earners in the household, and termination of a marriage. Indeed, the termination of a marriage has been identified as the single trigger event most strongly associated with termination of ownership. According to Reid, divorces raised the probability of leaving homeownership by a factor of 10.83 Haurin and Rosenthal reported that divorce increased the probability of an exit from homeownership by 40 percent.84 Other characteristics associated with the risk of leaving homeownership are age and education: younger households and those with lower levels of education are found to be at greater risk.85 Haurin and Rosenthal also observed that the likelihood of terminating a period of homeownership was associated with high state unemployment rates at the beginning of the period under study, increasing state unemployment rates during the period, and low household earnings at the time of home purchase.86 There is one further dimension of this topic that should be briefly mentioned: homeownership’s potential impact on labor market mobility. Some have argued that homeownership renders workers more vulnerable to severe downturns in the regional economy because it reduces the likelihood that they will move out of the area to one where job prospects are better.87 However, empirical evidence from Edward Coulson and Lynn Fisher, which is based on individuals in the United States, clearly showed that this hypothesis is rejected.88

Does Homeownership for Low-Income Households Typically Involve Terms and Conditions that, when Coupled with Their Other Instabilities, Render a High Probability of Financial Distress? The duration of homeownership as well as the financial returns to homeownership are sensitive to mortgage terms, mortgage interest rates and fees, the size of the mortgage relative to house value, and changes in the local macroeconomic climate.89 From 1993 to 1999, mortgage loans to low-income home buyers increased by 94 percent.90 While innovative mortgage products have enabled traditionally underserved populations to enter the housing market in unprecedented numbers, there are reasons for concern about the financial terms associ83. Reid (2004). 84. Haurin and Rosenthal (2005a). 85. Herbert and Belsky (2006). 86. Haurin and Rosenthal (2005a, 2005b). 87. Oswald (1999); Green and Hendershott (1999). 88. Coulson and Fisher (2002). 89. See Belsky, Retsinas, and Duda (2007); Haurin and Rosenthal (2005a, 2005b). 90. Pitcoff (2003).

Low-Income Homeownership

77

ated with this upsurge in low-income homeownership; some frequently cited ones include limited or negative equity at the time of home purchase, higher interest rates, and the rapid expansion of subprime lending in low-income neighborhoods. The first concern is limited initial equity in the home. In the past ten to fifteen years, greater shares of all first-time home buyers have secured loans with higher loan-to-value (LTV) ratios, making them more susceptible to fluctuations in house prices. Four out of ten first-time borrowers made down payments of 10 percent or less; nearly one in five put down 5 percent or less.91 Among first-time, low-income home buyers, 24 percent had LTV ratios exceeding 0.95.92 In a recent study of low-income homeowners who purchased homes through the Philadelphia 500 program, Harriet Newburger reported that 88 percent of the loans had an LTV value above 0.95, 50 percent were above 0.97, and 21 percent were above 1.00.93 Among buyers in the Section 8 Homeownership program, the average LTV was 0.997.94 Nearly one out of every four FFHO homebuyers had an LTV ratio of 1.00 or higher. Thus a growing number of low-income homeowners have little, no, or negative equity—a situation that can yield financial benefits in a booming housing market but is more likely to yield substantial losses and higher chances of default in declining housing markets. The second concern is that low-income borrowers are more vulnerable to short-term changes in interest rates. During the 1990s, fewer low-income home buyers held thirty-year, fixed interest rate mortgages in comparison with their higher-income counterparts.95 While six out of ten low-income buyers held such mortgages, approximately eight out of ten moderate- and high-income buyers did so. For those without a fixed rate, interest rate increases will have profound impacts on the costs of homeownership. The third concern is that even if the interest rate is fixed, it may be higher for lower-income households. Before 1995, there was a clear tendency for lowincome buyers rather than higher-income buyers to face higher interest rates.96 The average mortgage interest rate paid by low-income buyers was 8.81 percent; in contrast, moderate-income buyers paid, on average, 8.48 percent, and highincome buyers paid 8.46 percent. Between 1995 and 2003, the average interest rate for low-income buyers had declined to 7.24 percent.97 The average interest rate for FFHO homebuyers was 6.12 percent, although it varied from 5.23 per91. Pitcoff (2003). 92. Herbert and Belsky (2006). 93. Newburger (2006). 94. Locke and others (2006). 95. Herbert and Belsky (2006). 96. Herbert and Belsky (2006). 97. Herbert and Belsky (2006, p. 34).

78

George C. Galster and Anna M. Santiago

cent for White owners, 5.94 percent for Black owners, to 6.23 percent for Latino owners.98 Moreover, these differential interest rates significantly increase the housing cost burden for low-income households. James Carr and Jenny Schuetz estimated that every additional percentage point added to a thirty-year mortgage increases the total interest paid over the life of the mortgage by at least $20,000.99 As Belsky, Retsinas, and Duda demonstrated, the interest paid on a fifteen-year, $90,000 mortgage for a house purchased with a 5 percent down payment would be $135,646 at a 7 percent rate.100 However, at a 9 percent rate, the interest would be $179,629, and at 12 percent, the interest would be $249,709. For that same house, the authors estimated a $70,000 reduction in wealth over the course of fifteen years that was associated with the higher rates.101 The literature also suggests some racial differences in mortgage interest rates, although Raphael Bostic, Paul Calem, and Susan Wachter found that the quality of credit for homeowners improved for virtually all race and income groups.102 Indeed, Herbert and Belsky argue that no longer is there a significant tendency for low-income or minority home buyers to face higher interest rates.103 Using data from the Annual Housing Survey, however, Scott Susin and Thomas Boehm, Paul Thistle, and Alan Schlottmann separately found that Blacks pay higher interest rates than do Whites, which the two studies attributed to differences in rates obtained through refinancing as well as the greater likelihood of Black homeowners to use subprime lenders.104 Boehm, Thistle, and Schlottmann noted that about 87 percent of the difference between Black and White refinance interest rates is associated with differential treatment in the lending market, not differences in individual characteristics.105 Higher costs associated with higher mortgage interest rates also have been linked to the greater likelihood of slipping out of homeownership. Haurin and Rosenthal found that a 1 percentage point increase in the initial mortgage interest rate raises the hazard rate for termination of homeownership by 16 percent annually. They reported that the impact of increasing interest rates is even more striking after home purchase: a 1 percentage point increase in the postpurchase mortgage interest rate increased the rate of homeownership termination by 30 percent. Conversely, Haurin and Rosenthal noted that a 1 percentage point 98. Administrative data from the DHA FFHO program. 99. Carr and Schuetz (2001). 100. Belsky, Retsinas, and Duda (2005, pp. 10–11). 101. Belsky, Retsinas, and Duda (2005, p. 11). 102. Bostic, Calem, and Wachter (2004). 103. Herbert and Belsky (2006). 104. Susin (2003); Boehm, Thistle, and Schlottmann (2006). 105. Boehm, Thistle, and Schlottmann (2006).

Low-Income Homeownership

79

decline in postpurchase interest rates reduced the risk of ending homeownership by 15 percent.106 A fourth concern is that, despite the fact that they are likely to pay more for their mortgage, low-income and minority households are less likely than are higher-income households to capitalize on declining interest rates to refinance their primary mortgage.107 In their study of more than 2,000 homeowners, Glenn Canner, Karen Dynan, and Wayne Passmore reported that borrowers with incomes under $40,000 were 1.4 percent less likely to refinance.108 Herbert and Belsky found that only 12 percent of low-income homeowners had refinanced their primary mortgage by 2003—about a third of the share of highincome homeowners who refinanced.109 Of interest, more than one-quarter of the FFHO homeowners had refinanced their homes by 2006. Nonetheless, Frank Nothaft and Yan Chang estimated that 6.9 percent of low-income homeowners nationally miss out on favorable refinance opportunities, with an estimated total lost benefit of $21.9 billion.110 A fifth concern is the extent to which higher interest rates and increased likelihood of dropping out of homeownership can be traced to the proliferation of subprime mortgage products in the past decade.111 During the 1990s, the number of subprime home purchase and refinance loans made to all borrowers increased dramatically.112 However, subprime lending continued to be disproportionately concentrated among low-income and minority borrowers and neighborhoods.113 Belsky, Retsinas, and Duda estimated that subprime loans accounted for 10 percent of the home purchase loans and 21 percent of the refinance loans originated in low-income neighborhoods by 2001—shares that are approximately two to three times higher than the share of subprime loans originating in high-income areas.114 Subprime loans also are more common within low-income minority communities, accounting for approximately 13 percent of all purchase mortgages in 1993 and 28 percent of refinance loans in 2001.115 A recent study by the Neighborhood Housing Services of Chicago underscores that borrowers within these communities are increasingly obtaining high-risk 106. Haurin and Rosenthal (2005a). 107. See Canner, Dynan, and Passmore (2002); Van Order and Zorn (2002); Nothaft and Chang (2004). 108. Canner, Dynan, and Passmore (2002). 109. Herbert and Belsky (2006). 110. Nothaft and Chang (2004). 111. Willis (2006); Immergluck (2008). 112. Belsky, Retsinas, and Duda (2007). 113. Apgar and Herbert (2005); Fishbein and Woodall (2005); Immergluck (2008). 114. Belsky, Retsinas, and Duda (2007). 115. Belsky and Retsinas (2005).

80

George C. Galster and Anna M. Santiago

loans without consideration or understanding of how to manage that debt or when to ask for help.116 Moreover, subprime loans are costly and riskier.117 Amy Cutts and Robert Van Order estimated that the average rate of subprime loans was 2.5 to 3 percentage points higher than a prime loan rate.118 For low-income homeowners holding subprime loans, the burden of even higher interest rates and fees further exacerbates their financial difficulties and accentuates their vulnerability.119 Therefore, it is not surprising that delinquency and foreclosure rates for subprime loans are significantly higher than those for conventional prime mortgages. Apgar noted that the serious delinquency rate for subprime mortgages was 10.44 percent in 2002—nearly twenty times higher than that for conventional prime mortgages and twice as high as the serious delinquency rate for Federal Housing Administration (FHA) loans.120 Subprime foreclosure rates are approximately eight times as high as the rates for prime conventional loans.121 Of course, the ultimate indicators of financial distress are mortgage delinquency and default. Previous studies consistently have found that households with lower incomes are more likely to miss payments and default on their mortgages.122 Quercia and others found that 5 percent of very-low-income borrowers experienced ninety-day delinquencies; however, less than one-third of these loans eventually ended in foreclosure.123 In their analysis of nearly 40,000 loans originated under Freddie Mac’s Affordable Gold program, Abdighani Hirad and Peter Zorn found that the rate of seriously delinquent loans was 6.9 percent, relative to a portfolio average of 1.8 percent for all other Freddie Mac loans.124 Gretchen Locke and others reported that 86 percent of the 206 public housing authorities participating in the Section 8 Homeownership program had no delinquencies or loans in default.125 None of the FFHO program home buyers had experienced any serious delinquencies during the twelve months before their interview; however, nearly one in five had made a late mortgage payment at some point during the last twelve months. Delinquency rates vary by ethnicity, size of loan, and type of structure. Black borrowers were more likely to experience repayment difficulties than were 116. NHS (2004). 117. Immergluck (2008). 118. Cutts and Van Order (2005). 119. Wiranowski (2003); Willis (2006). 120. Apgar (2004, p. 31). 121. Danis and Pennington-Cross (2005). 122. See Quercia and Stegman (1992); Deng, Quigley, and Van Order (1996); Hirad and Zorn (2001, 2002); Quercia and others (2002); Van Order and Zorn (2002). 123. Quercia and others (2002). 124. Hirad and Zorn (2001). 125. Locke and others (2006).

Low-Income Homeownership

81

White borrowers. In contrast, delinquency rates are lower for borrowers with larger loan origination amounts and for those who took out mortgages on condominiums.126 As was the case for homeownership terminations, the most frequently cited triggers for Freddie Mac delinquencies included unemployment or curtailment of income (40 percent), illness or death of the borrower or other family member (24 percent), marital difficulties (10 percent), and excessive financial obligations (10 percent). However, Mark Wiranowski underscored another source of mortgage delinquencies, namely, the borrower’s financial choices and spending patterns that may lead to insolvency.127 Foreclosure among low-income homeowners remains a rare event, though it is more likely with loan products that offer lower down payments and higher debt ratios. Van Order and Zorn estimated foreclosure rates varying from less than 1 percent for prime Freddie Mac loans to 6 or 7 percent of FHA or Affordable Gold loans.128 Defaults on FHA loans have more than tripled from 1986 to 2004.129 The experience with government-supported low-income homeownership programs also shows a mixed record of defaults. A recent evaluation of purchases through the Section 8 Homeownership program revealed only ten foreclosures out of more than 3,400 purchases.130 Our evaluation of FFHO program purchases found three foreclosures out of the 131 homes purchased before 2006, reflecting a 2.3 percent foreclosure rate. Newburger reported, however, that nearly one out of five low-income households purchasing homes through the Philadelphia 500 program experienced at least one foreclosure with half of the these filings occurring within the three-year period after purchase.131 The likelihood of experiencing a foreclosure varies by race as well as by initial down payment levels and local macroeconomic conditions.132 In addition, minority homeowners experience greater likelihoods of losing their homes to foreclosure.133 Further, foreclosure rates have been increasing in specific markets and neighborhoods since the mid-1990s and have been linked to increases in subprime lending.134 Among our FFHO program participants, all of the foreclosures to date occurred with Latino homebuyers; further, they all have occurred since 2005. 126. Hirad and Zorn (2001, p. 14). 127. Wiranowski (2003). 128. Van Order and Zorn (2002); Hirad and Zorn (2001); Mortgage Bankers Association (2005). 129. Herbert and Belsky (2006). 130. Locke and others (2006). 131. Newburger (2006). 132. Quercia and Wachter (1996); Herbert and Belsky (2006). 133. Herbert and Belsky (2006). 134. Immergluck and Smith (2004).

82

George C. Galster and Anna M. Santiago

Instead of using their home equity as a “reserve” to buffer future unexpected claims, low-income homeowners appear to be drawing down on it to finance current consumption, thereby losing a potential buffering opportunity. Previous studies, such as those by Winton Pitcoff and by Rohe and Quercia, have underscored this use of refinancing as a means to reduce burgeoning consumer debt, thus shifting increasing amounts of consumer debt to mortgage debt as well as increasing the risk of mortgage default.135 Pitcoff estimated that one out of every five homeowners refinances each year. Among those who refinance, slightly less than one-third do so to pay down other—mostly credit card—debt. Rohe and Quercia found that while 15 percent of Neighborhood Reinvestment program participants refinanced their mortgage primarily to obtain a better interest rate, about the same percentage—14 percent—took out home equity loans, with nearly a third of these homeowners using the loans to obtain cash. Approximately 16 percent of FFHO home buyers took out home equity loans within the first few years of buying their homes, primarily for the purpose of debt consolidation, home repairs, or other types of consumption. The increased “cashing out” of home equity by homeowners corresponds with the rapid growth of both housing debt and consumer debt during the1990s. George Masnick, Zhu Di, and Eric Belsky observed that the median debt per owner household grew by 140 percent in the 1990s, from $13,700 to $33,100.136 Rohe and Quercia reported that the outstanding debt of homeowners in the Neighborhood Reinvestment program increased by 12 percent within the first few years of homeownership. Concomitantly, the monthly debt payments for these low-income homeowners increased from $397 at the time of home purchase to $589 by the end of their first eighteen months of homeownership.137 Outstanding debt increased among FFHO participants as well. By 2006 FFHO homeowners carried on average $10,007 in debts. Not surprisingly, monthly debt payments rose on average from $346 at the time of home purchase to $455.

Does Homeownership Provide Benefits to Children in Low-Income Households that Might Lead to Superior Economic Outcomes in the Next Generation? Conventional wisdom would probably lead one to answer “of course!” Indeed, the scholarly literature has generally supported the conventional wisdom here. However, one must be careful in leaping to this conclusion. The statistical study of homeownership must confront three serious challenges that relate essentially 135. Pitcoff (2003); Rohe and Quercia (2003). 136. Masnick, Di, and Belsky (2005). 137. Rohe and Quercia (2003).

Low-Income Homeownership

83

to omitted control variables. First, parents who select homeownership may do so because they seek a home-centered, supportive environment as one component of a multifaceted effort to raise children in the best way possible. These parents are more prone to be future oriented and more willing to invest time and resources in their children. Unfortunately, these parental attributes are poorly measured in surveys available to researchers, and thus they remain as uncontrolled variables in a multivariate model testing the relationship between homeownership status and child outcomes. This raises the problem of selection bias: Are observed positive correlations between child outcomes and homeownership truly caused by owning a home, or is the correlation merely spurious because of omitted parental characteristics affecting both tenure and outcomes? Second, because of their preferences for stability and their substantial investment in transactions costs, homeowners typically move less frequently than renters. If residential mobility is not controlled, is the positive association between homeownership and child outcomes due to the fact of owning a home, or is it due to residential stability? Finally, the parents’ choice of whether to rent or own is made simultaneously with the choice of neighborhood and expectations of residential mobility. Those who can only find feasible options of purchasing a home in neighborhoods unlikely to appreciate in the future and those who expect to move soon are unlikely to choose homeownership. But neighborhood environment is also likely to exert some independent effect on child outcomes. So, once again, models must control for neighborhood environment if they hope to gain an unbiased estimate of the true causal impact of homeownership on children. Unfortunately, few extant studies meet all these challenges, and hence the literature in this area must be treated with caution. Several causal mechanisms have been advanced through which parental homeownership status may produce felicitous outcomes for children living in the home —directly and indirectly (via mobility). The direct effects that have been posited include the following: —Homeowners maintain their dwellings to higher standards than do otherwise identical households who are renting, which may affect differentially the health and cognitive and social development of resident children.138 —Homeowners may acquire a distinctive set of skills, such as those related to do-it-yourself home repairs; negotiating with contractors, plumbers, and so on; or seeking refinancing. Insofar as these may be transferable to children, the latter will benefit.139 138. For home maintenance standards kept by homeowners compared with those by renters, see Galster (1983, 1987); Mayer (1981); regarding the effects of standards on resident children, see Parcel and Menaghan (1994a, 1994b). 139. Green and White (1997); Boehm and Schlottmann (1999).

84

George C. Galster and Anna M. Santiago

—Homeowners may have more financial stake in the occupied residence and thus more motivation to monitor and control activities of children (both their own as well as neighbors’) that might threaten the neighborhood’s property value.140 —Homeowners may invest more in developing social capital and participating actively in the neighborhood, from which their children may benefit in a variety of ways.141 —Buying a home may yield gains in parental satisfaction and self-esteem, which in turn translate into a more supportive, positive sociopsychological environment for children.142 —Homeowners may gain a superior equity position than renters to the extent that home appreciation outperforms other financial instruments (as discussed above in the section on homeownership and building wealth), and thus they would be able to invest more in the educational and nurturing aspects of the children’s environment. —Homeowners may experience lower levels of stress because of greater security of tenure, which produces more positive behavioral and cognitive outcomes for children in the household.143 The indirect consequences putatively transpire through the effect of homeownership status on residential stability. The argument proceeds as follows. Given the high transaction costs of home sale and purchase, owners typically reside in any given unit longer than do renters.144 In turn, this enhanced residential stability can have numerous positive impacts on children in the areas of educational achievement and credential attainment, substance use, social functioning, mental health, and sexual and deviant behaviors.145 At least one potential reason for this relationship is that as children remain longer in a neighborhood they are likely to become better known to adult neighbors, thus rendering them more subject to behavioral modifications through the exercise of neighbors’ “collective efficacy.”146 Unfortunately, from extant empirical work it is not possible to distinguish definitively among the above causal hypotheses. As we shall see below, there is a 140. Haurin, Parcel, and Haurin (2002a, 2002b); Hoff and Sen (2005). 141. Hunter (1975); Cox (1982); Jeffers and Dobos (1984); Coleman (1988, 1990); Austin and Baba (1990); Rohe and Stegman (1994); Verba, Schlozman, and Brody (1995); Rossi and Weber (1996); DiPasquale and Glaeser (1999). 142. Balfour and Smith (1996); Rossi and Weber (1996). 143. Cairney and Boyle (2004). 144. See Haurin, Hendershott, and Ling (1988) regarding high transaction costs. Regarding length of occupancy, see Lee, Oropesa, and Kanan (1994); Rohe and Stewart (1996). 145. Buerkle (1997); DeWitt (1998); Huffines (2003); Potter and others (2001); Rumberger (2003); Rumberger and Larson (1998); Stack (1994); Temple and Reynolds (1999). 146. Sampson, Morenoff, and Earls (1999).

Low-Income Homeownership

85

substantial body of credible empirical work to suggest that homeownership probably does something positive for children. Unfortunately, it must be left to future studies to ascertain precisely what this something is. The earliest statistical research in this area, by Richard Green and Michelle White, found that current parental homeownership was associated with a reduced probability of a resident seventeen-year-old dropping out of high school or giving birth.147 Past residential mobility and homeownership status were not controlled, however, and subsequent analyses by Daniel Aaronson and by Joseph Harkness and Sandra Newman showed that most of the relationships were explained by the greater residential stability associated with homeownership.148 Michael Boyle’s study of longitudinal samples of children in Ontario, Canada, revealed significantly lower rates of emotional-behavioral problems among children of homeowners, controlling for a variety of family and neighborhood characteristics.149 However, he did not attempt to deal with selection bias. Indeed, a good deal of skepticism remains about the true independent causal impact of homeownership. Most notably, David Barker and Eric Miller found that many of the reported effects of homeownership disappear when more controls and alternative estimation techniques for selection effects are introduced.150 However, a core of studies is available that employ sufficiently robust methodologies such that their findings can be interpreted as credible support for direct causal effects of homeownership on children. Most of these works use the Panel Study of Income Dynamics, whereas Donald Haurin, Toby Parcel, and Jean Haurin use the National Longitudinal Survey of Youth. These works employ a variety of techniques to deal with the aforementioned selection problem, including a mobility-tenure model by Green and White, sibling comparisons by Aaronson, fixed effects by Haurin, Parcel, and Haurin, and various instrumentation strategies by Boehm and Schlottmann, Aaronson, Harkness and Newman, and George Galster and others.151 Although all employ a variety of controls that differ in their comprehensiveness, they consistently provide support for a direct, independent, nontrivial relationship between homeownership and several outcomes for children and young adults, controlling for residential stability and wealth. Specifically, this set of methodologically robust studies consistently finds a positive impact of homeownership status on: 147. Green and White (1997). 148. Aaronson (2000); Harkness and Newman (2002). 149. Boyle (2002). 150. Barker and Miller (2005). 151. The Panel Study of Income Dynamics is used by Green and White (1997); Boehm and Schlottmann (1999); Aaronson (2000); Harkness and Newman (2002, 2003); Galster and others (forthcoming). The National Longitudinal Survey of Youth is used in Haurin, Parcel, and Haurin (2002a, 2002b).

86

George C. Galster and Anna M. Santiago

—Early childhood cognitive and social development152 —Educational attainment153 —Teen childbearing154 —Earnings and welfare usage, both directly and indirectly, through the effects of homeownership on education155 —Buying a home as a young adult156 Moreover, Green and White and Harkness and Newman found stronger effects for children from lower-income households; Harkness and Newman also found that this is true even in unstable neighborhood contexts.157 All this bodes well for the conventional wisdom for expanding homeownership to lowerincome households.

What Are the Primary Barriers to Attaining and Sustaining Homeownership for Low-Income Households? As a foundation for analyzing programmatic initiatives to enhance and support the homeownership opportunities for low-income households, it is important to establish what research has identified as the major impediments to such opportunities. Michael Collins and Jeffrey Lubell separately provided comprehensive descriptions of the barriers to attaining homeownership:158 —Low and unstable incomes relative to the costs of periodic mortgage payments, insurance, and property taxes159 —Savings and wealth that are inadequate to meet minimum down payment requirements and to stay under the ceiling of total debt-to-income ratios160 —Weak credit ratings that lead to higher-priced mortgages or outright denial of mortgage credit161 —Inadequate information on the part of prospective home buyers concerning how to buy a home, repair their credit, obtain the least-expensive mortgage, and so on162 152. Haurin, Parcel, and Haurin (2002a, 2002b). 153. Green and White (1997); Boehm and Schlottmann (1999); Aaronson (2000); Harkness and Newman (2002, 2003); Galster and others (forthcoming). 154. Green and White (1997); Harkness and Newman (2002, 2003). 155. See Harkness and Newman (2002, 2003) for direct effects; see Boehm and Schlottmann (1999); Galster and others (forthcoming) for indirect ones. 156. Boehm and Schlottmann (1999); Galster and others (forthcoming). 157. Green and White (1997); Harkness and Newman (2003). 158. Collins (2004b); Lubell (2005). 159. Haurin, Hendershott, and Wachter (1997). 160. Listokin and others (2002). 161. Rosenthal (2002). 162. Collins and Dylla (2001).

Low-Income Homeownership

87

—Discrimination in the for-sale housing and mortgage markets directed against low-income minority borrowers163 —High costs of entry-level homes due to a variety of restrictive building codes, environmental regulations, and land use policies164 There have been numerous opinion polls and descriptive studies of the characteristics of low-income and minority households related to financial and home buying literacy, wealth, income, and credit histories, for example.165 Unfortunately, one must make a leap of faith from these studies regarding how these characteristics translate into differentials in homeownership rates. Only a few rigorous empirical studies have attempted to quantify which of the above constraints constitute the largest impediments.166 The work of Quercia and others and David Listokin and others concluded that the savings and wealth constraint is most important, although they differed in their estimates of how much lowincome homeownership rates might increase were this constraint to be relaxed. Quercia and others concluded that the income constraint (mortgage payment– to-income ratio) is next-most critical (and roughly half the size of the savings and wealth effect), though Listokin and others found that relaxing mortgage underwriting will have virtually no impact on the chances of lower-income households to buy a home.167 Neither of the prior works viewed interest rates as a serious constraint, which is supported by the statistical study of Gary Painter and Christian Redfearn.168 Rosenthal’s analysis does not disentangle the contributions made by the first three constraints above but concludes that they represent a substantial barrier in combination.169 George Galster, Laudan Aron, and William Reeder examined the combined effect of inadequate information, discrimination, and inadequate housing supply and concluded that, collectively for lower-income households, they represent roughly as important a barrier as do savings and wealth.170 Unfortunately, each component in the bundle of constraints analyzed by Rosenthal and by Galster, Aron, and Reeder is appropriately 163. Turner and Skidmore (1999). 164. Advisory Commission on Regulatory Barriers to Affordable Housing (1991); Collins, Crowe, and Carliner (2002). 165. For reviews, see Cortes and others (2006); Herbert and others (2005); Herbert and Belsky (2006). 166. For a review, see Herbert and others (2005). They make an important methodological distinction within this literature between studies employing “synthetic underwriting,” such as Listokin and others (2002), and those using “constrained tenure choice models,” such as Galster, Aron, and Reeder (1999); Rosenthal (2002); Quercia and others (2002). The former tend to produce biased estimates of the severity of constraints but nevertheless can produce unbiased relative rankings of constraints. 167. Quercia and others (2002); Listokin and others (2002). 168. Painter and Redfearn (2002). 169. Rosenthal (2002). 170. Galster, Aron, and Reeder (1999).

88

George C. Galster and Anna M. Santiago

confronted by a distinctive set of programmatic approaches, so it is difficult to draw pointed policy implications from both works. It seems clearer, however, that low-income minority home seekers face additional, illegal barriers associated with discrimination because of their racial and ethnic status. Thorough reviews and analyses of the evidence indicate that discrimination against minorities in the home mortgage lending process occurs in many forms and at nontrivial incidences.171 Before loan applications are filed, minorities often receive less assistance and information from loan officers and are quoted higher interest rates.172 After applications are submitted, minorities are more likely to be denied than are equally qualified Whites.173 This discrimination in mortgage markets may be abetted by discrimination in the home sales market, as real estate agents also give less information about housing finance to both Black and Latino home seekers. Moreover, they often steer them through their actions and commentary to neighborhoods that have higher proportions of minority and lower-income residents, characteristics that have been associated with lower home appreciation rates.174 As for barriers to the continuation of homeownership, the sobering fact underpinning the discussion is that these barriers may be substantial inasmuch as roughly half of low-income homeowners return to renter status within five years of buying, as compared with at most a third of high-income buyers.175 Haurin and Rosenthal have been the only ones to econometrically model the transitions of low-income owners out of homeownership. They found that this is more likely to occur when the owner is not married; is younger, poorly educated, and Black; and has a larger family and falling income. It is also more likely in contexts in which home values are falling and mortgage interest rates and unemployment rates are high.176 Unfortunately, these correlations mask considerable detail about the particular stressors faced by these homeowners and thus provide only the broadest policy guidance. A few surveys of recent, low-income home buyers have looked at what these stressors might be.177 Generally, the most frequently cited concern of the lowincome homeowner is home maintenance and repairs, particularly in terms of the ability to address any unexpected, emergency home repairs. Concerns about 171. Turner and Skidmore (1999); Ross and Yinger (2002). 172. Turner and others (2002). 173. Ross and Yinger (2002). 174. See Galster and Godfrey (2005) with respect to targeting minority and low-income residents and Flippen (2004) for the association between neighborhoods with high rates of minority and lower-income residents and lower home appreciation rates. 175. Reid (2004); Haurin and Rosenthal (2005a, 2005b). 176. Haurin and Rosenthal (2005a, 2005b). 177. Mitchell and Warren (1998); Rohe and Quercia (2003); Santiago and Galster (2006).

Low-Income Homeownership

89

high or rising utility costs and about their ability to keep up with mortgage payments are the next-most frequently cited. Moreover, we have found that FFHO home buyers during the first few years after home purchase often acquire a significant level of consumer debt not connected with home repairs or improvements. Thus the same features of income and savings that serve as barriers to attaining homeownership persist as barriers to sustaining ownership. In addition, a rising concern in recent years has been that predatory lenders may exploit low-income homeowners and, in the worst scenarios, induce them to undertake such onerous financing schemes that their home equity is stripped and they are forced to default.178 Several studies have employed various methods for estimating how much wealth is siphoned off (presumably from lowerincome homeowners in the main) by predatory lenders. Elizabeth Renuart estimated, for example, that the actions of predatory lenders annually cost consumers $2.1 billion in equity stripping, $1.8 billion in exorbitant up-front fees, $2.3 billion in prepayment penalties, and $2.9 billion in rate-risk disparities.179 Christopher Richardson estimated that households refinancing through the subprime market would retain from $7.6 billion to $9.5 billion more equity were it not for predatory lenders.180 There have been no multivariate statistical studies that have estimated increases in the probability of foreclosure relative to the receipt of a predatory (as opposed to subprime) loan, though descriptive statistics suggest that a high proportion of subprime loans that foreclose have predatory-like characteristics.181 Certainly, defaults can ensue when unexpected major repairs, utility bills, income losses, or predatory refinance lending occurs. However, the observed rates of default (as discussed above) are insufficient to explain the exceptional attrition of half of low-income homeowners over five years. This suggests that the financial strains associated with homeownership may often create such tenuous family budgets or psychological stresses or both that many opt out of this tenure, sell their home, and do not buy again immediately. By implication, sustaining low-income homeownership will require more than a myopic focus on default prevention. It will require a more comprehensive view of the financial position of the homeowner and the ambient levels of stress associated with this form of tenure.

178. Willis (2006); Goldstein (2006). For a review, see Zimmerman (2006). 179. Renuart (2004, p. 487). 180. Richardson (2002). 181. Stock (2001); Quercia, Stegman, and Davis (2005).

90

George C. Galster and Anna M. Santiago

What Have We Tried to Expand and Sustain Homeownership among Low-Income Households, and What Has Worked? Federal, state, and local governments, as well as nonprofit and philanthropic organizations, have been engaged for many years in a dizzying array of initiatives designed to lower the aforementioned barriers to attaining and sustaining lowincome homeownership. Primary examples (either tried or proposed) related to each constraint include —Low and unstable incomes: HUD housing vouchers applied to home purchase; subsidized home construction and rehabilitation through Community Development Block Grants, the HOME Investment Partnerships program, or local sources of financing; requirements for Fannie Mae and Freddie Mac to purchase loans originated to low-income borrowers; employer-assisted housing subsidies; lease-purchase programs; state and local housing trust funds, mortgage revenue bonds, and other subsidy-financing devices; self-help housing —Inadequate savings and wealth: no or low down payment mortgages, riskbased mortgage pricing, down payment assistance grants and loans, assetbuilding tools such as Individual Development Accounts, public housing authority–run programs for income enhancement and wealth accumulation (such as the Family Self-Sufficiency program), foreclosure prevention services, major home repair insurance —Weak credit ratings: credit counseling and repair services, schemes whereby rent and other payment histories bolster credit ratings —Inadequate information: financial literacy education, homeownership preand postpurchase counseling, programs to expand participation in the banking sector —Discrimination: federal resources for fair housing and lending education and complaint investigation, paired testing used as enforcement tool, lender regulatory oversight under auspices of Community Reinvestment Act and Equal Credit Opportunity Act, state and local antipredatory lending legislation, Home Mortgage Disclosure Act (HMDA) reporting —Restrictions in housing supply: community land trusts, grant allocation incentives for localities to reduce regulatory barriers, inclusionary zoning, manufactured housing and other reduced-cost construction techniques182 Despite this wealth of policies, there are very few evaluation studies that employ appropriate methodologies for plausibly assessing their degree of independent, causal impact. Virtually all assessments of programs designed to promote low-income homeownership have merely described programmatic opera182. For overviews, see Collins (2004b); Lubell (2005); Herbert and Belsky (2006); Cortes and others (2006).

Low-Income Homeownership

91

tions and reported outcomes for participants.183 They have not tried to compare these outcomes against some carefully specified counterfactual ones.184 As a result, skepticism about observed outcomes being tainted by nonprogrammatic selection effects (that is, unobserved characteristics of participants) or other uncontrolled forces is well founded. Sadly, this means that for the vast majority of policies listed above we cannot answer the question “What works?” with any assurance. The only partial exceptions to this generalization come in two areas: liberalization of mortgage terms and homeownership education and counseling.

Studies Related to Liberalization of Mortgage Terms Three studies that differ in approach and conclusions look at mortgage terms and focus on Federal Housing Administration lending to low-income borrowers. FHA loans offer low down payment requirements and offer more flexible underwriting terms regarding required debt and income ratios. Thus their greater availability should, in principle, allow more low-income buyers into the market by relaxing both down payment and income constraints. John Goodman and Joseph Nichols, however, concluded that FHA loans may at best accelerate the transition from renting into homeowning and do not significantly increase the aggregate homeownership rates beyond that.185 Zeynep Onder’s study produced a somewhat more sanguine conclusion that greater FHA activity in a metropolitan area is positively related to homeownership rates, but the size of the implied effect is small: a 1 percent higher share in the FHA market is associated with a 0.02 percent higher homeownership rate.186 A considerably stronger relationship is uncovered in the study by Albert Monroe, however, who used an approach that avoids the potential selection biases of the prior works. Monroe found that a 1 percent higher share of homes rendered affordable because of FHA mortgage availability is associated with a 0.2 percent increase in homeownership rates, and this relationship is seven times greater for Blacks and those with a high school diploma or less.187 To the extent that FHA loans in the periods under study above required substantially lower down payments than those required by conventional mortgages, Monroe’s findings are consistent with the aforementioned evidence on the importance of the down payment constraint for lower-income home seekers. Another small set of studies has examined whether intensified efforts by the government-sponsored enterprises (GSEs, that is, Fannie Mae and Freddie Mac) 183. Turnham and others (2004); Locke and others (2006); Quercia, Gorham, and Rohe (2006), for example. 184. Quercia and Wachter (1996). 185. Goodman and Nichols (1997). 186. Onder (2002). 187. Monroe (2001).

92

George C. Galster and Anna M. Santiago

to purchase mortgages originated to lower-income individuals and neighborhoods (in response to HUD regulatory requirements) affect their homeownership chances. In two studies, Brent Ambrose and Thomas Thibodeau found that the gap in homeownership rates between low- and high-income households dropped in Metropolitan Statistical Areas (MSAs) where GSEs expanded their purchases more strongly, though in the national sample they appeared to have no impact on the growth of low-income homeownership rates from 1991 to 1997.188 This combination of results seems implausible; perhaps the best conclusion is that results for low-income groups are inconclusive.189 Xudong An, Raphael Bostic, and Yongheng Deng and Bostic and Stuart Gabriel reached even less sanguine conclusions.190 The An, Bostic, and Deng study of California found only limited evidence of any impact of the GSEs’ affordable housing purchasing regulations on homeownership rates. The Bostic and Gabriel nationwide companion study found no relationship between the intensity of GSE purchasing rates in a census tract in 1995 and changes in the homeownership rate there from 1990 to 2000, controlling for changes in metropolitan economic conditions and a variety of tract conditions in 1990. These studies yielded similar implications as the aforementioned studies, showing relatively little importance of interest rates as a barrier to homeownership. Ironically, the combined policies of offering FHA loans and expanded GSE mortgage purchasing requirements may actually work at cross-purposes. An, Bostic, and Deng argued that more aggressive purchasing activity by the GSEs in response to the HUD purchasing goals has merely substituted for some FHA lending that would have occurred in these areas previously.191 They demonstrated that conventional and FHA home purchase loan products compete for many borrowers (especially those with less-than-perfect credit histories), and that GSEs generally do not purchase FHA loans. Thus, they argued, if GSEs pursue more aggressive purchasing of conventional loans, some former FHA borrowers with the best credit may be induced by increasingly liquid primary lenders to obtain conventional products instead. In response, FHA will need to apply stricter underwriting standards to keep the overall risk profile of its portfolio acceptable, whereupon they will lower their loan origination volume. This induced FHA response blunts the impact of the increased liquidity provided by the GSEs and creates a zero sum scenario, they hypothesized. There have been two rigorous studies that offer estimates of the impact from programs that provide down payment assistance. Gary Engelhardt econometri188. Ambrose and Thibodeau (2004); Thibodeau and Ambrose (2002). 189. Herbert and others (2005). 190. An, Bostic, and Deng (2005); Bostic and Gabriel (2006). 191. An, Bostic, and Deng (2005).

Low-Income Homeownership

93

cally estimated a tenure choice model for Canadian households to see how the probability of choosing homeownership changed after the Registered Home Ownership Savings Plan was eliminated in 1985.192 This plan offered an annual tax reduction of up to $1,000 per individual (with a $10,000 cumulative limit) on savings committed to the purchase of a first home. He found that the elimination of this savings plan significantly reduced the transition into homeownership, though it is unclear whether the effect would be as strong for those with lower marginal tax rates and whether the subsidy only affected the timing, not the overall chances, of moving into homeownership. Christopher Herbert and Winnie Tsen provided a powerful empirical estimate of the potential of down payment assistance to spur low-income homeownership.193 Although it is not an impact evaluation of an extant program, it can be interpreted as an evaluation of one that simply provides a direct down payment grant, analogous to, for example, the recently enacted American Dream Downpayment Initiative of 2003. Using the 1996 Survey of Income and Program Participation, they estimated a statistical model that relates individual renters’ income, wealth, and other characteristics over time to the probability of the renter’s transitioning into homeownership. Remarkably, they found that liquid wealth has a decreasing positive impact on the probability of this transition. Using the estimated model parameters, they then simulated how many more low-income renters could have made the leap into homeownership during the 1996–2000 period with different hypothetical levels of down payment subsidies. The alternative subsidy amounts and simulated increases over the baseline (no subsidy) transition rates are $1,000, 19 percent; $5,000, 34 percent; and $10,000, 41 percent.

Studies Related to Homeownership Education and Counseling There are two broad categories of education and counseling programs: prepurchase and postpurchase.194 Prepurchase counseling is designed to remove barriers to obtaining information related to basic financial literacy, homeowning, and the home buying process, all of which may be hindering low-income households. There have been numerous studies that have described the wide variety of homeownership education and counseling programs and provided descriptive statistics of outcomes of participants.195 However, there are remarkably few evaluation stud192. Engelhardt (1997). 193. Herbert and Tsen (2005). 194. For histories of these programs, see McCarthy and Quercia (2000); Quercia and Wachter (1996). 195. Examples include McCarthy and Quercia (2000); Wiranowski (2003); Collins (2004b); Turnham and others (2004); Quercia, Cowan, and Moreno (2005); Lubell (2005); Quercia, Gorham, and Rohe (2006).

94

George C. Galster and Anna M. Santiago

ies that have tried to assess the degree to which such counseling actually enhances the willingness and ability of such households to buy a home compared with an otherwise comparable group of low-income households who do not receive counseling. Two unpublished studies from the 1970s summarized by Leslie Strauss and Sue Phillips claimed that counseling improved home purchase rates, but in one case the counseling was offered in conjunction with substantial financial incentives to buy a home, and in the other there was significant prescreening of participants in counseling.196 The strongest methodological effort here was provided by Judith Feins, Dixon Bain, and John Kirlin, who investigated counseling in conjunction with the HUD Section 235 homeownership program.197 They divided participants into three groups: those who received written information about the home buying process only—the putative control group; those who received written information plus group counseling; and those who received written information, one-on-one counseling, and advocacy activities like home inspections and attendance at closing. The only statistically significant difference among the three groups was that those receiving written information and the group counseling were somewhat less likely to purchase within nine months of counseling than the other two groups, perhaps because the group process proved intimidating to participants. That the group getting the most intensive, individualized counseling performed no better than those perfunctorily given literature suggests either that the counseling provides little value added or that its effects only become manifested after the first nine months of the homeowning experience. The only other aspect of prepurchasing counseling effectiveness that has been systematically probed is the degree to which it reduces the risk of mortgage delinquency and default. Valentina Hatarska, Claudio Gonzalez-Vega, and David Dobos and Hartarska and Gonzalez-Vega found that those counseled under a Midwest-based program evinced lower defaults, controlling for a wide variety of personal, life event, and mortgage characteristics.198 However, the results may have been produced by the rigorous screening into the program of only those most likely to be successful homeowners, not the counseling per se. Hirad and Zorn provided a comprehensive and more methodologically sophisticated study of the effectiveness of prepurchase counseling in reducing delinquency rates in Freddie Mac’s affordable loan portfolio. They found that, all else being equal, several types of counseling decrease the ninety-day delinquency rate: provision of individual counseling lowers it by 34 percent, classroom counseling by 26 percent, and home study by 21 percent.199 The type of organization 196. Strauss and Phillips (1997). 197. Feins, Bain, and Kirlin (1980). 198. Hatarska, Gonzalez-Vega, and Dobos (2002); Hartarska and Gonzalez-Vega (2005). 199. Hirad and Zorn (2002, p. 164).

Low-Income Homeownership

95

delivering the counseling had no independent effect. These basic results appear robust to methodological adjustments for participant selection effects. Thus both of the studies providing the most convincing controls for selection (Feins, Bain, and Kirlin and Hirad and Zorn) come to a similar conclusion about the relative superiority of one-on-one counseling to group prepurchase counseling in raising the likelihood of purchase and lowering the likelihood of delinquency, respectively. In the realm of postpurchase programs, there are two further categories: those that try to prevent new owners from falling into arrears on their mortgage payments and those that try to help those that have.200 Programs in the former category typically try to provide additional instruction in budgeting, home maintenance and repair techniques, energy conservation, and avoidance of predatory lenders, but they may in some cases provide emergency financing in the face of unexpected major home repair needs. Programs in the latter category help clients develop mortgage workout or refinancing plans, often with close collaboration with major lenders. Early evaluation work in this field has been uniformly characterized as “inconclusive” because of severe methodological shortcomings and variance in findings.201 There seems, however, to be a current consensus that programs designed to prevent foreclosure are at least somewhat efficacious, given that many major mortgage lenders partner with and financially support community-based organizations providing such assistance. This impression is bolstered by the favorable outcomes cited by program directors and expert evaluators.202 Similarly, Roberto Quercia, Spencer Cowan, and Ana Moreno’s cost-effectiveness analysis of the participants in the Mortgage Foreclosure Prevention Program in Minnesota found that time to resolution and rate of recidivism both compared favorably with national norms.203 Unfortunately, no studies related to postpurchase counseling have attempted to quantify the independent impacts of extant programs convincingly because of data limitations and methods that do not control for selection effects.204 Conclusions about impacts from postpurchase counseling studies cited above and the programmatic recommendations and best practices that have been advanced from them should thus be viewed with caution.205

200. Rohe and Quercia (2003); Wiranowski (2003); Lubell (2005); Quercia, Gorham, and Rohe (2006). 201. Wiranowski (2003); Quercia, Gorham, and Rohe (2006). 202. Rohe and Quercia (2003); Quercia, Gorham, and Rohe (2006). 203. Quercia, Cowan, and Moreno (2005). 204. Turnham and others (2004). 205. Wiranowski (2003); Quercia, Gorham, and Rohe (2006).

96 George C. Galster and Anna M. Santiago

Conclusion and Policy Implications We have addressed six questions in this chapter and provide our answers in summary form. —Does Homeownership Typically Enable Low-Income Households to Build Wealth? It can enable them to do so, although this is less likely for households who have lower incomes, hold minority status, stay in their home for only short periods, or buy at an inappropriate point in the housing price cycle. —Does Homeownership Typically Provide Low-Income Households with a Financial Buffer to Weather Financial Setbacks? No. Even those who manage to gain a financial buffer of home equity often spend down that equity by incurring more consumer debt, thereby rendering themselves continuously vulnerable to many housing, mortgage, and labor market shocks and family disruptions and producing a high attrition rate among low-income homeowners. —Does Homeownership for Low-Income Households Typically Involve Terms and Conditions That, When Coupled with Their Other Instabilities, Render a High Probability of Financial Distress? Yes. The terms of the mortgage (especially when predatory lending is involved) often place purchasers in a tenuous financial situation that, coupled with the above challenges, leads to a high rate of default and exits from homeownership even when no defaults occur. —Does Homeownership Provide Benefits to Children in Low-Income Households That Might Lead to Superior Economic Outcomes in the Next Generation? Yes. Studies show economically significant gains in certain behaviors, cognitive skills, and educational attainments of children. —What Are the Primary Barriers to Attaining and Sustaining Homeownership for Low-Income Households? Down payment constraints are typically the overwhelming obstacle to attaining homeownership, but combinations of other constraints are also important. Limited financial reserves, less favorable mortgage terms and reluctance to refinance, and increasing consumer debt render recent low-income homeowners especially vulnerable to interruptions of income or unexpected expenses associated with illness or home repair that make homeownership unsustainable. Discrimination in mortgage and home sales markets impose added constraints on Black and Latino home seekers. —What Mechanisms Have We Tried to Expand and Sustain Homeownership among Low-Income Households, and What Has Worked? There have been numerous initiatives from all levels of government aimed at making more homes more affordable, making mortgage financing available to more customers at more favorable terms regardless of race or ethnicity, and better informing consumers about the processes of home buying and sustaining a home once purchased. Attempts have been made to attack discrimination. Unfortunately, few rigorous program

Low-Income Homeownership

97

impact studies have been conducted on this panoply of efforts; they reveal that down payment assistance subsidies have substantial impacts on helping lowincome households attain homeownership and prepurchase counseling reduces their probability of default, especially if delivered in a one-on-one situation. All of this leads us to the following recommendations for policymakers. We initially recommend that the public sector should be interested in enhancing the opportunities of homeownership for low-income households, given its sizable potential for building wealth under the right circumstances and its demonstrated benefits for children. However, it is also clear from our analysis that the desirability of low-income homeownership as a strategy for growing the middle class is highly contingent. That is, increasing homeownership among low-income households has to be part of a more comprehensive asset building program; if not, desirable results will not be achieved. In particular, policymakers must recognize two overarching findings: —Building and sustaining a modicum of liquid wealth before and after home purchase (instead of using home equity to finance consumer spending, for example) is the key to attaining and sustaining the pattern of homeownership that ultimately proves financially beneficial for the low-income buyer in the long run. —Low-income minority households face special challenges in attaining and sustaining homeownership that are not associated with their household or financial characteristics. This suggests a strategy that embeds enhancing opportunities for low-income homeownership within a more comprehensive suite of synergistic initiatives designed to build income and assets for low-income households, without regard for race and ethnicity. More specifically, we would recommend that homeownership policy be nested within programs that help low-income households simultaneously to raise earnings, reduce consumer indebtedness while increasing savings, budget more effectively, and raise skills to do their own home repairs. In addition, the programmatic suite should include efforts to reduce discrimination against minorities. We think several proposals offer promise here.206 Raising earnings could effectively and swiftly be accomplished by an expansion and deepening of the well-tested Earned Income Tax Credit, so that it provided higher benefits to more working households. Encouraging wealth building could be accomplished by expanding Individual Development Accounts (or an equivalent program targeted to matched savings for use as down payments) and retargeting various forms of federal assistance to better focus on assisting potential first-time home 206. See Dreier and Atlas (1996); Lubell (2005); Collins (2004a, 2004b, 2007); Rohe (2007).

98

George C. Galster and Anna M. Santiago

buyers. This should include replacement of the current IRS mortgage interest and property tax deductibility rules with a refundable, progressive federal tax credit for homeowners and an expansion of Family Self-Sufficiency and Rent Voucher Homeownership program opportunities for recipients of federal housing assistance. For enhancing consumer education concerning buying and owning a home, we should expand the availability of free pre- and postpurchasing counseling programs and publicize their benefits. The former would aim to improve the chances that low-income households would buy homes under the most favorable, feasible circumstances regarding timing, terms, and locations. The latter would aim to help new owners avoid equity erosion through excessive spending and assuming consumer debt or predatory lending, take advantage of favorable refinancing options, build their home repair and improvement skills, and work out situations of mortgage delinquency before they escalate to default. Finally, we must expand the use of paired testing investigations and stiffen penalties for those found to violate the fair housing and fair lending laws, as there is evidence that such action would strengthen their deterrent effect.207

207. Galster (1990); Ross and Galster (2007).

Low-Income Homeownership

99

References Aaronson, Daniel. 2000. “A Note of the Benefits of Homeownership.” Journal of Urban Economics 47, no. 3: 356–69. Advisory Commission on Regulatory Barriers to Affordable Housing. 1991. “Not In My Backyard: Removing Barriers to Affordable Housing.” Washington: U.S. Department of Housing and Urban Development. Ambrose, Brent W., and William N. Goetzmann. 1998. “Risks and Incentives in Underserved Mortgage Markets.” Journal of Housing Economics 7, no. 3: 274–85. Ambrose, Brent W., and Thomas G. Thibodeaux. 2004. “Have the GSE Affordable Housing Goals Increased the Supply of Mortgage Credit?” Regional Science and Urban Economics 34, no. 3: 263–73. An, Xudong, Raphael W. Bostic, and Yongheng Deng. 2005. “Have the Affordable Housing Goals Improved Housing Markets?” University of Southern California, School of Policy, Planning and Development. Apgar, William. 2004. “Rethinking Rental Housing: Expanding the Ability of Rental Housing to Serve as a Pathway to Economic and Social Opportunity.” Working Paper W04-11. Cambridge, Mass.: Harvard University Joint Center for Housing Studies. Apgar, William C. Jr., and Christopher E. Herbert. 2005. “Subprime Lending and Alternative Financial Service Providers: A Literature Review and Empirical Analysis.” Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. Austin, D. Mark, and Yoko Baba. 1990. “Social Determinants of Neighborhood Attachments.” Sociological Spectrum 10, no. 1: 59–78. Barker, David, and Eric Miller. 2005. “Homeownership and Child Welfare.” Paper presented at the 33rd annual Mid-Year Meeting, American Real Estate and Urban Economics Association. Washington, May 31–June 1. Balfour, Danny, and Janet Smith. 1996. “Transforming Lease-Purchase Programs for LowIncome Families.” Journal of Urban Affairs 18, no. 2: 173–88. Belsky, Eric S., and Mark Duda. 2002. “Asset Appreciation, Timing of Purchases and Sales, and Returns to Low-Income Homeownership.” In Low Income Homeownership: Examining the Unexamined Goal, edited by Nicolas P. Retsinas and Eric S.Belsky, pp. 208–38. Brookings. Belsky, Eric S., and Nicholas P. Retsinas. 2005. “New Paths to Building Assets for the Poor.” In Building Assets, Building Credit: Creating Wealth in Low-Income Communities, edited by Nicolas P. Retsinas and Eric S. Belsky, pp. 1–9. Brookings. Belsky, Eric S., Nicolas P. Retsinas, and Mark Duda. 2005. “The Financial Returns to LowIncome Homeownership.” In Chasing the American Dream: New Perspectives on Affordable Homeownership, edited by William M. Rohe and Harry L. Watson, pp. 191–212, Cornell University Press (2007) [Working Paper W05-9. Cambridge, Mass.: Harvard University Joint Center for Housing Studies]. Blank, Rebecca M. 2002. “Evaluating Welfare Reform in the U.S.” Journal of Economic Literature 40, no. 4: 1105–166. Boehm, Thomas P. 2004. Wealth Accumulation and Homeownership: Evidence for Low-Income Households. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research.

100

George C. Galster and Anna M. Santiago

Boehm, Thomas P., and Alan M. Schlottmann. 1999. “Does Home Ownership by Parents Have an Economic Impact on Their Children.” Journal of Housing Economics 8, no. 3: 217–32. ———. 2002. “Housing and Wealth Accumulation: Intergenerational Impacts.” In Low Income Homeownership: Examining the Unexamined Goal, edited by Nicolas P. Retsinas and Eric S.Belsky, pp. 407–26. Brookings. ———. 2004a. “The Dynamics of Race, Income, and Homeownership.” Journal of Urban Economics 55, no. 1: 113–30. ———. 2004b. Is Manufactured Housing a Good Alternative for Low-Income Families? Evidence from the American Housing Survey. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. ———. 2004c. Wealth Accumulation and Homeownership: Evidence for Low-Income Households? Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. Boehm, Thomas P., Paul D. Thistle, and Alan Schlottmann. 2006. “Rates and Race: An Analysis of Racial Disparities in Mortgage Rates.” Housing Policy Debate 17, no. 1: 109–49. Bostic, Raphael W., Paul S. Calem, and Susan M. Wachter. 2004. “Hitting the Wall: Credit as an Impediment to Homeownership.” Working Paper 04-5. Cambridge, Mass.: Harvard University Joint Center for Housing Studies (February). Bostic, Raphael W., and Stuart A. Gabriel. 2006. “Do the GSEs Matter to Low-Income Housing Markets?” Journal of Urban Economics 59, no. 3: 458–75. Boyle, Michael. 2002. “Home Ownership and the Emotional and Behavioral Problems of Children and Youth.” Child Development 73, no 3: 883–92. Buerkle, Karla K. 1997. “Mobile Children and Families: Qualitative and Quantitative Explorations of the Meaning and Impact of Residential Mobility and School Changes.” Dissertations Abstracts International, A: The Humanities and Social Sciences 19, no. 2: 161–74. Cairney, John, and Michael Boyle. 2004. “Home Ownership, Mortgages and Psychological Distress.” Housing Studies 19, no. 2: 161–74. Canner, Glenn, Karen Dynan, and Wayne Passmore. 2002. “Mortgage Refinancing in 2001 and Early 2002.” Federal Reserve Bulletin (December): 469–81. Carr, James H., and Jenny Schuetz. 2001. Financial Services in Distressed Communities: Framing the Issues, Finding Solutions. Washington: Fannie Mae Foundation. Coleman, James S. 1988. “Social Capital and the Creation of Human Capital.” American Journal of Sociology 94 (Supplement): S95–S120. ———. 1990. Foundations of Social Theory. Harvard University Press. Collins, J. Michael. 2002. Pursuing the American Dream: Homeownership and the Role of Federal Housing Policy. Report prepared for the Millennial Housing Commission. Washington (January). ———.2004a. “An Assessment of Policies to Promote Financial Savings and Asset Building in the U.S.” Ithaca, N.Y.: PolicyLab Consulting Group. ———. 2004b. “Expanding the American Dream: A Homeownership Guide for Grantmakers.” Grantmakers Supporting Community Change Public Policy Paper. Washington: Neighborhood Funders Group (Fall). ———. 2007. “An Assessment of Policies to Promote Financial Savings and Asset Building in the U.S.” Ithaca, N.Y.: PolicyLab Consulting Group. (www.policylabconsulting.com/).

Low-Income Homeownership

101

Collins, J. Michael, David Crowe, and Michael Carliner. 2002. “Supply-Side Constraints on Low-Income Homeownership.” In Low Income Homeownership: Examining the Unexamined Goal, edited by Nicolas P. Retsinas and Eric S.Belsky, pp. 175–200. Brookings. Collins, Michael J., and Doug Dylla. 2001. Mind the Gap: Issues in Overcoming the Information, Income, Wealth and Supply Gaps Facing Potential Buyers of Affordable Homes. Home Ownership Summit Research Series. Washington: LISC Center for Home Ownership (October). Cortes, Alvaro, and others. 2006. Improving Homeownership Opportunities for Hispanic Families: A Review of the Literature. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. Coulson, N. Edward, and Lynn M. Fisher. 2002. “Tenure Choice and Labour Market Outcomes.” Housing Studies 17, no. 1 (January): 35–49. Cox, Kevin. 1982. “Housing Tenure and Neighborhood Activism.” Urban Affairs Quarterly 18, no. 1: 107–29. Cutts, Amy Crews, and Richard K. Green. 2004. “Innovative Servicing Technology: Smart Enough to Keep People in Their Houses?” Working Paper 04-03. McLean, Va.: Freddie Mac (July). Cutts, Amy Crews, and Robert A. Van Order. 2005. “On the Economics of Subprime Lending.” Journal of Real Estate Finance and Economics 30, no. 2: 167–96. Danis, Michelle A., and Anthony N. Pennington-Cross. 2005. “The Delinquency of Subprime Mortgages.” Working Paper Series 2005-022A. Federal Reserve Bank of St. Louis. Deng, Yongheng, John M. Quigley, and Robert Van Order. 1996. “Mortgage Default and Low Downpayment Loans: The Costs of Public Subsidy.” Regional Science and Urban Economics 26, nos. 3–4 (January): 263–85. DeWit, David J. 1998. “Frequent Childhood Geographic Relocation: Its Impact on Drug Use Initiation and the Development of Alcohol and Other Drug-Related Problems among Adolescents and Young Adults.” Addictive Behaviors 23, no. 5: 623–34. Di, Zhu Xiao, Yi Yang, and Xiadong Liu. 2003. “The Importance of Housing to the Accumulation of Household Net Wealth.” Cambridge, Mass.: Harvard University Joint Center for Housing Studies. Dietz, Robert D., and Donald R. Haurin. 2003. “The Social and Private Micro-Level Consequences of Homeownership.” Journal of Urban Economics 54, no. 6: 401–50. DiPasquale, Denise, and Edward L. Glaeser. 1999. “Incentives and Social Capital: Are Homeowners Better Citizens?” Journal of Urban Economics 45, no. 2: 354–84. Dreier, Peter and Atlas, John. 1996. “U.S. Housing Policy at the Crossroads: Rebuilding the Housing Constituency.” Journal of Urban Affairs 18, no. 4: 341-70. Duda, Mark, and Eric S. Belsky. 2000. “Asset Appreciation, Timing of Purchases and Sales, and Returns to Low-Income Homeownership.” Paper prepared for the Symposium on Low-Income Homeownership as an Asset-Building Strategy. Harvard University Joint Center for Housing Studies, November 14–15. Also as Working Paper LIHO-01.6 (2001). Engelhardt, Gary V. 1997. “Do Targeted Savings Incentives for Homeownership Work? The Canadian Experience.” Journal of Housing Research 8, no. 2: 225–48. Feins, Judith D., Dixon Bain, and John A. Kirlin. 1980. Results of the Prepurchase Homeownership Counseling Demonstration: Final Report. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. Fishbein, Allen, and Patrick Woodall. 2005. “Subprime Cities: Patterns of Geographic Disparity in Subprime Lending.” Washington: Consumer Federation of America (September 8).

102

George C. Galster and Anna M. Santiago

Flippen, Chenoa. 2004. “Unequal Returns to Housing Investment? A Study of Real Housing Appreciation among Black, White and Hispanic Households.” Social Forces 82, no. 4: 1523–551. Ford, Janet, and Deborah Quilgars. 2001. “Failing Home Owners? The Effectiveness of Public and Private Safety-nets.” Housing Studies 16, no. 2: 147–62. Galster, George C. 1983. “Empirical Evidence on Cross-Tenure Differences in Home Maintenance and Conditions.” Land Economics 59, no. 1: 107–13. ———. 1987. Homeowners and Neighborhood Reinvestment. Duke University Press. ———. 1990. “The Great Misapprehension: Federal Fair Housing Policy in the Eighties.” In Building Foundations: Housing and Federal Policy, edited by Denise DiPasquale and Langley C. Keyes, pp. 137–57. University of Pennsylvania Press. Galster, George C., Laudan Aron, and William Reeder. 1999. “Encouraging Mortgage Lending in ‘Underserved’ Areas: The Potential for Expanding Home Ownership in the US.” Housing Studies 14, no. 6: 777–801. Galster, George C., and Jennifer Danielle. 1996. “Housing.” In Reality and Research: Social Science and U.S. Urban Policy Since 1960, edited by George Galster, pp. 85–112. Washington: Urban Institute. Galster, George C., and Erin Godfrey. 2005. “By Words and Deeds: Racial Steering by Real Estate Agents in the U.S. in 2000.” Journal of the American Planning Association 71, no. 3: 1–19. Galster, George C., and others. Forthcoming. “The Impacts of Parental Homeownership on Children’s Outcomes During Early Adulthood.” Housing Policy Debate. Goetzmann, William N., and Matthew Spiegel. 2002. “The Policy Implications of Portfolio Choice in Underserved Mortgage Markets.” In Low-Income Homeownership: Examining the Unexamined Goal, edited by Nicholas P. Retsinas and Eric S. Belsky, pp. 257–74. Brookings. Goldstein, Ira J. 2006. Lost Values: A Study of Predatory Lending in Philadelphia. Philadelphia: The Reinvestment Fund. Goodman, John L., and Joseph B. Nichols. 1997. “Does FHA Increase Home Ownership or Just Accelerate It?” Journal of Housing Economics 6, no. 2: 184–202. Green, Richard K., and Patric H. Hendershott. 1999. Home Ownership and Unemployment in the U.S. Working Paper. Washington: National Multifamily Housing Council (December). Green, Richard K., and Michelle J. White. 1997. “Measuring the Benefits of Homeowning: Effects on Children.” Journal of Urban Economics 41, no. 3: 441–61. Harkness, Joseph M., and Sandra J. Newman. 2002. “Homeownership for the Poor in Distressed Neighborhoods: Does This Make Sense?” Housing Policy Debate 13, no. 3: 597–630. ———. 2003. “Effects of Homeownership on Children: The Role of Neighborhood Characteristics and Family Income.” Economic Policy Review 9, no. 2: 87–107. Hartarska, Valentina, and Claudio Gonzalez-Vega. 2005. “Credit Counseling and Mortgage Termination by Low-Income Households.” Journal of Real Estate Finance and Economics 30, no. 3: 227–43. Hartaska, Valentina, Claudio Gonzalez-Vega, and David Dobos. 2002. “Credit Counseling and the Incidence of Default on Housing Loans by Low-Income Households.” Ohio State University, Rural Finance Program (February).

Low-Income Homeownership

103

Haurin, Donald R., Robert D. Dietz, and Bruce A. Weinberg. 2002. “The Impact of Neighborhood Homeownership Rates: A Review of the Theoretical and Empirical Literature.” Journal of Housing Research 13, no. 1: 119–51. Haurin, Donald R., Patric H. Hendershott, and David C. Ling. 1988. “Home Ownership Rates of Married Couples: An Econometric Investigation.” Housing Finance Review 7, no. 2: 85–108. Haurin, Donald R., Patric H. Hendershott, and Susan M. Wachter. 1997. “Borrowing Constraints and the Tenure Choice of Young Households.” Journal of Housing Research 8, no. 2: 137-54. Haurin, Donald R., Toby L. Parcel, and R. Jean Haurin. 2002a. “Does Home Ownership Affect Child Outcomes?” Real Estate Economics 30, no. 4: 635–66. ———. 2002b. “Impact of Home Ownership on Child Outcomes.” In Low Income Homeownership: Examining the Unexamined Goal, edited by Nicolas P. Retsinas and Eric S.Belsky, pp. 427–46. Brookings. Haurin, Donald R., and Stuart S. Rosenthal. 2005a. The Growth Earnings of Low-Income Households and the Sensitivity of Their Homeownership Choices to Economic and SocioDemographic Shocks. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development & Research. ———. 2005b. The Impact of House Price Appreciation on Portfolio Composition and Savings. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. Herbert, Christopher E., and Eric S. Belsky. 2006. The Homeownership Experience of LowIncome and Minority Families: A Synthesis and Review of the Literature. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research (February). Herbert, Christopher E., and Winnie Tsen. 2005. The Potential of Downpayment Assistance for Increasing Homeownership among Minority and Low-Income Households. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. Herbert, Christopher E., and others. 2005. Homeownership Gaps among Low-Income and Minority Borrowers and Neighborhoods. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. Hirad, Abdighani, and Peter M. Zorn. 2001. “A Little Knowledge Is a Good Thing: Empirical Evidence of the Effectiveness of Pre-Purchase Homeownership Counseling.” Working Paper. McLean, Va.: Freddie Mac. (www.freddiemac.com/corporate/reports/pdf/homebuyers_study.pdf ). ———. 2002. “Prepurchase Homeownership Counseling: A Little Knowledge is a Good Thing.” In Low-income Homeownership: Examining the Unexamined Goal, edited by Nicholas P. Retsinas and Eric S. Belsky, pp.146–74. Brookings. Hoff, Karla, and Arijit Sen. 2005. “Homeownership, Community Interactions, and Segregation.” American Economic Review 95, no. 4: 1167–189. HUD (U.S. Department of Housing and Urban Development). 2002. Barriers to Minority Homeownership. Washington. Huffines, Margaret E. 2003. “A Study of the Impact of Mobility on Student Social Adjustment.” Dissertation Abstracts International, A: The Humanities and Social Sciences 64, no. 4-A: 1184.

104

George C. Galster and Anna M. Santiago

Hunter, Albert. 1975. “The Loss of Community: An Empirical Test through Replication.” American Sociological Review 40, no. 5: 537–51. Immergluck, Daniel. 2008. “From the Subprime to the Exotic: Expanded Mortgage Market Risk and Implications for Metropolitan Areas and Neighborhoods.” Journal of the American Planning Association 74, no. 1: 59–76. Immergluck, Daniel, and Geoff Smith. 2004. “Risky Business: An Econometric Analysis of the Relationship between Subprime Lending and Neighborhood Foreclosures.” Chicago: Woodstock Institute. Jeffers, Leo, and Jean Dobos. 1984. “Communication and Neighborhood Mobilization.” Urban Affairs Quarterly 20, no. 1: 97–112. Lee, Barrett A., R. S. Oropesa, and James W. Kanan. 1994. “Neighborhood Context and Residential Mobility.” Demography 31, no. 2: 249–70. Listokin, David, and others. 2002. The Potential and Limitations of Mortgage Innovation in Fostering Homeownership in the United States. Washington: Fannie Mae Foundation. Locke, Gretchen, and others. 2006. Voucher Homeownership Study: Volume 1 Cross-site Analysis. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. Louie, Josephine, Eric S. Belsky, and Nancy McArdle. 1998. “The Housing Needs of LowerIncome Homeowners.” Working Paper W98-8. Cambridge, Mass.: Harvard University, Joint Center for Housing Studies (August). Lubell, Jeffrey M. 2005. Strengthening the Ladder for Sustainable Homeownership. Washington: National Housing Conference and Annie E. Casey Foundation. Masnick, George S., Zhu Xiao Di, and Eric S. Belsky. 2005. “Emerging Cohort Trends in Housing Debt and Home Equity.” Working Paper W05-1. Cambridge, Mass.: Harvard University, Joint Center for Housing Studies (January). Mayer, Neil S. 1981. “Rehabilitation Decisions in Rental Housing.” Journal of Urban Economics 10, no. 1: 76–94. McCarthy, George W., and Roberto G. Quercia. 2000. Bridging the Gap between Supply and Demand: The Evolution of the Homeownership, Education and Counseling Industry. Washington: Research Institute for Housing America. McCarthy, George W., Shannon Van Zandt, and William M. Rohe. 2001. The Economic Benefits and Costs of Homeownership. Washington: Research Institute for Housing America. Meyer, Peter B., Jerry Yeager, and Michael Burayidi. 1994. “Institutional Myopia and Policy Distortions: The Promotion of Homeownership for the Poor.” Journal of Economic Issues 28 (June): 567–76. Mills, Edwin S. 1990. “Housing Tenure Choice.” Journal of Real Estate Finance and Economics 3, no. 4: 459–68. Mitchell, Maxine, and S. Paige Warren. 1998. Making Home Ownership a Reality: Survey of Habitat for Humanity International (HFHI), Inc. Homeowners and Affiliates. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. Monroe, Albert. 2001. “How the Federal Housing Administration Affects Homeownership.” Harvard University, Department of Economics (www.jchs.harvard.edu/publications/governmentprograms/monroe_w02-4.pdf ). Mortgage Bankers Association. 2005. “Residential Foreclosures and Delinquencies Down from Last Year, According to MBA National Delinquency Survey.” Washington.

Low-Income Homeownership

105

Neighborhood Housing Services of Chicago. 2004. “Homeownership Preservation Initiative: Mid Term Report for the Chicago Homeownership Preservation Initiative.” Chicago. Newburger, Harriet. 2006. “Foreclosure Filings and Sheriff ’s Sales Experienced by LowIncome First-Time Home Buyers.” Housing Policy Debate 17, no. 2: 341–87. Nothaft, Frank E., and Yan Chang. 2004. “Refinance and the Accumulation of Home Equity Wealth.” Working Paper 04-02. McLean, Va.: Freddie Mac (February). ———. 2005. “Refinance and the Accumulation of Home Equity Wealth.” In Building Assets, Building Credit: Creating Wealth in Low-Income Communities, edited by Nicholas P. Retsinas and Eric S. Belsky, pp.71–102. Brookings. Onder, Zeynep. 2002. “Homeownership and FHA Mortgage Activity in Neighborhoods and Metropolitan Areas.” Journal of Housing Economics 11, no. 2: 152–81. Orr, James A., and Richard W. Peach. 1999. “Housing Outcomes: An Assessment of Longterm Trends.” FRBNY Economic Policy Review 3, no. 5 (September): 51–61. Orzechowski, Shawna, and Peter Sepielli. 2003. “Net Worth and Asset Ownership of Households: 1998 and 2000.” Current Population Reports P70-88: 1–23. Washington: U.S. Department of Commerce, Census Bureau. Oswald, Andrew J. 1999. The Housing Market and Europe’s Unemployment. Working Paper. United Kingdom: University of Warwick. Painter, Gary, and Christian Redfearn. 2002. “The Role of Interest Rates in Influencing Long-Run Homeownership Rates.” Journal of Real Estate Finance and Economics 25, nos. 2–3 (September): 243–67. Parcel, Toby L., and Elizabeth G. Menaghan. 1994a. “Early Parental Work, Family, Social Capital and Early Childhood Outcomes.” American Journal of Sociology 99, no. 4: 972–1009. ———. 1994b. Parents’ Jobs and Children’s Lives. New York: Aldine De Gruyter. Pitcoff, Winton. 2003. “Has Homeownership Been Oversold?” Shelterforce Online 127 (January/February), National Housing Institute. Potter, Lloyd B., and others. 2001. “The Influence of Geographic Mobility on Nearly Lethal Suicide Attempts.” Suicide and Life-Threatening Behavior 32 (Supplement): 42–48. Quercia, Roberto G., Spencer M. Cowan, and Ana B. Moreno. 2005. “The CostEffectiveness of Community-Based Foreclosure Prevention.” Minneapolis, Minn.: Family Housing Fund (www.fhfund.org/_dnld/reports/MFP_Full-Report.pdf ). Quercia, Roberto G., Lucy S. Gorham, and William M. Rohe. 2006. “Sustaining Homeownership: The Promise of Postpurchase Services.” Housing Policy Debate 17, no. 2: 309–39. Quercia, Roberto G., and Michael A. Stegman. 1992. “Residential Mortgage Default: A Review of the Literature.” Journal of Housing Research 3, no. 2: 341–79. Quercia, Roberto G., Michael A. Stegman, and Walter R. Davis. 2005. The Impact of Predatory Loan Terms on Subprime Foreclosures: The Special Case of Prepayment Penalties and Balloon Payments. University of North Carolina–Chapel Hill, Center for Community Capitalism in the Kenan Institute of Private Enterprise. Quercia, Roberto G., and Susan M. Wachter. 1996. “Homeownership Counseling Performance: How Can It Be Measured?” Housing Policy Debate 7, no. 1: 175–200. Quercia, Roberto G., and others. 2002. “Performance of Community Reinvestment Loans: Implications for Secondary Market Purchases.” In Low-income Homeownership: Examining the Unexamined Goal, edited by Nicholas P. Retsinas and Eric S. Belsky, pp. 348–74. Brookings.

106

George C. Galster and Anna M. Santiago

Reid, Carolina Katz. 2004. “Achieving the American Dream? A Longitudinal Analysis of the Homeownership Experiences of Low-Income Households.” Dissertation Discussion Paper. University of Washington, Department of Geography. (www.nonprofithousing.org/ 2005conference/Reid_Paper.pdf ). ———. 2007. “Locating the American Dream: Where Do Low-Income Homeowners Live?” In Chasing the American Dream: New Perspectives on Affordable Homeownership, edited by William M. Rohe and Harry L. Watson, pp. 233–62. Cornell University Press. Renuart, Elizabeth. 2004. “An Overview of the Predatory Lending Process.” Housing Policy Debate 15, no. 3: 467–502. Retsinas, Nicolas P., and Eric S. Belsky, eds. 2002. Low-Income Homeownership: Examining the Unexamined Goal. Brookings. ———. 2005. Building Assets, Building Credit: Creating Wealth in Low-Income Communities. Brookings. Richardson, Christopher A. 2002. “Predatory Lending and Housing Disinvestment.” U.S. Department of Justice, Civil Rights Division, Housing & Civil Enforcement Section (May) (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=338660). Rohe, William M. 2007. “Conclusion: Toward More Efficient and Equitable Homeownership Policies.” In Chasing the American Dream: New Perspectives on Affordable Homeownership, edited by William M. Rohe and Harry L. Watson, pp. 263–77. Cornell University Press. Rohe, William M., and Victoria Basolo. 1997. “Long-Term Effects of Homeownership on the Self-Perceptions and Social Interactions of Low-income Persons.” Environment and Behavior 29, no. 6: 793–819. Rohe, William M., and Roberto G. Quercia. 2003. Individual and Neighborhood Impacts of Neighborhood Reinvestment’s Homeownership Pilot Program. University of North Carolina–Chapel Hill, Center for Urban and Regional Studies. Rohe, William M., and Michael A. Stegman. 1992. “Public Housing Homeownership: Will It Work and for Whom?” Journal of the American Planning Association 58: 144-57. ———. 1994. “The Effects of Homeownership on the Self-Esteem, Perceived Control, and Life Satisfaction of Low-Income People.” Journal of the American Planning Association 60, no. 2: 173–84. Rohe, William M., and Leslie S. Stewart. 1996. “Home Ownership and Neighborhood Stability.” Housing Policy Debate 7, no. 1: 37–81. Rohe, William M., Shannon Van Zandt, and George McCarthy. 2000. “The Social Benefits and Costs of Homeownership: A Critical Assessment of the Research.” Working Paper 0102. Washington: Research Institute for Housing America (October). Rohe, William M., and Harry L. Watson, eds. 2007. Chasing the American Dream: New Perspectives on Affordable Homeownership. Cornell University Press. Rosenthal, Stuart. 2002. “Eliminating Credit Barriers: How Far Can We Go?” In Low-Income Homeownership: Examining the Unexamined Goal, edited by Nicolas P. Retsinas and Eric S. Belsky, pp. 111–45. Brookings. Ross, Stephen L., and George C. Galster. 2007. “Fair Housing Enforcement and Changes in Discrimination between 1989 and 2000: An Exploratory Study.” In Fragile Rights within Cities: Government, Housing, and Fairness, edited by John Goering, pp. 177–202. Lanham, Md.: Rowman and Littlefield. Ross, Stephen L., and John Yinger. 2002. The Color of Credit. MIT Press.

Low-Income Homeownership

107

Rossi, Peter H., and Eleanor Weber. 1996. “The Social Benefits of Homeownership: Empirical Evidence from National Surveys.” Housing Policy Debate 7, no. 1: 1–35. Rumberger, Russell W. 2003. “The Causes and Consequences of Student Mobility.” Journal of Negro Education 72, no.1 (Winter): 6–21. Rumberger, Russell W., and Katherine A. Larson. 1998. “Student Mobility and the Increased Risks of High School Dropout.” American Journal of Education 107, no.1 (November): 1–35. Sampson, Robert J., Jeffrey D. Morenoff, and Fenton Earls. 1999. “Beyond Social Capital: Spatial Dynamics of Collective Efficacy for Children.” American Sociological Review 64 (October): 633–60. Santiago, Anna M., and George C. Galster. 2006. “There’s No Place Like Home: The Experiences of Recent, Low-Income Homebuyers.” Paper presented at the 36th Annual Meeting of the Urban Affairs Association, Neighborhoods and Urban Transformation: The New Global Context. Montréal, April 19–22. Sherraden, Michael. 1991. Assets and the Poor: A New American Welfare Policy. Armonk, N.Y.: M. E. Sharpe. Shlay, Anne B. 2006. “Low-Income Homeownership: American Dream or Delusion?” Urban Studies 43, no. 3: 511–31. Stack, Steven. 1994. “The Effect of Geographic Mobility on Premarital Sex.” Journal of Marriage & the Family 56 (February): 204–08. Stegman, Michael A., Roberto G. Quercia, and Walter Davis. 2007. “The Wealth-Creating Potential of Homeownership: A Preliminary Assessment of Price Appreciation among Low-Income Home Buyers.” In Chasing the American Dream: New Perspectives on Affordable Homeownership, edited by Willam M. Rohe and Harry L. Watson, pp. 171–90. Cornell University Press. Stegman, Michael A., and others. 1991. “Designing Better Homeownership Assistance Programs Using the Panel Study of Income Dynamics (PSID): An Exploratory Analysis.” Journal of Housing Research 2, no. 1: 39–85. Stock, Richard D. 2001. “Predation in the Sub-prime Lending Market: Montgomery County.” Working Paper 1097. Dayton, Ohio: Center for Business and Economic Research (www.mvfairhousing.com/cber/). Strauss, Leslie, and Sue Phillips. 1997. “Housing Counseling in Rural America.” Washington: Housing Assistance Council (www.ruralhome.org/pubs/counseling/tableofcon.htm). Stuart, Guy. 2000. “Segregation in the Boston Metropolitan Area at the End of the 20th Century.” Cambridge Mass.: Harvard Civil Rights Project. Susin, Scott. 2003. “Mortgage Interest Rates and Refinancing: Racial and Ethnic Patterns.” Paper presented at the 30th annual Midyear Meeting, American Real Estate and Urban Economics Association. Washington, May 27–28. Temple, Judy A., and Arthur J. Reynolds. 1999. “School Mobility and Achievement: Longitudinal Findings from an Urban Cohort.” Journal of School Psychology 37, no. 4: 355–77. Thibodeau, Thomas G., and Brent W. Ambrose. 2002. An Analysis of the Effects of the GSE Affordable Goals on Low- and Moderate-Income Families. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. Tong, Zhong Yi, and John L. Glascock. 2000. “Price Dynamics of Owner-Occupied Housing in the Baltimore-Washington Area: Does Structure Type Matter?” Journal of Housing Research 11, no. 1: 29–66.

108

George C. Galster and Anna M. Santiago

Turner, Margery Austin, and Felicity Skidmore. 1999. Mortgage Lending Discrimination: A Review of Existing Evidence. Washington: Urban Institute. Turner, Margery Austin, and others. 2002. All Other Things Being Equal: A Paired Testing Study of Mortgage Lending Institutions. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. Turnham, Jennifer, and others. 2004. Study of Homebuyer Activity through HOME Investment Partnership Program. Washington: U.S. Department of Housing and Urban Development, Office of Policy Development and Research. Vandell, Kerry D. 1995. “How Ruthless Is Mortgage Default? A Review and Synthesis of the Evidence.” Journal of Housing Research 6, no. 2: 245–64. Van Order, Robert, and Peter Zorn. 2002. “Performance of Low-Income and Minority Mortgages.” In Low-Income Homeownership: Examining the Unexamined Goal, edited by Nicholas P. Retsinas and Eric S. Belsky, pp. 322–47. Brookings. Verba, Sidney, Kay Lehman Schlozman, and Henry Brady. 1995. Voice and Equality: Civic Voluntarism in American Politics. Harvard University Press. Willis, Lauren E. 2006. “Decisionmaking and the Limits of Disclosure: The Problem of Predatory Lending: Price.” Maryland Law Review 65, no. 3: 707–840. Wiranowski, Mark. 2003. Sustaining Home Ownership through Education and Counseling. Washington: Neighborhood Reinvestment Corporation and Harvard University, Joint Center for Housing Studies. Zimmerman, Tonya. 2006. “Predatory Lending: A Comprehensive Review of the Literature.” Paper presented at the 36th Annual Meeting of the Urban Affairs Association, Neighborhoods and Urban Transformation: The New Global Context. Montréal, April 19–22.

4 Tax and Expenditure Limitations and Their Effects on Local Finances and Urban Areas david brunori, michael bell, joseph cordes, and bing yuan

I

t has now become part of public finance lore that the property tax is the “worst tax.” During the latter half of the twentieth century, the Advisory Commission on Intergovernmental Relations (ACIR) conducted an annual public opinion poll to gauge the people’s views on the federal, state, and local tax systems. One of the most cited aspects of the poll was the request for people to identify the tax that they disliked the most. Over the course of the ACIR polling, the property tax was annually listed as the worst tax or the second worst tax following the federal income tax. So it is not surprising that virtually all states have some limitations on the local government’s ability to impose property taxes. States have used three primary types of limitation to constrain the taxing authority of local government, particularly with respect to property taxation. States have placed limits, through either constitutional amendment or statutory enactment, on rates, assessment increases, or general revenue and expenditure increases. Some states are subject to more than one type of limitation, and a few states, such as California, are subject to all three. Such tax and expenditure limitations (TELs) are controversial. Some see TELs as a necessary means of protecting the public from politicians who set taxing and spending levels higher than the public actually prefers. Others see TELs as an unwarranted attack on local autonomy.

109

110

Brunori, Bell, Cordes, and Yuan

Although TELs have been characterized in these somewhat black and white terms in the political debate, the fact that such measures enjoy political traction suggests that perhaps they are better viewed as attempts to craft political arrangements that balance, if not reconcile, competing objectives and interests. Indeed, local officials who one might think would be uniformly critical of measures intended to limit their autonomy have mixed reactions to TELs. According to a recent survey of elected city officials, respondents were pretty evenly split on the desirability of TELs.1 Just more than half of city officials responding to the survey felt TELs were sometimes a good idea (47 percent) or always a good idea (6 percent). Alternatively, 24 percent of respondents felt TELs were sometimes a bad idea, and 16 percent felt they were always a bad idea. Moreover, support for TELs is based on more than a desire by some voters and taxpayers to limit government spending and taxation, since these objectives can, in principle, be achieved through local elections by voting for candidates who promise such outcomes. The political argument made on behalf of TELs rests on the added belief that normal local electoral processes by themselves are an inadequate means of expressing preferences for fiscal restraint and that, hence, they need to be supplemented by additional “self-imposed” constraints on local taxing and spending decisions. The apparent willingness of many voters and public officials to support TELs raises a number of questions. Why are voters and public officials willing to enact laws that are intended to limit their discretion to raise taxes and spending, even though such measures may turn out in the future to be rather blunt instruments that not only put checks on unnecessary spending but also may limit the ability of communities to raise revenue to meet genuine future needs? How have such measures affected local finances and public services? Are there perhaps less blunt instruments for ensuring that there is appropriate restraint in local spending and taxation? The objective of this chapter is to summarize what we know about the answers to these questions, with special emphasis on the effects of TELs on urban public finance.2 We begin by discussing the history of TELs and present a typology of the variants of TELs that are presently in existence. We then examine the political economy of TELs with a special emphasis on whether adopting limitations on taxing and spending potentially improves or worsens the wellbeing of the citizens and taxpayers in jurisdictions that adopt such measures. This discussion is followed by a survey of the rather extensive empirical literature on TELs in which we examine research on taxpayer motivations for supporting (or opposing) TELs and the various effects of TELs on urban public 1. Hoene (2005). 2. For a recent analysis of state-level TELs, see Bae and Gais (2007).

Tax and Expenditure Limitations

111

finances. We conclude with a summary of the main findings of the empirical research on TELs, a discussion of what insights the research offers about possible policy alternatives to TELs, and a discussion of research remaining to be done.

A Brief History of Tax and Expenditure Limitations These various limitations arose from the public’s well-documented unhappiness with the property tax. But contrary to what many believe, the tax limitation movement did not start with California’s Proposition 13. Rather, the initial push to limit property tax increases began during the Great Depression. Despite the harsh economic realities of the Depression, the underemployed and unemployed were still faced with paying their property taxes. While property values (and hence property tax burdens) were falling, the dramatic loss of income forced many homeowners into or to near bankruptcy. By 1933 real estate values fell by 92 percent.3 But tax assessments did not fall nearly as far or as fast. Moreover, the share of income absorbed by the property tax doubled between 1929 and 1932, reaching 11.3 percent.4 Local governments remained heavily dependent on the property tax. In 1932, for example, property taxes accounted for 85.2 percent of local government “own-source revenue.”5 The Depression produced an upsurge in tax delinquencies, bankruptcies, and foreclosures. Nationwide, localities had property tax delinquency rates of more than 26 percent.6 The sheer volume of delinquencies and the threat of losing homes gave rise to the most serious tax revolts in America since the Whiskey Rebellion of the 1790s. As early as 1930, unhappiness with property tax burdens caused a storm of protest across the country. Thousands of taxpayer organizations were formed, all created specifically to fight for property tax relief. These protests had significant repercussions. In 1932 and 1933 alone, sixteen states and numerous localities adopted some form of limitations on property taxation.7 Throughout the Great Depression, states began to limit local property taxation. Michigan (1933), Nevada (1936), Ohio (1929), and Oklahoma (1933) placed statutory or constitutional limits on property tax rates during the early years of the Depression. Later, West Virginia (1939) and Washington (1944) would place limits on rates. 3. Beito (1989). 4. O’Sullivan (2000). 5. Beito (1989). 6. O’Sullivan (2000). 7. O’Sullivan (2000).

112 Brunori, Bell, Cordes, and Yuan The property tax unrest during the Great Depression also spurred states to adopt property tax relief measures such as homestead exemptions. But the early limitation movement and the proliferation of homeowner relief did not quell the public’s dislike of the property tax. That dislike, combined with a growing cynicism and distrust of government, led to the most significant development in American property tax history: the tax revolts of the late 1970s and early 1980s. Proposition 13 and its progeny not only dramatically changed property taxation but also were a defining moment in the public’s attitudes toward taxation in general in the United States. The tax revolts changed the way many local governments raised revenue. But they also signaled the beginning of a new and decidedly antitax political philosophy that continues to this day. The causes of Proposition 13 were varied. The public was frustrated by continuously rising property tax burdens. California real estate values were increasing 25 percent per year in the decade before passage of Proposition 13. The public was equally frustrated with local government leaders who refused to lower tax rates and state government leaders who refused to offer relief. Political leaders around the state were aware of the property tax problem for at least a decade before 1978. Governor Ronald Reagan proposed limiting property taxes in 1973. Los Angeles county assessor Philip E. Watson led two property tax limitation drives in 1968 and 1972. These efforts were unsuccessful, and as Clarence Lo noted, the California legislature refused to provide property tax relief for four straight years before the proposition passed.8 Another cause for Proposition 13, and indeed for other property tax protests, was school finance litigation. In 1971 the Supreme Court of California declared that the system of financing education through local property taxes was unconstitutional. The court ordered that the state assume the primary role in financing the schools. That decision had the effect of diminishing public support for property taxes and is arguably one of the reasons for the public’s willingness to approve Proposition 13.9 On June 6, 1978, two-thirds of California voters chose to radically reduce and limit property taxes in the state. Proposition 13 rolled back assessment values to 1976 levels. It limited increases in assessed value to 2 percent per year as long as the property was not sold. It imposed a 1 percent limit on the property tax rate. The measure also required that all state tax increases be approved by a two-thirds vote of the legislature and that all local tax increases be approved by a vote of the electorate. The effect was dramatic. Property tax revenue immediately fell by 57 percent across the state. Local governments in California collected more than $6.6 bil8. Lo (1995). 9. Fischel (1989).

Tax and Expenditure Limitations

113

lion less in property tax revenue in 1979 than they did in 1978.10 California property taxes went from being 51 percent above the national average in 1978 to being 22 percent below the average in 1981. California local governments became much more dependent on state aid as a result of Proposition 13. They also began increasing user fees and charges significantly (which were not subject to limitation). Between 1978 and 1981, local government user fee revenue increased by 48 percent, and Los Angeles increased user fee revenue by 67 percent.11 The immediate impact of Proposition 13 was significant. Within six months after passage of Proposition 13, tax limitation measures were on the ballots in seventeen states, and all but five were approved. There were fifty-eight ballot measures during the 1979–84 period concerning property tax classification, exemptions, assessment reform, and rollbacks. Among the most successful were tax and expenditure control measures. Forty-three states adopted new property tax limitations or relief plans between 1978 and 1980. Idaho and Massachusetts followed California’s lead and adopted measures that both cut and limited property taxes. New state spending limits were set in New Jersey and Colorado. Several states (Arizona, Michigan, Louisiana, Oregon, Utah, and Washington) tied growth in local government spending or revenue to growth in personal income or population. Michigan restricted growth in local property tax revenues to the rate of inflation, and state revenues were limited to the share of personal income they represented in 1978–79. Although the tax revolt movement lost momentum in the latter half of the 1980s, continued dislike of the property tax together with the fiscal pressures resulting from the recent recession have served to maintain interest in changing the tax and spending activities of state and local governments. In 1992, voters in Florida approved a 3 percent limit on assessed value increases for homeowner property until sale. In a historic move, the Michigan legislature in 1993 voted to eliminate all property taxes for school operations.

Property Tax Limitations Today California’s Proposition 13, which many would regard as the “paradigmatic” example of a tax expenditure limitation, is, in fact, a specific form of a tax and expenditure limitation among several different variants. Just as states differ in their decision to adopt a limitation measure or not, states and localities that have chosen to adopt limits differ in the design of limitations that they choose to implement. Important distinctions are whether limitations are imposed at the 10. Citrin (1984). 11. Richter (1984).

114

Brunori, Bell, Cordes, and Yuan

state or at the local level, how the limitation is administratively imposed, and whether the constraint implied by the limitation is considered to be binding or nonbinding. A listing of the various forms of TELs currently in existence is provided in appendix tables A-1, A-2, and A-3.

State and Local Limits One important distinction is between state TELs and local TELs.12 State TELs refer to limitations imposed on the state government, while local TELs refer to those imposed on local governments by the state. Both types of limitations were simultaneously adopted in a number of states. The focus of much of this chapter is on limits on local rather than on state spending and taxation, although we do discuss empirical evidence on the effect on spending of both types of TELs. Not surprisingly, state TELs are found to be less effective than local TELs in controlling the size and growth of government because states have greater capacity to circumvent those limitations.13

Administrative Implementation of Limitation In the case of property taxes, three options for limitation present themselves: limiting the base of the property tax through assessment limitation, limiting the rate at which the base may be taxed, and limiting revenues or expenditures or both. assessment limitations. Twenty states have some form of limitation on the amount that assessed values can increase each year (see appendix table A-1, which includes a brief description and legal authority for the limits). The assessment limits usually apply only to residential property and rarely to other uses. For example, there are few commercial property assessment limitations in the United States. Assessment limitations vary widely. For example, some states such as California, Florida, Oklahoma, and New Mexico have flat percentage limitations on yearly increases in assessed value. California’s Proposition 13 limits the increase in assessed value for residential property to 2 percent a year unless the property is sold. When the property is sold, it acquires a market value for assessment purposes. But not all assessment limits are so straightforward. Colorado mandates that residential property makes up no more than 45 percent of total assessed value. This requirement serves to limit the growth of residential assessment, but property owners have a difficult time determining how much. Georgia limits only conservation use assessments. New York only limits Nassau County assessments. 12. Joyce and Mullins (1991); Mullins and Joyce (1996); Shadbegian (1996, 2003). 13. Joyce and Mullins (1991); Mullins and Joyce (1996); Shadbegian (1996, 2003).

Tax and Expenditure Limitations

115

Virginia has a 1 percent annual assessment limitation, but that limit can be overridden by a simple majority vote of the legislature of the taxing jurisdiction. rate limitations. Forty states have some form of limitation on the property tax rate that can be levied by a local government (see appendix table A-2). Like assessment limitations, rate limitations are set by statute, by the constitution, or by both. Rate limitations also vary from state to state. Some states (California and Wyoming, for example) have rate limit laws that do not allow for any increases. But many states have rate limit laws that can be overridden in particular circumstances. Alabama, Ohio, and Michigan, for example, allow their rate limits to be increased after a majority vote of the electorate. Oregon and Nebraska require a supermajority vote of the electorate to override a rate limit. In some states (Maryland and Illinois) the property tax rate limit limitation is a local option. revenue and expenditure limitations. Thirty-seven states have some form of limitation on revenue and expenditure increases in local governments (see appendix table A-3). As in the case of limitations on tax rates, many states allow tax and expenditure limits to be overridden by either the electorate or the legislative body in the taxing jurisdiction. In Alaska local governments cannot collect more than $1,500 per capita. In Arkansas property tax revenue cannot increase more than 10 percent from the previous year. Like other limitations, there are wide variations among the states that limit revenue and expenditure growth. In many cases, expenditures are limited by some combination of population growth and inflation. Some states place a flat percentage limit on growth; some tie expenditure growth to personal income growth. And some states (Massachusetts, Arizona, Arkansas, Indiana, Maine, and South Dakota) limit revenue growth only to the property tax.

Binding and Nonbinding Limits Scholars also distinguish between different types of local TELs according to their stated target and their stringency or the degree to which the constraint is binding.14 Joyce and Mullins identified six categories of TELs, namely, overall property tax rate limit, specific property tax rate limit, assessment increase limit, property tax levy limit, general revenue or general expenditure limit, and full disclosure or truth-in-taxation.15 Tax rate limits and assessment limits are 14. Anderson (2006); Brown (2000); Joyce and Mullins (1991); Mullins and Joyce (1996); Preston and Ichniowski (1991); Poterba and Rueben (1995); Shadbegian (2003); Sokolow (1998, 2000). 15. See Joyce and Mullins (1991). An overall tax rate limit sets a ceiling on the aggregate tax rate of all local governments. A specific tax rate limit applies to specific types of local government, such as school districts, or narrowly defined service areas. An assessment increase limit caps the growth rate of assessed values and is intended to control the ability of local gov-

116

Brunori, Bell, Cordes, and Yuan

expected to be potentially binding if combined with each other. Limits on the overall property tax levy or on general revenues or expenditures are also considered potentially binding because of the fixed nature of the ceiling, whereas full disclosure is considered nonbinding, as the local legislative body can easily raise the tax through a formal vote.16 Some states have adopted a combination of the potentially binding limitations, while some have implemented only the leastbinding type. As states with potentially binding TELs vary in the size of their cap on the growth rate of property taxes, Ronald Shadbegian suggested 5 percent as the threshold to distinguish between stringent and constringent limitations.17 James Poterba and Kim Rueben, on the other hand, classified limits on property tax revenues, property tax rates, or general revenues or expenditures as effective limitations and assessment limits—along with full disclosure—as ineffective.18 In Nathan Anderson’s review, assessment limits offer better insurance against large property tax increases for homeowners than limits on rates and revenues do.19 TELs may also be divided into those with an override mechanism and those without one.20 Override provisions provide a method for overcoming tax and expenditure limitations. Override provisions can be as simple as allowing a majority vote of the governing body. But most are much more difficult to implement, often requiring a supermajority vote of the electorate in a special election.

The Political Economy of TELs The various types of tax and expenditure limitations have different effects on local government policies in general and on tax policy in particular. The most obvious policy issue emanating from the tax and expenditure limitation movement is the effect on local government autonomy. The modern political landscape remains dominated by the belief that localities are critical to the governance of America. As Donald Norris has noted, “Local government autonomy—or the ability of these governments to exercise ernments to raise revenue by reassessment of property or through natural or administrative escalation of property values. A property tax levy limit constrains the growth rate of total revenue that can be raised from the property tax, independent of the rate. A general revenue or general expenditure limit sets the maximum growth rate of total revenue or spending or both. Full disclosure, or truth-in-taxation, requires public discussion and a specific legislative vote before enactment of tax rate or levy increases. 16. Joyce and Mullins (1991). 17. Shadbegian (2003). 18. Poterba and Rueben (1995). 19. Anderson (2006). 20. Cutler, Elmendorf, and Zeckhauser (1999); Figlio and O’Sullivan (2001).

Tax and Expenditure Limitations

117

their police powers broadly within their territories—is sacrosanct in the United States.”21 An influential report issued by the National Conference of State Legislatures asserted that public service responsibilities should be assigned to the lowest level of government to foster accountability and best meet local citizens’ needs.22 Such views of localism are broadly accepted by political leaders, academics, and the general public.23 There is little debate that local governments are the most efficient providers of certain public services. Scholars have long recognized that federal, state, and local governments are each capable of providing some services more effectively than the other levels of government. Public services should be provided by the jurisdiction covering the smallest area over which benefits are distributed.24 As Richard Bird asserted, “So long as there are variations in tastes and costs, there are clearly efficiency gains from carrying out public sector activities in as decentralized a fashion as possible.”25 Another rationale for localism is that it promotes democratic values and practices.26 That is, government that is closer to the people not only will reflect citizen desires better, but it will encourage them to participate in public affairs and participate in the democratic process. As one prominent political scientist observed, “The bedrock of American local democratic theory is that the role of the local government is to reflect the will of the people and that direct individual participation in local government is the best means of achieving this end.”27 And there is scholarly evidence that the public desires local government because of the democratic ideals that such government fosters.28 The theory of localism outlined above is dependent upon local governments having an independent source of revenue within their political control that is also adequate to meet local needs for goods and services.29 An implication is that local governments, through their elected officials, must have the ability to impose taxes on their citizens without undue interference from state or federal law. Without such ability, local governments cannot effectively or efficiently provide public services or respond to the needs of their citizens. As Bird stated, “Local governments should not only have access to those revenue sources that they are best equipped to exploit—such as residential property taxes and user 21. Norris (2001, p. 566). 22. NCSL (1997). 23. Brunori (2003). 24. Oates (1972); Gramlich (1993). 25. Bird (1993, p. 211). 26. Frug (1980). 27. Wolman (1995, p. 136). 28. Haselhoff (2002). 29. See Peterson (1995).

118

Brunori, Bell, Cordes, and Yuan

charges for public services—but they should also be both encouraged and permitted to exploit these sources without undue central supervision.”30 The system has worked because the three levels of government have traditionally enjoyed access to their own sources of revenue. The federal government has used primarily income taxes to fund its operations. The states have used a combination of income and consumption taxes. And local governments have relied primarily on property taxes to pay for the services that their citizens demand.31 The connection between successful local government and the ability to raise revenue is undeniable. Indeed, there is widespread agreement that local governments need tax authority to successfully provide the services demanded by their citizens. Virtually every economist who has studied the subject has concluded that local governments must have access to own-source revenue to fulfill their responsibilities in an efficient and effective manner.32 The proliferation of tax and expenditure limitations has curtailed local taxing authority and as a consequence hindered local political autonomy. There is some evidence that locally elected officials are concerned about this loss of autonomy. For example, in a recent survey by the National League of Cities, 54 percent of elected city officials said they would not trade local tax revenue authority for a larger share of state revenues.33 The empirical research summarized below, in the section “Effects of TELs on Urban Public Finances,” generally finds that tax limitations have constrained the ability of local governments to raise revenue from property taxes, which remain the most important source of independent revenue for local governments.34 The amount of property tax revenue forgone as a result of the various limitations is in the tens of billions of dollars nationwide.35

TELs and Restraining Leviathan If local TELs have the potential to reduce local autonomy and limit jurisdictions’ abilities to respond to citizen demands for public goods and services, why would voters voluntarily agree to impose such potentially costly constraints, when they 30. Bird (1993, p. 211). 31. Musgrave (1983) suggested that ideally local government would be financed by user fees and property taxes, state governments through consumption taxes, and federal government through income taxes. That “ideal” has more or less been accomplished in the United States. 32. For examples, see Oates (1991); Bird (1993); Sokolow (1998); Bowman, MacManus, and Mikesell (1992); Giertz and McGuire (1991); McGuire (1995). 33. Hoene (2005). 34. McCabe and Feiock (2000); Shadbegian (1996, 1999); Poterba and Rueben (1995); Preston and Ichniowski (1991); Sexton, Sheffrin, and O’Sullivan (1999); Dye and McGuire (1997); Brown (2000); Cutler, Elmendorf, and Zeckhauser (1999). 35. O’Sullivan (2000).

Tax and Expenditure Limitations

119

already have mechanisms in place such as referenda and general and special elections to make their preferences for public spending known, and discipline officials who fail to act on such preferences? Scholars have sought answers to this question by applying theoretical models of public choice and public finance and by examining empirical data on taxpayer motivations for supporting TELs. As initially noted by Geoffrey Brennan and James Buchanan, adoption of tax and expenditure limitations cannot be explained by traditional normative models of public finance in which a benevolent government pursues public policies that maximize the well-being of taxpayers and voters.36 As the authors trenchantly observed: In this institutional setting, any constraints such as tax limitation (or electoral processes!) can only be viewed as undesirable restrictions, ranging from the mildly irritating to the wildly perverse. Tax limitation cannot be understood within such a political setting without retreat into the assumption of irrational political behavior.37 The implication is that popular support for TELs can be analyzed and understood only in the context of public expenditure models that assume something other than governments that can be counted on to pursue only the public interest. The most frequently employed variant of such a model is the so-called Leviathan model proposed by Brennan and Buchanan, which interprets the passage of TEL laws as citizens’ attempts to rein in a wasteful and inefficient government.38 In one form or other, this model has informed much of the empirical and conceptual research on TELs. The model of political behavior proposed by Brennan and Buchanan posits the existence of a government that is initially given a monopoly on the use of coercive power for the mutual benefit of individual citizens but that over time comes to use that monopoly power to foster its own ends. In this so-called Leviathan model of government, government officials (understood to be elected politicians and civil servants) have strong incentives to maximize public budgets because doing so can bring them both pecuniary and nonpecuniary rewards. Public officials are able to act on these incentives because they have the legal (monopoly) power to tax and spend and because they enjoy an informational 36. Brennan and Buchanan (1979). See also McGuire (1999). 37. Brennan and Buchanan (1979, p. 11). 38. Scholars use different names for the theory, but the arguments embodied are essentially the same: for example, the Leviathan model is referred to as agency loss theory in Cutler, Elmendorf, and Zeckhauser (1999), as principal-agent model in Temple (1996), and as budgetmaximizing bureaucrat model in Downes and Figlio (1999). Besley and Smart (2007) have proposed a more general model of local political and electoral behavior that includes many of these conceptual models .

120 Brunori, Bell, Cordes, and Yuan advantage relative to the electorate, inasmuch as they have access to better and more complete information about the true cost of satisfying the electorate’s demands for public goods and services. In such an environment, even a democratic political process has the potential to produce levels of government spending and taxation that are higher than is socially optimal. The challenge facing citizen voters and taxpayers therefore is to devise a set of institutional arrangements that allow them to obtain the benefits of government action, while at the same time minimizing the risk of monopoly exploitation by Leviathan. One set of such arrangements are tax and expenditure limitations. Namely, by imposing a cap on taxes and spending, voters expect to constrain the size and growth of government and to improve its efficiency and financial accountability.

Political Agency Model of TELs The normative model of an entirely benevolent government and the BrennanBuchanan Leviathan model can be seen as opposite ends of a continuum of models of government behavior. Timothy Besley and Michael Smart examine the role of TELs in a game theoretic model that incorporates features of both— the purely benevolent and the Leviathan models of government behavior.39 The Besley-Smart model is populated by a world of politicians who must initially be elected to office and who must then stand for reelection. These politicians are made up of good politicians, who if elected can be counted on to make tax and expenditure decisions that are intended to maximize the expected wellbeing of voters and taxpayers, and bad politicians, who if elected will exploit their monopolistic power to tax and spend and their informational advantage about the true cost of producing public goods to pursue their own self-interest. In such an environment, voters must elect or reelect politicians to office without knowing whether a particular candidate or incumbent is good or bad. In this context, the role of the political process is twofold. Elections serve to sort out and elect politicians who are most likely to act in the interest of voters and taxpayers (for example, elect good rather than bad politicians) and discipline incumbents who are ultimately revealed as acting in ways that do not further the public interest. Besley and Smart show that in such a setting the fact that incumbents must stand for reelection has two positive effects. It improves the average quality of office holders by electing good politicians to office with some positive probability (the selection effect), and it creates incentives for otherwise bad incumbent politicians to behave more like good politicians so that they are reelected (the discipline effect). 39. Besley and Smart (2007).

Tax and Expenditure Limitations

121

The question then becomes whether adding tax limitations as an additional fiscal constraint can improve upon these outcomes, which are achievable simply from having periodic elections. If one assumes that imposing such constraints does not change the political equilibrium of the game, the answer is yes, as in the case of the Brennan-Buchanan Leviathan model. In that simple case, imposing a TEL does not change the proportion of good politicians who are normally (though without certainty) elected to office, while providing an added restraint on the behavior of bad politicians. However, as Besley and Smart point out, by restraining the behavior of otherwise bad politicians, the imposition of a tax limitation somewhat paradoxically also makes bad politicians look more like good politicians, which serves to dilute the selection effect of elections. An important implication is that when the selection effect is more important than the discipline effect, TELs may not necessarily make voters better off because, in effect, what voters and taxpayers “gain” in the form of discipline of bad politicians is more than offset by a loss in the ability of elections to sort out bad from good politicians in the first place. A direct implication of this result is that TELs will be more likely to improve citizen welfare in settings in which there is a high proportion of bad politicians among all politicians, because it is in those settings that the discipline effect of elections is more important than the selection effect. The implication of the analysis is that TELs will be most effective in enhancing the well-being of voters and taxpayers in jurisdictions with a high proportion of “self-interested” (as distinct from “public-spirited”) politicians.

Taxpayer Self-Interest Other explanations for taxpayer support for TELs place a greater emphasis on taxpayer (perceptions of ) self-interest as a motivating factor. Anderson, for example, hypothesized that homeowners employ property tax limitations, assessment limits in particular, as insurance against spikes in their property tax payments, which may occur even when revenues remain constant.40 Still other forms of taxpayer self-interest have been proposed as explanations for voter support of state-imposed tax expenditure limits on localities. The question is why local residents would find it in their interest to support statewide limitations on local property taxes at the expense of their own local fiscal autonomy. Voter support for state-level restrictions on local taxing and spending authority can be rationalized within the framework of the Leviathan model as one means by which local voters seek to restrain Leviathan government.41 However, 40. Anderson (2006). 41. This point is made by Vigdor (2004). Similarly, such behavior would fit also within the model proposed by Besley and Smart (2007).

122

Brunori, Bell, Cordes, and Yuan

such behavior can also be explained as attempts by different groups of taxpayers not only to prevent exploitation by politicians and bureaucrats but also to prevent exploitation from other taxpayers or to extract fiscal gains for themselves. Judy Temple focused on the desire of the “median voter” to ensure that his or her preferences for local public goods matter.42 She argued that median-income voters in heterogeneous communities will find it in their self-interest to support the state imposition of fiscal constraints to forestall the formation of winning coalitions by other taxpayers that may lead to spending increases above their preferred level. Jacob Vigdor focused on the self-interest of nonresident taxpayers, arguing that individuals in their capacities as absentee landlords, nonresident employees, or potential new residents use statewide limitations to extend their voting franchise in local jurisdictions where they do not live but where they own property, work, or intend to relocate in the future.43

Fiscal Illusion Last, it is also possible that support for TELs among voters, if not necessarily among the elected officials responding to the previously cited survey, may reflect a certain degree of fiscal illusion. Taxpayers and voters may not necessarily see the connection between taxes imposed and services received. This may be due partly to simple ignorance, but also may be due to the asymmetrical visibility of taxes and expenditures. Taxes, although they vary in visibility by type, are nonetheless quite visible to those who pay them. Expenditures, however, are generally visible more with respect to the outcomes they produce. Some expenditures produce outcomes that are visible only in the future and only if they are not made (for example, maintenance); others are hardly visible to the public at all (for example, planning, evaluation, research, administration). In some cases the relationship between spending and output and outcome is complex and, again, if successful, may be visible only over time (for example, education). 42. Temple (1996). 43. Vigdor argues that nonresident employees have a stake in lowering property taxes in the jurisdiction where they work because higher capital costs induced by high property taxes can lower the marginal product of labor wages. Firms may also exit a jurisdiction in response to tax breaks offered elsewhere. Several participants at the conference noted that the political science literature may offer venue-shifting as an explanation for TELs. The influence of various interests differs across levels of government (local, state, federal) as well as across institutions. Groups representing those interests that are unable to gain their objectives through efforts at one level will attempt to shift the issue to a venue for decisionmaking that is more favorable to them. In the case of TELs, interest groups (such as higher-income households, households with high property values, and businesses) that cannot assemble a majority within the local decisionmaking arena may try to shift their efforts to the state level where the composition of the electorate as well as the decisionmaking rules differ. TELs imposed by the state might be an application of this. See Vigdor (2004).

Tax and Expenditure Limitations

123

A possible implication is that taxpayers may see the imposition of TELs as offering a kind of “fiscal free lunch” in which taxes are reduced with little effect on spending. Of course, this motivation is to some extent observationally equivalent to support for TELs as an attempt to “restrain Leviathan,” because if politicians engage in wasteful spending it should be possible, in principle, to restrain simultaneously wasteful spending and taxes, without at the same affecting the true quality of local public services.

Why Taxpayers Support TELs: Empirical Evidence Evidence bearing on why taxpayers support TELs comes from two main sources. Responses to public opinion surveys are a source of direct evidence. In addition, indirect inferences about taxpayer motives can be drawn from decisions made by individual jurisdictions when given choices between supporting or rejecting TELs.

Public Opinion Surveys Public opinion surveys have frequently found a substantial number of voters who respond that taxes or government spending could be cut without reducing the quantity or quality of public services, reflecting an implicit belief that government is wasteful and inefficient.44 These responses could indicate that many voters truly believe it is possible to restrain taxes without paying a significant price in the form of reduced public services. There is, however, also evidence in the survey data that voters may be ignorant about the likely impact of TELs and may suffer from a form of free lunch illusion that taxes can be reduced without affecting public services.45

Decisions by Jurisdictions to Support or Reject TELs David Cutler, Douglas Elmendorf, and Richard Zeckhauser draw inferences about voter motives for supporting or rejecting TELs by examining the correlation between an individual’s property tax payments and his or her attitude toward TELs. If individuals with high property tax bills either see themselves as being exploited by self-interested politicians (Leviathan) or view these bills as a signal that incumbent politicians are from the “bad” group (political agency), one might expect to find a positive correlation between an individual’s support 44. Courant, Gramlich, and Rubinfeld (1980); Downes and Figlio (1999); Joyce and Mullins (1991); Ladd and Wilson (1982); Temple (1996); Wallin (2004). 45. For evidence regarding the ignorance of the electorate concerning TELs, see Courant, Gramlich, and Rubinfeld (1980). Concerning free lunch illusion, see Ladd and Wilson (1982); Temple (1996); Wallin (2004).

124

Brunori, Bell, Cordes, and Yuan

for TELs and the size of his or her property tax bill.46 Such a correlation would also be consistent with the insurance hypothesis advanced by Anderson. Using data on votes for Proposition 21⁄2 and for overrides in Massachusetts, Cutler, Elmendorf, and Zeckhauser found that, consistent with theory, voters are indeed more likely to support Proposition 21⁄2 and less likely to support overrides of TELs the higher their property tax bill is.47 Both Temple and Vigdor examined factors that explain support for TELs in different jurisdictions to test their hypotheses that voter support of statewide TELs is motivated by taxpayer self-interest.48 In Illinois, local jurisdictions may choose home rule status and thereby be exempted from state-imposed TELs. Temple exploited this natural experiment to test her hypothesis that limitations should be more likely to be accepted by communities with relatively heterogeneous populations, which is taken as an indicator of intracommunity variation in individual preferences for local public services.49 Her empirical analysis demonstrated that age-heterogeneous, lower-income municipalities are more likely to forgo the opportunity to adopt home rule, while home rule is more likely to be adopted in local governments with city or village managers. Vigdor used data on the interjurisdictional incidence of Proposition 21⁄2 in Massachusetts, combined with data on support for this ballot measure, to test the hypothesis that support for statewide limitations on local taxes can be attributed to self-interested behavior by nonresident taxpayers. As predicted by the “selfinterested nonresident hypothesis,” support for the proposition was higher among voters who live near, rather than in, communities with high tax rates and whose tax rates would be reduced by Proposition 21⁄2. In addition, Vigdor found that there was significantly less support for the statewide tax limitation in communities with a greater ability to export local property taxes to nonresidents. Vigdor also tested the prediction that adopting statewide tax limitations would cause a re-sorting of residents among communities. In a world with no statewide limitations on local taxing power, jurisdictions that enjoy special locational advantages for industry might capitalize on this advantage by raising local property tax rates, effectively capturing a portion of the locational rent for the local public sector. Such favored communities would be able to offer individual taxpayers more “public goods per dollar of noncommercial tax burden” and would be attractive to households with strong tastes for public goods and services who would be willing to outbid other households for land and housing in such communities. Imposing tax limits on such communities would make them 46. Cutler, Elmendorf, and Zeckhauser (1999). 47. Cutler, Elmendorf, and Zeckhauser (1999). 48. Temple (1996);Vigdor (2004). 49. Temple (1996).

Tax and Expenditure Limitations

125

relatively less attractive to “public goods preferring households” and more attractive to other households and over time should produce some re-sorting of household types among communities. Vigdor tested this hypothesis by constructing a control group of communities in Connecticut, which are not subject to TELs and which have characteristics similar to those of Massachusetts communities that are constrained by Proposition 21⁄2. A comparison of the demographics in both communities before and after the adoption of Proposition 21⁄2 indicates that community composition changed in the Massachusetts communities but not in the Connecticut communities.

Effects of TELs on Urban Public Finances Because local governments have traditionally been so dependent on the property tax, limitations have had several effects on urban public finances. Some local governments have responded to tax and expenditure limitations by increasing their reliance on user charges and fees. In other cases reduced locally and independently generated revenues have been replaced by revenues from the state, leading to greater centralization in the finance of local public services. Sokolow noted that the state share of state-local tax revenue in the United States increased from 59.6 percent to 63.3 percent between 1970 and 1994 and has argued that local restrictions on property taxes are a cause of diminished local government autonomy and increased fiscal centralization.50 Also, there is evidence that TELs have led to reduced spending on local public education, although there is not uniform agreement that such reduced spending necessarily has reduced the actual output of (as distinct from spending on) local public schools. As Sokolow found, state control over local government finances has been increasing for more than two decades.51 And the rising state share of state-local revenue best illustrates the increased level of state control. The various limitations have forced local governments to increasingly rely on state aid to fund services that once were paid for exclusively by the localities. There is also a substantial empirical literature on the various effects of TELs on the public finances of the communities that are subject to these limitations. Broadly speaking, these empirical studies fall into three broad groups: examination of whether TELs have indeed had their intended effect of restraining government revenue and spending, examination of the effect of TELs on the level and composition of government services, and analysis of the distributional effect of TELs. 50. Sokolow (2000). 51. Sokolow (1998).

126

Brunori, Bell, Cordes, and Yuan

Effects of TELs on Taxes and Spending Because most TELs have the explicit purpose of constraining property tax revenues and containing the growth of government, there is an extensive empirical literature focusing on the effects of TELs on the fiscal structure of state and local governments.52 These studies in turn can be grouped into those that focus on the extent to which different types of TELs impose binding or nonbinding constraints and those that focus on the effect of TELs on revenue from property taxes, the level of government spending, and the mix of local financing sources. are tels binding or not? The degree of interstate variation in the design of TELs as discussed earlier implies heterogeneity in their effects on government finance. Not surprisingly, binding limitations are more likely to assert an influence than those that are less binding.53 One indirect way of testing this conjecture is to compare the effects of TELs imposed on state spending and taxes with the effects of those that constrain local spending and taxes. One might expect that the former would have less of a constraining effect than the latter because of the breadth and diversity of state revenue and spending compared with those of local revenue and spending. Shadbegian examined the impact of limitations on state and local taxes and expenditures for school finance.54 His analyses of the state-level data provide support for a negligible impact of state-level limits on local education spending per student in contrast to local-level limitations that have a substantial impact. In an earlier study of local TELs, Shadbegian also found that the effects of tax caps appear to increase with their stringency. For example, he found that less stringent local TELs reduced property taxes by about $32 per capita and “other” taxes by roughly $12 per capita, but local governments facing less stringent TELs raised about $23 per capita in additional miscellaneous revenues for a net reduction of $22 per capita (about 4 percent) of own-source revenues.55 Moreover, he found that more stringent TELs reduced property taxes by $44 per capita, raised “other” taxes by $3, and raised $6 per capita less in miscellaneous revenues than expected, leading to an overall net reduction in own-source revenues of $47 per capita (about 9 percent).56 52. See Bland and Laosirirat (1997); Brooks and Phillips (2006); Cornia and Walters (2005, 2006); Cutler, Elmendorf, and Zeckhauer (1997, 1999); De Tray and Fernandez (1986); Dye and McGuire (1997); Dye, McGuire, and McMillen (2005); Elder (1992); Figlio (1998); Figlio and O’Sullivan (2001); Galles and Sexton (1998); Glickman and Painter (2004); Joyce and Mullins (1991); Merriman (1986); Mullins and Joyce (1996); O’Sullivan, Sexton, and Sheffrin (1994, 1995); Preston and Ichniowski (1991); Shadbegian (1996, 2003); Sokolow (1998, 2000); Waters, Holland, and Weber (1997). 53. Joyce and Mullins (1991). 54. Shadbegian (2003). 55. Shadbegian (1999). 56. Shadbegian (1999, pp. 233–34).

Tax and Expenditure Limitations

127

The cross-state analysis by Ann Preston and Casey Ichniowski empirically corroborated Philip Joyce and Daniel Mullins’ findings about the degree to which various types of limitations impose binding or nonbinding constraints.57 Limits on property tax rates, when coupled with assessment limits, have resulted in the largest reduction in the growth of per capita property tax revenue. Property tax levy limits and general revenue limits have had significant yet smaller impacts. Similarly, Tom Brown argued that the comprehensive revenue and expenditure limit in Colorado has been more effective than the earlier assessment limit in controlling the growth of government.58 effect of tels on property tax revenue. Empirical studies of TELs based on data from individual states and studies based on cross-state comparisons marshal compelling evidence that state-imposed limitations on local property tax rates, levies, and assessments have a significant impact on property tax revenue—reducing these revenues absolutely or reducing the share of local ownsource revenue or general revenue.59 Richard Dye, Therese McGuire, and Daniel McMillen also found evidence in Illinois that these effects became more pronounced over time.60 Illinois school districts with the cap in effect for four to nine years saw a more dramatic decline in the growth rate of property tax revenue than did those districts in which the cap was in place for one to three years. Findings regarding the effects of full disclosure laws, as distinct from the effects of explicit restrictions, are more mixed. On the one hand, consistent with the expectation that full disclosure is a nonbinding type of limitation, Robert Bland and Phanit Laosirirat concluded from their analysis of city-level panel data in Texas that truth-in-taxation requirements have little or no effect on real per household property tax burdens.61 Gary Cornia and Lawrence Walters, on the other hand, contended that the Utah full disclosure law strikes a balance between restraint and local discretion by allowing property tax revenues to grow, though less rapidly than would have been expected, given the growth in property values.62 An earlier paper by the same authors suggested that full disclosure in Utah also may have provided a benefit initially unforeseen by legislators; namely, it has promoted uniformity in the administration of the property tax.63 57. Preston and Ichniowski (1991); Joyce and Mullins (1991). 58. Brown (2000). 59. For individual state studies, see Cornia and Walters (2006); Cutler, Elmendorf, and Zeckhauer (1999); Dye and McGuire (1997); Dye, McGuire, and McMillen (2005); Elder (1992); Galles and Sexton (1998, 2000); O’Sullivan, Sexton, and Sheffrin (1995). For crossstate comparisons, see Joyce and Mullins (1991); Mullins and Joyce (1996); Preston and Ichniowski (1991); Sokolow (1998, 2000). 60. Dye, McGuire, and McMillen (2005). 61. Bland and Laosirirat (1997). 62. Cornia and Walters (2006). 63. Cornia and Walters (2005).

128 Brunori, Bell, Cordes, and Yuan effects of tels on government spending. Despite evidence that at least some variants of TELs have had a significant constraining effect on property tax revenue, aside from Poterba and Rueben who found that effective property tax limits are associated with slower wage and employment growths for local government employees, there is little empirical evidence that TELs have had much effect on the overall size of government, as measured by indicators such as the level and growth of general revenue and general expenditure.64 There is, however, evidence that TELs have had a significant effect on the composition and structure of government revenues as well as on the mix of state and local financing of local public services. A number of studies have found that the adoption of TELs has fostered increased local reliance on narrow-based nontax revenues and state-level centralization of school financing.65 Local governments have sought to replace revenue forgone due to TELs with alternative revenue sources such as user fees and charges and with intergovernmental aid from the states, thus eliminating the need to cut down spending. Using multistate cross-sectional data, Joyce and Mullins showed that in response to the recent tax revolt nontax revenue has grown as a share of state and local revenue.66 Shadbegian concluded that “evidence indicates that local governments located in TEL states substitute miscellaneous revenue for tax revenue, but that this substitution is less than dollar-for-dollar. In particular, the point estimates indicate that for each $1 reduction in taxes per capita, there is a corresponding $0.27 increase in miscellaneous revenue per capita.”67 In addition, Joyce and Mullins found that states with TELs increased their aid to local governments, which led to the shifting of responsibility for certain expenditure functions from local government to state government.68 Joyce and Mullins confirmed these findings in a 1996 study with updated data.69 Gary Galles and Robert Sexton’s comparative analysis of California and Massachusetts reported that local revenues and expenditures were initially reduced by the limitations but that state and local governments soon made up lost revenues through increased use of fees and charges and were able to exceed 64. Poterba and Rueben (1995); Cutler, Elmendorf, and Zeckhauer (1997); Galles and Sexton (1998, 2000); Joyce and Mullins (1991); Mullins and Joyce (1996). 65. Dye and McGuire (1997); Galles and Sexton (1998, 2000); Joyce and Mullins (1991); Mullins and Joyce (1996); Shadbegian (1999); Sokolow (1998, 2000). 66. Joyce and Mullins (1991). 67. Shadbegian (1999, p. 233). 68. Joyce and Mullins (1991). 69. Mullins and Joyce (1996).

Tax and Expenditure Limitations

129

their prelimit real per capita revenues and expenditures in 1990.70 Mark Glickman and Gary Painter also have found that there is a positive relationship between the presence of TELs and a state’s adoption of a lottery.71 The authors’ explanation is that the median voter may strategically use the combination of limitation and state lottery to lower his or her tax burden without affecting the desired level of public services, an interpretation consistent with the various voter self-interest models of support for TELs. Another fiscal response to the presence of TELs has been the replacement of lost property tax revenues with grants from higher levels of government, associated with increases in state aid to local governments and in the share of state taxes in total state-local revenue.72 The trend is most pronounced in the financing of K–12 education, which traditionally has depended heavily on the local property tax as a revenue source. Public finance experts have expressed their concerns over the viability of nontax revenue and state aid as a long-term substitute for property tax revenues forgone under TELs. Unlike broader-based property taxes, user fees and charges have very limited revenue-generating capacity.73 Moreover, the shift to nontax revenue may change fundamentally the way in which public services are financed. While property tax revenues go into a general fund for general services, user fees and charges are collected to cover the costs of specified services and are imposed on those who consume those services. Thus, even when TELs do not change the amount of taxes collected, they do change how specific goods and services are financed. Over time, such a shift toward more dedicated sources of revenue tied to the provision of specific local public goods and services has the potential to alter the mix of goods and services provided by local government. When TELs shift greater reliance from localities to states, a potential risk is that this revenue source is highly susceptible to substantial cuts at times of an economic downturn. In addition, state centralization of traditional local expenditure functions, such as elementary and secondary education, undermines local fiscal control and increases the distance between the locus of responsibility for delivering public services and the population to be served.74 Since state government is less responsive to diverse local preferences, the shift to state finance of local services reduces the potential gains in efficiency to be had from providing local public goods in a decentralized manner.

70. Galles and Sexton (1998). 71. Glickman and Painter (2004). 72. Sokolow (1998, 2000). 73. Joyce and Mullins (1991). 74. Joyce and Mullins (1991); Mullins and Joyce (1996); Sokolow (1998, 2000).

130

Brunori, Bell, Cordes, and Yuan

Effects of TELs on Public Services Although the weight of the evidence suggests that local governments may have had some success in replacing revenue forgone because of TELs with other revenue sources, TELs nonetheless have affected revenues available for specific public goods and services. Given that a predictable effect of TELs has been to reduce property taxes, which have traditionally been a main, often earmarked, source of revenue for education, it is natural that empirical studies of the effects of TELs on the provision of local public services have focused on whether and how TELs have affected the provision of public education.75 The general question to be answered is “Does the enactment of TELs sufficiently constrain revenue for (and in some cases expenditure on) local public schools to materially affect the “amount” of public education provided? Answering this question empirically requires that one have an operational measure of educational output or outcomes. Although outcome measures are hard to come by in practice, several empirical studies have made use of quantitative measures of outcomes as measured by indicators such as test scores. David Figlio’s crosssectional analysis suggested that tax limitations worsen student performance on standardized tests in mathematics and various subjects.76 Thomas Downes, Richard Dye and Therese McGuire found a modest negative impact of the tax cap in Illinois on mean math scores of third graders but not on reading scores of eighth graders.77 Conventional wisdom is that per pupil spending, student-teacher ratios, starting teacher salaries, and teacher quality should be related to student educational outcomes. Although these indicators are ultimately measures of inputs rather than outputs or outcomes, other empirical studies have substituted such measures of educational inputs for measures of actual output or outcomes to assess the effects of TELs. One way to gauge the amount of input in public education is to see how much money local governments have invested in public schools and how funds have been spent on different parts of local education budgets. In their study of the limits on property tax revenue in Illinois, Dye and McGuire found that the state-imposed cap had a constraining effect on school district operating expenditures but no effect on school district instructional spending.78 To the extent that operating expenditures are regarded as having more to do with the administra75. Dye and McGuire (1997); Dye, McGuire, and McMillen (2005); Bradbury, Case, and Mayer (1998); Downes, Dye, and McGuire (1998); Downes and Figlio (1999); Figlio (1997, 1998); Figlio and O’Sullivan (2001); Figlio and Rueben (2001); Shadbegian (2003). 76. Figlio (1997). 77. Downes, Dye, and McGuire (1998). 78. Dye and McGuire (1997).

Tax and Expenditure Limitations

131

tive overhead of operating a school system while instructional expenditures are viewed as having more to do with the provision of actual educational output, the authors interpret the differential effect as evidence that TELs improved efficiency by reducing less essential public spending while preserving essential spending. In a subsequent study, however, Dye, McGuire, and McMillen found that the cap reduced operating and instructional spending, especially in the long run, and thus had no differential effect.79 On the basis of this evidence, the authors concluded that while the short-run effect of TELs may have been to enhance efficiency by reducing only the fat, but not the muscle, of public spending on education in Illinois jurisdictions, the longer-run effect may have been to cut both less essential and arguably essential spending. The notion of using the effects of TELs on different components of educational spending to infer whether TELs achieve their stated purpose of trimming only wasteful spending is intriguing. However, Downes and Figlio cautioned against taking for granted the assumed linkage between inputs and outcomes in education.80 In particular, they noted that instructional spending might actually be more susceptible to cuts in jurisdictions in which administrative costs are already at a minimum or in districts that are populated by budget-maximizing bureaucrats who see cutting essential educational spending as a strategy to inflict enough pain to encourage voters to subsequently support overrides of TELs. The differential effect of TELs on different kinds of services is further highlighted by David Figlio and Arthur O’Sullivan, who suggested that local politicians may strategically manipulate the teacher-administration ratio (that is, instructional as opposed to administrative spending) to enlist citizen support for overrides of tax limits.81 The larger the demonstrated loss in public goods (that is, teachers), the more likely citizens are to approve an override.82 Such strategic manipulation of spending reductions in response to TELs makes interpretation of prior studies more difficult. Student-teacher ratios, teacher salaries, and teacher quality are other input measures that have been used to assess the effects of TELs. Using country-wide school district data between 1988 and 1991, Figlio demonstrated that local property tax limitations led to higher student-teacher ratios and lower starting salaries for teachers.83 In a comparative study of Oregon and Washington, Figlio also found that state-imposed limitations significantly increased student-teacher 79. Dye, McGuire, and McMillen (2005). 80. Downes and Figlio (1999). 81. Figlio and O’Sullivan (2001). 82. In the same vein, local politicians may reduce the ratio between uniformed police and police administration in exchange for voter support for limitation overrides. 83. Figlio (1997).

132

Brunori, Bell, Cordes, and Yuan

ratios while the ratio of administrative to educational spending remained unchanged or increased.84 Using test scores and college selectivity as measures of ability, Figlio and Rueben also found evidence that tax limits significantly reduce the average quality of new public school teachers, who may face longer queues for an opening as a result of limitations on education spending.85 Shadbegian found that stringent local limits increase student-teacher ratios only slightly but exert no significant impact on teacher salaries.86 The author cited this as evidence for improvement in efficiency by local governments that are subject to caps.

Effects of TELs on Property Values In principle, TELs should affect property values, with the direction of the effect indicative of the underlying political economy of TELs. In one scenario, there is a Tiebout equilibrium in which voters with varying preferences for public goods, services, and taxes locate in communities that offer a mix of taxes and spending that matches these preferences. In this case, TELs would have the undesirable effect of constraining tax rates below levels truly desired by citizens. Moreover, since housing prices would also reflect basic voter satisfaction with the tax and service mix, artificially constraining the ability of politicians and civil servants to satisfy voter demands would lower property values in constrained communities by making properties less attractive to current and potential residents. If, however, consistent with the Leviathan hypothesis, the tax and spending mix is more reflective of the self-interest of politicians and public employees, TELs would act to bring taxes and spending more in line with residents’ preferences. In these circumstances, TELs would be predicted to have the opposite effect and would raise property values in constrained jurisdictions by making housing more attractive to their residents. A still different set of predictions about TELs and property values is implied by the nonresident taxpayer self-interest model. In that case the value of property other than owner-occupied housing would be predicted to increase since, as Vigdor noted, lower tax rates would shift some of the rents garnered from properties owned by nonresidents away from the public fisc and back to nonresident owners.87 In contrast, the impact of restrictions on resident owner-occupiers is theoretically ambiguous. Resident owners would gain from a reduction in tax rates but would lose from any associated reductions in public services. 84. Figlio (1998). 85. Figlio and Rueben (2001). 86. Shadbegian (2003). 87. Vigdor (2004).

Tax and Expenditure Limitations

133

In view of the potentially important effects of TELs on the value of taxable property, and hence on the tax base of local jurisdictions as well as the relevance of directions and patterns of change in property values for assessing the effect of TELs on voter well-being, surprisingly little empirical research exists on the effect of TELs on changes in local property values. Vigdor’s paper is a notable exception in attempting to provide such estimates. In principle, to estimate the effects of TELs on property values, one should observe the time path of property values in communities affected by a TEL with and without the presence of the TEL. In practice, such a counterfactual is impossible to observe. Vigdor constructed a “synthetic” counterfactual by comparing the time path of property values in Massachusetts communities affected by Proposition 21⁄2 with that of property values in communities in the neighboring state of Connecticut with characteristics similar to those in the Massachusetts communities. With this comparison, he found that the net impact of Proposition 21⁄2 on values of owner-occupied housing was positive. Vigdor also found that the estimated increase in property values was greater in communities that initially opposed the ballot measure: a finding consistent with the hypothesis that it is nonresident property owners who favor statewide TELs rather than residents (who benefit from the ability to export local property taxes to nonresidents).

Distributional Effects of TELs In their assessment of the impact of limitations, a number of researchers also address the incidence of TELs among communities with differing characteristics and among taxpayers in different income classes. For example, David Merriman concluded from his analysis that high tax capacity and low-density communities in New Jersey experienced the most severe spending cuts as a result of the statewide expenditure limit.88 In their examination of the fiscal effects of the Illinois property tax limit, Dye and McGuire found that, in addition to their other results, the magnitude of the cap’s impact varied across different types of jurisdictions, including park districts, fire districts, library districts, municipalities, and school districts.89 Brown found that smaller-sized municipalities were more constrained by the TABOR amendment in Colorado than larger municipalities were, even though the measure applied uniformly to all jurisdictions.90 A few studies have attempted to estimate the effect of TELs on the distribution of tax burdens and public services among different income groups. The findings of these studies are mixed. 88. Merriman (1986). 89. Dye and McGuire (1997). 90. Brown (2000).

134

Brunori, Bell, Cordes, and Yuan

Dennis De Tray and Judith Fernandez drew on the new view of property tax incidence, which holds that the burden of the component of local property taxes that corresponds to the national average effective property tax rate falls on owners of capital, while the incidence of the component that deviates from that national average falls on local immobile land, labor, and consumers.91 Using incidence assumptions consistent with the new view, De Tray and Fernandez allocated local property taxes collected in four California and four New Jersey cities to individual taxpayers before and after the passage of TELs in each state. A key assumption of the authors’ analysis was that what matters for the distribution of relative local tax burdens is mainly the effect of TELs on the deviation of local property taxes from the national average. Using this procedure to allocate and compare local tax burdens before and after the enactment of TELs, the authors found that when the local burden is assumed to be borne entirely by owners of land, the initial effect of adopting TELs in both California and New Jersey was to shift state and local tax burdens from lower- to higher-income households (in part by shifting what remains of state and local tax burdens toward relatively more progressive state income taxes), although the effect was smaller in New Jersey (because of a less progressive state income tax). These findings are robust for alternative shifting assumptions: for example, that local property taxes are also borne by local consumers and labor in the case of California cities but not of New Jersey cities, where TELs have a regressive impact under some alternative local shifting assumptions. Subsequent research on the distributional effects of California Proposition 13 by Arthur O’Sullivan, Terri Sexton, and Steven Sheffrin reached conclusions broadly consistent with those of De Tray and Fernandez with regard to vertical equity among residential homeowners.92 One important feature of Proposition 13 is that it severely limits annual growth in annual assessments to a maximum of 2 percent per year unless the property is sold, at which time its taxable value is reassessed according to actual market conditions. While not quite freezing the assessed value, the result is to base assessments more on acquisition value than on actual market value. In a real estate market such as California’s, the effect of such a system is to benefit less mobile homeowners, who the authors argue are more likely to be in low-income and older households. Of course, a direct corollary, as Sexton, Sheffrin, and O’Sullivan noted, is that, while less mobile households benefit from the fact that their property taxes will rise quite slowly, the fact that the taxable value of property can increase quite significantly upon sale creates a substantial horizontal inequity among households in the same income class depending on when a property is purchased.93 For 91. De Tray and Fernandez (1986). 92. O’Sullivan, Sexton, and Sheffrin (1994). 93. Sexton, Sheffrin, and O’Sullivan (1999).

Tax and Expenditure Limitations

135

example, Sheffrin noted that in Los Angeles County in 1992, 43 percent of homeowners had been in their homes since 1975 and faced an effective property tax rate of 0.2 percent compared with an effective rate of 1.0 percent (five times greater) for newly purchased homes.94 (Downes and Figlio caution, however, that the advantages of remaining “locked-in” to one’s home under Proposition 13 is a somewhat unique feature of this particular form of TEL.)95 In contrast to the studies discussed above, other research finds less progressive effects of TELs. When evaluating the effects of limits on the quality of education, Downes and Figlio found that among communities subject to TELs, economically disadvantaged communities suffered larger reductions in student performance.96 Using a computable general equilibrium (CGE) model to simulate the economic and fiscal effects of Oregon’s Measure 5, Edward Waters, David Holland, and Bruce Weber estimated that higher-income Oregon households garnered disproportionate benefits from Measure 5 in the form of higher factor incomes resulting from reductions in production costs caused by lower property taxes.97 The simulated effect of Measure 5 on the overall progressivity of Oregon’s tax system was mixed, reducing progressivity at the top of the income distribution while increasing it at the bottom.

Conclusions and Directions for Further Research The emergence of the tax revolt in the 1970s and the subsequent support for measures intended to limit local taxes and spending have caused much to be written about such measures. Tax and expenditure limitations restrict local government autonomy when they are legally and administratively structured to be binding. Since local fiscal autonomy is widely recognized to be critically important to the American federalist system, and local governments are seen to be the most efficient and effective methods of providing local services, what should one make of widespread, if not uniform, support for TELs in many communities?

Motives for Supporting TELs The answer may be as simple as that voters suffer from a degree of fiscal illusion and believe that taxes can be reduced without reducing public services. A more complex answer, however, lies in the recognition that even democratic systems of government are human, and hence, imperfect institutions. When citizens believe, for good or ill, that elected leaders and bureaucrats are apt to use the tax 94. Sheffrin (2005). 95. Downes and Figlio (1999). 96. Downes and Figlio (1999). 97. Waters, Holland, and Weber (1997).

136

Brunori, Bell, Cordes, and Yuan

and spending powers of the government to pursue a variety of self-interested objectives, or when one group of taxpayers seeks to insulate itself from paying what it believes to be unjustified taxes imposed by other groups of taxpayers, incentives are created for voters to support the democratic imposition of fiscal restraints on elected officials and bureaucrats.

Effects of TELs on Local Finances and Public Services From the standpoint of both positive and normative urban public finance, the embrace of such democratically imposed limitations by state and local governments raises a number of questions: How have such restrictions operated in practice? How have they affected local public finances? On balance, have such measures had desirable or undesirable effects? The empirical literature on TELs summarized above provides some answers to these questions: —TELs that are legally and administratively structured to be binding are most likely to have measurable effects on urban public finances than are TELs whose constraints are more readily circumvented. —When TELs are binding they have constrained growth in property tax revenue, which has long been the main broad-based source of revenue for local governments. Property tax rate limits that are coupled with assessment limits are particularly binding, resulting in the greatest reduction in the growth of per capita property tax revenue. —Local governments have reacted to such constraints by substituting for property taxes other local, though more narrow, revenue sources such as fees and charges and by increasing their reliance on intergovernmental grants from state government. —TELs have constrained local spending on public schools, as measured by a variety of indicators such as student-teacher ratios, teacher salaries, and teacher quality. —TELs are not only associated with reduced spending on education inputs but also with lower educational outcomes, as measured by test scores. —Evidence from Proposition 21⁄2 in Massachusetts suggests that TELs may have actually raised property and home values in constrained jurisdictions. —Enacting TELs creates both winners and losers among taxpayers. Potential winners include nonresident taxpayers, who include both nonresident workers and abstentee property owners, while residents of communities with the ability to export tax burdens to nonresidents are potential losers. —Evidence on whether TELs favor lower- or higher-income taxpayers is mixed. California’s Proposition 13 may have benefited lower-income homeowners. There is somewhat weaker evidence that the TEL enacted in New Jersey had a similar effect. In contrast, there is also evidence that lower-income communi-

Tax and Expenditure Limitations

137

ties experienced larger reductions in educational outcome from TELs, and Oregon’s Measure 5 was estimated to have benefited higher-income taxpayers and to a lesser extent the lowest-income taxpayers. In terms of the broader themes of this volume, the literature surveyed does not directly address questions such as whether TELs have differential impacts on different types of local areas, such as large cities as opposed to smaller cities, cities as opposed to suburbs, or fiscally healthy communities as opposed to distressed communities. Nonetheless, some speculation is possible. First and foremost, the results indicate that revenue diversification matters. Other things being equal, communities able to draw on revenue sources other than local property taxes have more ability to adapt when the property taxes are subject to TELs. One might conjecture that fiscally healthy communities are also likely to have more diverse revenue sources, and hence, they would enjoy somewhat greater flexibility in responding to TELs. Larger cities are also more likely to have more diverse revenue sources as well, although this advantage might be offset by a correspondingly greater set of spending demands. In addition, to the extent that TELs limit spending on education and that educational spending affects educational outcomes, one might conjecture that TELs are more likely to have a negative effect on education in central cities than they would on education in suburbs. The implicit model supporting this hypothesis is one in which educational outputs depend heavily on family socioeconomic status in addition to public education spending, coupled with the fact that suburban families, on average, generally have higher incomes and more education than do families in central cities. The implication is that school systems in more affluent communities are likely to have a greater ability to absorb the impact of lower spending that may be triggered by TELs. Finally, to the extent that TELs serve as a means whereby communities with locational advantages are able to shift local property tax burdens elsewhere, as argued by Vigdor, TELs create winners and losers among communities.98 Whether, on balance, there are likely to be more losers (and more winners) among central cities than there are among suburbs, however, will depend on the distribution of locational advantages and the abilities to export taxes among the two types of communities.

Future Research These broad findings provide important insights for gauging the broad effects of TELs on urban public finances. Further research, however, is needed to provide more definitive quantitative answers to such questions as how TELs affect urban 98. Vigdor (2004).

138

Brunori, Bell, Cordes, and Yuan

public finances and, more generally, the place of TELs as policy instruments for balancing competing local fiscal goals and interests. empirical evidence on effects of tels. Attempts to quantify the effects of TELs on urban public finances confront the usual challenges associated with establishing whether statistical relationships estimated from nonexperimental data are in fact causal relationships. It is difficult to isolate the effects of TELs vis-à-vis other confounding factors that affect urban public finances and to rule out the possibility that some unobserved characteristics of local jurisdictions are responsible for both the adoption of TELs and changes in local fiscal systems and public services. The research summarized above has employed the usual statistical approaches for addressing the problem including the use of “difference-in-difference” designs; fixed effects estimators to account for unobservable factors; and the use of quasi-experimental control groups, such as the use of Connecticut as a comparison group for Massachusetts by Vigdor. Much of this literature, however, has drawn upon a relatively small number of statespecific or jurisdictional-specific “natural experiments” with TELs. Further replication of these findings using more natural experiments drawn from a larger number of states and types of TELs would be of value, especially if such natural experiments can be shown to deal “decisively” with the problem of inferring causation. distributional effects of tels. There has also been relatively little conceptual and empirical modeling of how TELs affect the distribution of tax burdens. Such distributional effects are apt to vary considerably with the type of TEL and the specific offsetting fiscal response to the TEL by local and state governments. Hence, more analysis of different types of TELs is warranted, drawing on microsimulation modeling of tax burdens along the lines of De Tray and Fernandez and general equilibrium modeling along the lines of Waters, Holland, and Weber. In addition to exploring the distributional effects of TELs across income groups, more analysis needs to be done on the differential spatial consequences of TELs across local governments. For example, the overall fiscal conditions of jurisdictions differ as a result of different revenue structures and expenditure needs. According to a recent survey of city fiscal conditions, 46 percent of central cities indicated they were in excellent or good fiscal condition. This is less than the share of rural areas rating themselves in similar conditions (63 percent), and substantially different from that of suburban areas (67 percent). In central cities the fiscal situation is probably more difficult because they have limited, or declining, revenue bases, but high expenditure needs.99 In rural areas, there are 99. Hoene (2005).

Tax and Expenditure Limitations

139

probably fewer expenditure needs relative to the limited ability of rural areas to generate revenues from own-sources. Suburbs are probably better off because there is a stronger economic base and even fewer expenditure needs. These differences are even more pronounced when the size of the jurisdiction is considered. For example, 63 percent of cities with populations of less than 50,000 indicate that they have excellent or good fiscal conditions, while only 48 percent of cities with populations of more than 100,000 characterize their fiscal conditions as good or excellent. A primary factor in determining the effect of TELs on individual local governments is the composition of local revenues. A recent study of the fiscal capacity of local governments within metropolitan areas documented substantial variation of the relative importance of property taxes to municipal own-source revenues. For example, in the Miami metropolitan area, Miami City relied on property taxes for 54.8 percent of its own-source revenues, while the relative importance of property taxes in suburban jurisdictions within the metropolitan area ranged from 95.9 percent in Golden Beach Town and 71.2 percent in Key Biscayne City to 14.7 percent in Homestead City and 16.4 percent in Aventura City. Similarly, in the San Francisco metropolitan area, local property taxes accounted for 34.1 percent of San Francisco’s own-source revenues but ranged from 79.8 and 83.3 percent of own-source revenues in San Mateo and Marin Counties, respectively, to 10.6, 17.0, and 19.5 percent of own-source revenues in Half Moon Bay City, Burlingame City, and San Anselmo Town, respectively.100 TELs will have substantially different effects across individual jurisdictions because, in part, of the difference in the revenue structures of those jurisdictions. assessing policy options. The fact that a sizable percentage of local officials find some merit in TELs suggests that it may be worthwhile to identify a range of mechanisms for achieving local fiscal restraint and to undertake a more systematic analysis of the alternatives. The decision to enact TELs involves choosing imperfect measures that are meant to check self-interested public spending and taxation but do so at the price of tying the hands of communities that desire to increase levels of spending and taxation to meet genuine local needs. Are there either features that can be incorporated into TELs or alternative institutional arrangements that can be used to achieve the benefits of TELs (that is, restricting public spending and taxation that is not in the broader public interest) at a lower cost (not also limiting spending and taxation that is in the public interest)? TEL Overrides. The empirical evidence summarized above shows that the impact of TELs depends on how binding the constraints actually are. The legislation summarized in appendexes A-1 to A-3 indicates that many states have provi100. Atkins and others (2005).

140

Brunori, Bell, Cordes, and Yuan

sions that allow TELs to be overridden. Combining TELs with meaningful overrides seems to provide a check on local spending while at the same time allowing voters the ability to relax the preimposed constraints on a more targeted case-bycase basis. There is, however, relatively little systematic empirical evidence bearing on the question of how overrides of TELs have been used by voters. One study summarized above found that Massachusetts voters were less likely to support overrides the higher their individual property tax bill was, but this result sheds little light on whether override measures serve their intended purpose of providing a kind of safety valve to the constraints imposed by TELs.101 Better Information, Full Disclosure, and Best Practices. As Howard Wial noted, if the underlying concern motivating the adoption of TELs and similar measures is that local officials need to be held more accountable for spending decisions, perhaps this objective might be achieved more directly by making information to taxpayers more transparent and providing information that would allow local government to be benchmarked against best practices.102 In their theoretical model, Besley and Smart caution that, while increased transparency and benchmarking are shown to have the intended effect of making it easier for voters to discipline self-interested politicians once greater transparency and benchmarking reveal the true costs of providing services, such measures also may have the unintended consequence of increasing incentives for self-interested incumbents to behave irresponsibly “before they are found out.”103 Ultimately, however, the question of whether reforms such as increased transparency and benchmarking have their intended effect must be resolved on the basis of empirical evidence. The findings summarized above from the limited research on the effects of full disclosure measures are mixed, although the generally positive findings of the effects of such provisions found by Cornia and Walters suggest that further study of the effects of such measures would be worthwhile.104 benefit-cost evaluation of tels. Last, as noted by Downes and Figlio, it would be useful to derive quantitative estimates of the net benefits and costs of TELs.105 The framework for such an assessment should be the standard public finance paradigm of estimating the size of tax reductions due to TELs and the reduction in the economic costs (excess burden) of taxation attributable to TELinduced reductions in tax liabilities and then comparing these magnitudes with the economic value of public services forgone due to lower tax revenues. 101. Cutler, Elmendorf, and Zeckhauser (1999). 102. Wial (2004). 103. Besley and Smart (2007). 104. Cornia and Walters (2005). 105. Downes and Figlio (1999).

None None Assessed value cannot increase by more than the greater of 10 percent or 25 percent of the difference between the last year’s assessed value and the current year fair market value Homestead value limited to 5 percent; all other real property capped at 10 percent Assessment increase cannot exceed 2 percent per year Constitution requires residential property to make up no more than 45 percent of total assessed value None None Residential assessments limited to 3 percent or CPI Conservation use property cannot exceed 3 percent year None None 7 percent cap on annual equalized assessment valuation None Increase of total statewide assessments cannot exceed 4 percent annually None None None None No statewide limit, but local option limits are permitted None Total assessed tax base cannot exceed the lessor or 5 percent or the rate of inflation without a rate rollback

Alabama Alaska Arizona

Connecticut Delaware Florida Georgia Hawaii Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan

Arkansas California Colorado

Limit

State

Table 4A-1. Assessment Limitations

Michigan Constitution Art. IX, Sec. 3

… … Florida Constitution Art. VII, Sec. 4 O.C G.A. 48-5-7.4 and 48-5-269 … … 35 ILCS 200/15-170; expires in 2006 … Iowa Code Sec. 44121 (4)-(5) … … … … Maryland Statutes Title 9, Sec. 9.105

Arkansas Constitution Amend. 79 California Constitution Proposition 13 Colorado Constitution Art. X

… … Arizona Statutes 42-13301

Citation

Limit

North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island

Minnesota

None None None Assessment cannot increase more than 5 percent per year unless title to property is sold Taxable assessed value cannot increase more than 3 percent each year None None

The value is limited to the greater of: a) 15 percent increase over last year’s limited value or, b) 25 percent of the difference between this year’s estimated value and last year’s limited value. This only applies to agricultural, residential, timberland, or noncommercial seasonal recreational residential (cabins) property Mississippi None Missouri None Montana None Nebraska None Nevada None New Hampshire None New Jersey None New Mexico 3 percent cap on residential property unless title changes New York Nassau County: property cannot increase more than 5 percent per year

State

Table 4A-1. Assessment Limitations (continued)

… … … … … … … New Mexico Statutes 7:36, Sec. 7-36-21.2 Authorized by NYSA Chapter 50-A Art. 18, Sec. 1805 … … … Oklahoma Constitution Art. 10, Sec. 8B Oregon Constitution Art. XI … …

Minnesota Statutes Chap. 273

Citation

Property value cannot increase more than 15 percent every 5 years None 6 percent cap on assessments of Greenbelt properties Residential assessed value limited to an annual increase of not more than 10 percent None None 1 percent on any reassessment; override by simple legislative vote after public hearing Total assessments for the district cannot increase by more than 25 percent of previous total assessments (special assessment districts only) None None None

…= Not applicable.

West Virginia Wisconsin Wyoming

South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington … … …

South Carolina Constitution Art. 10., Sec. 6 … Tennessee Statutes 67-5-1008 Texas Tax Code Title 1, D 23.23 … … Virginia Code 58.1-3321 Revised Code of Washington 35.85.030

Residential tax rates cannot exceed 1 percent full market value Municipalities: 3 percent of assessed value; second class cities: 2 percent of assessed value Cannot exceed 1 percent of full cash value All governments 5 mills

Alabama

None Municipalities (sliding scale): $0.75–$1.50/$100; county: $0.50/$100; school district: $1.50/$100 Orleans Parish: 7 mills; Jackson Parish: 5 mills None No state limit, but Prince George’s County: $0.96/$100 None All governments: 15 mills; except counties: 18 mills

Kansas Kentucky

Louisiana Maine Maryland Massachusetts Michigan

Iowa

Hawaii Idahoa Illinois Indiana

Cannot exceed 1 percent All governments limited to rate of previous year None Kent County (only): $0.50/$100 of assessed value County and municipal governments: 10 mills Rate for school districts: 20 mills; no limits for independent cities and counties None See note a, attached table Local option Municipal: $0.6667/$100; all other local governments: $0.4167/$100 Counties: 3.5 mills; municipalities: 8.1 mills

California Colorado Connecticut Delaware Florida Georgia

Arizona Arkansas

Alaska

Limit

State

Table 4A-2. Property Tax Rate Limits Override

Majority vote of electorate … … … Can be increased to 50 mils

Majority vote of electorate in special election … Voter approval

… See note a, attached table Referendum None

None Majority vote of electorate … None Majority vote of electorate Majority vote of electorate

Majority vote of electorate None

…

Majority vote in a special election

Citation

Louisiana Constitution Art. 6, Prt. 2, Sec. 26 … … … Michigan Constitution Art. 9, Sec. 6

… Kentucky Constitution Sec. 157

Iowa Code Sec. 331.423; Sec. 384.1

Arizona Constitution Art. 9, Sec. 18 (1) Arkansas Constitution Art. 12, Sec. 4 (municipalities); Arkansas Code 26-25-101 (counties) California Constitution Proposition 13 Colorado Constitution Art. X, Sec. 20(7) … Delaware Statutes Title 9, Chap. 80 Florida Constitution Art. VII, Sec. 9 O.C.G.A. 48-5-8, GA Constitution Art. VIII, Sec. 6 … See note a, attached table 35 ILCS 200/18-125 Indiana Statutes Sec. 6-1.1-18-3

Alaska 29.45.090

Alabama Constitution Amend. 373 of Sec. 217

None Municipalities: $1/$100 assessed valuation; counties: $0.35/$100 assessed valuation; school districts formed of cities and towns: $2.75/$100; all other school districts: $0.65/$100. The maximum number of mills that a governmental entity may impose is established by calculating the number of mills required to generate the amount of property tax actually assessed in the governmental unit in the prior year based on the current year taxable value, less the current year's value of newly taxable property, plus one-half of the average rate of inflation for the prior 3 years. County government: $0.50/$100; schools: $1.05/$100; city government: $0.45/$100; natural resource districts: $0.045/$100; community colleges: $0.08/$100 $0.05/$1 assessed value

Mississippi Missouri

New Hampshire None New Jersey None New Mexico County: $11.85/$1,000; school district: $0.50/$1,000; $7.65/$1,000 New York The New York State constitution sets a limit on the amount raised by the property tax for county purposes. This tax limit is 1.5 percent of the full market value of taxable real estate, averaged over the last 5 years. For municipalities the limit is 2 percent. For New York City the limit is 2.5 percent. North Carolina Counties may levy taxes for several specified purposes up to a combined rate of $1.50/$100 assessed value North Dakota County: 23 mills; various other special levy limits: city: 38 mills; township: 18 mills

Nevada

Nebraska

Montana

Local option

Minnesota

New York Constitution Art. 8, Sec. 10

North Carolina Statutes 153A-149 North Dakota SA 57-15-06; 57-15-06.7; 57-15-08; 57-17-01; 57-15-20

None

Majority vote in referendum Two-thirds vote of government body plus majority vote of electorate

Nevada Constitution Art.10, Sec. 2; NRS 354.5982 … … New Mexico SA 7-37-7

Nebraska Statutes 77.3442

Montana Code 15-10-420

… Missouri Constitution Art. 10, Sec. 11b; Sec. 11C

Minnesota SA Sec. 275.011

School boards can vote in an additional 1 percent over limit, which requires a supermajority Can vote on additional levy; proposal placed on general or special election ballot … … Majority of electorate

Majority of electorate

Referendum with voter approval … Two-thirds vote of electorate; majority of electorate for schools

Mill levy limits: common schools: 5–15; vocational and technical schools: 5; county government 2.5–10; municipal government: 5; special districts: 3–4; assessment districts: 3–10 School property tax rates: 0.5 percent of full market value (FMV); nonschool tax rates: 1 percent of FMV Tax rates cannot exceed 12 mills of market value None None General county: $12/$1,000; snow removal fund: $1.20/$1,000; highway reserve: $1.20/$1,000; courthouse: $0.90/$1,000; agriculture building $0.30/$1,000; fire fighting: $0.60/$1,000; railroad authority: $2.40/$1,000; airport authority: $2.40/$1,000; ambulance district: $0.60/$1,000; water project district: $1/ $1,000; sanitary district: $0.27/$1,000; hospital fund: $0.60/$1,000 None County and municipality: 8 mills; school districts: 13.3 mills County: .0032; library: .001; health: .0004; tort liability: .0001; state A&C: .0003; local A&C: .0002; school district set by legislature; school capital outlay: .0024; school reading program: .000121; city and town: .007; water, light, power: .0008; city library: .001; city tort liability: .0001; special cemetery: .0004; special mosquito: .0004; special fire: .0008; county water: .0008; flood control: .0008; special county service: .0014 None None

Oklahoma

Vermont Virginia

Tennessee Texas Utah

Pennsylvania Rhode Island South Carolina South Dakota

Oregon

1 percent of taxable value

Ohio

… …

… Majority vote of electorate Under limited circumstances

None … … Three-fourths majority of electorate; school district: majority

Double majority

Majority vote of electorate (school districts only)

Override Majority vote of electorate

Limit

State

Table 4A-2. Property Tax Rate Limits (continued)

…. …

… Texas Constitution Art. VII; Art. VIII Utah Code 59-2-908; 53A-17a-135; 10-6-133; 17A-2-222

Pennsylvania.Statutes 53.6917 … … South Dakota CL 10-12-21 et seq.

Oregon Constitution Art. XI

Ohio Revised Code 5705.19; 5705.02; Ohio Constitution Art. 12.02 Oklahoma Constitution Art. 10, Sec. 9; Art. 10, Secs. x-9–x-10B

Citation

Tax rates limited to 1 percent Class one property: $0.50/$100; class two property: $1/$100; class three property: $1.50/$100; class four property: $2.00/$100 1 mill or rate in effect in 1992, whichever is greater County: 1.2 percent of assessed value; city and town: 0.8 percent of assessed value

Cite

21-404 31-3901 46-722 31-1008 31-863 31-3503 63-805 63-805 31-867 31-823 31-822 22-206 36-1702

Function limit

Airport: .0004 Ambulance: .0002 Armory construction: .0002 Building funds: .0006 Charities and indigents: .0010 Medical building and equipment: .0006 Current expenditures fund: .0026 County justice fund: .0020 District court fund: .0004 Fair exhibits: .0002 Fair buildings: .0001 Fair operation: .0001 Fish hatchery: .00005 31-862 25-2401 31-864 31-3613 63-314 39-1334 33-2110a 31-4706 22-2406 63-805 25-2602 40-801 (1)(a)

Cite

Majority of electorate None

Special Tax: .00084 Joint county bridges: .000024 Seeding burned areas: .0002 Solid waste disposal: .0004 Veteran’s Memorial: .00001 Veteran’s Memorial construction: .00005 Warrant redemption: .002 LID guarantee: .0002 Tort insurance premiums: 6-927 Claims or judgments: 6-928

40-801 (1) (b) 40-807 38-509 31-4404 65-103 65-104 63-806 50-1762 None None

Cite

Wisconsin Statutes 59.685 Wyoming Statutes Sec. 39-13-104

Washington Constitution Art. VII, Sec. 2. West Virginia Constitution Sec. 11-8-6; Sec. 11-8-17

Function limit

Three-fifths majority of electorate All limits can be overridden by majority vote of electorate

Health prevention: .0004 Herd: .0002 Historical societies: .00012 Hospital, operational: .0006 Revaluation program: .0004 Sinking fund: .0002 Junior college tuition: .0006 Museums: .0003 Noxious weeds: .0006 Parks and recreation: .0001 Pest fund: .0002 Roads and bridges: .002

Function limit

… = Not applicable. a. Idaho has numerous rate limitations for its county governments' functions.

Wisconsin Wyoming

Washington West Virginia

Kentucky

Iowa Kansas

Indiana

Illinois

Georgia Hawaii Idaho

Florida

Connecticut Delaware

Colorado

California

4 percent annual increase limit for school districts

Tax cap limits levy increases of taxing bodies to lower of 5 percent or rate of inflation Property tax revenue increases are limited to not more than the 6-year annual growth of nonfarm personal income None Property tax revenue is limited to amounts from previous year

Alabama Alaska Arizona

Arkansas

Limit

None No municipality may collect more than $1,500 per resident Property taxes levied by any local government cannot increase more than 2 percent Property tax revenue cannot increase more than 10 percent from previous year Annual appropriations growth linked to population growth and per capita personal income growth Increases cannot exceed the lessor of local growth or inflation or 5.5 percent of the previous year’s revenue (suspended through 2010) None Property tax revenue cannot increase more than 15 percent annually Revenue growth limited to average personal income growth rate over previous five years None None Revenue growth factor capped at 3 percent

State

Table 4A-3. Revenue and Expenditure Limits Override

Local government can appeal through state administration agency … Majority vote of governing body after notice and publication Vote by school board after public notice

… … Additional amounts to be approved by 60 percent of voters in cities; in other taxing districts, must be approved by two-thirds Majority vote of governing body

Majority of electorate

… None

Majority of electorate

…

None

… None Popular vote

Citation

KRS 160.470

… SB45, Sec. 21.

IC 6-1.1-18.5-3

35 ILCS 200/18-55

… … Idaho Statutes Title 63, Chap. 8

Florida Constitution Art. VII, Sec. 1

… Delaware Statutes Title 9, Ch. 80

Colorado Constitution Art. X

California Constitution Proposition 13

Arkansas Constitution Art.16, Sec. 14

… Alaska 29.45.090 Arizona Constitution Art. 9, Sec.19

Nevada Local taxes cannot be raised more than 6 percent annually New Hampshire None New Jersey County property tax revenue cannot increase more than 2.5 percent over previous year's plus COLA. New Mexico Property tax revenue cannot increase more than 5 percent annually New York None North Carolina None North Dakota School district property tax revenue limited to previous year's plus 18 percent. Statute freezes amount of revenue from real property Ohio until the property is reappraised

Nebraska

Mississippi Missouri Montana

Minnesota

Michigan

Local option and property tax revenue cannot exceed previous year's 10 percent cap on increase in all tax collections No net revenue increases Tax revenue cannot increase more than half of average rate of inflation for prior 3 years School district revenue increase limits vary by size of district from 2.5 to 5.5 percent annually

No state limit; but Anne Arundel County: revised limit of 4.5 percent of CPI growth; Montgomery County: revised limit of CPI; Talbot and Wicomico Counties: revised increase limit of 2 percent of CPI. Property tax levies cannot increase by more than 2.5 percent annually No net tax increases without voter approval

Maryland

Massachusetts

Property tax revenue cannot exceed previous year's Property tax levy increases limited to formula based on inflation, assessment growth, and income.

Louisiana Maine

New Mexico SA Sec. 7-37-7.1 … … North Dakota CC Sec. 57-14-14 (law subseqeuntly changed) Ohio Constitution 319.301

None … … Majority of electorate …

Nevada RS 354.5982; 354.59811 … New Jersey SA 40A Sec. 4.45.4

LB 989

Missippi Code Sec. 27-39-320 Missouri Constitution Art. 10, Sec. 22 Montana Codes Annotated, 15-10-420.

Massachusetts Constitution Chapter 59, Sec. 21C (Prop 2 ?) Michigan Constitution Art. IX, Sec. 31 (Headlee Amendment) Minnesota SA Sec. 275.001

School board can vote in an additional 1 percent that requires a three-fourths majority vote Majority of popular vote … Referendum

Majority vote of electorate Majority of voters within jurisdiction Majority of electorate

Referendum

Majority vote of electorate

Majority vote of electorate

Two-thirds vote of governing body Louisiana Constitution Art. 7, Sec. 23 Limit can be overridden by majority MRSA 5721 (LD 1) vote of local government, but 10 percent of electorate can call for referendum … …

None New law limits property tax revenue increases to inflation rate New law lowers the cap on local property tax increases from the current 5.5 percent to 4 percent in fiscal 2013 and reduces the amount school budgets can rise over the previous year's from 5.25 percent in fiscal 2008 to 4 percent in fiscal 2012. The percentage limit by which school budgets can increase will drop by 0.25 percentage point per year until reaching 4 percent. None Property tax revenue may not increase more than the lesser of 3 percent or rate of inflation None Property tax revenue limited to previous year's

Oklahoma

Oregon Pennsylvania Rhode Island

… = Not applicable.

Wyoming

Wisconsin

West Virginia

Levies can increase in 2006 and 2007 by either 2 percent or percentage change in municipality’s equalized value due to net new construction, whichever is greater None

None Total revenue capped at 101 percent of previous year’s revenue Property tax revenue limited by formula based on population and inflation Property tax revenue limited to previous year's

Vermont Virginia

Washington

Property tax revenue limited to previous year's

Utah

Tennessee Texas

South Carolina South Dakota

Limit

Spending increases limited to 12 percent adjusted for inflation

State

Table 4A-3. Revenue and Expenditure Limits (continued)

…

By local government vote after notice and hearing Revenue above limit by referendum

… Majority of electorate after two-thirds governing body vote … Vote of governing body after notice and hearing Vote of governing body after notice and hearing … By local government vote after holding a public hearing Majority vote of electorate

Three-fourths majority vote of local legislature … By vote of electorate Four-fifths majority of local legislature

Override Citation

…

Wisconsin Statutes Art. 25 (2005–2007), Sec. 66.0602

West Virginia Constitution Sec. 11-8-6

RCW 84.55

… Code of Virginia Sec. 58.1-3321

Utah Code Sec. 59-2-911.

… South Dakota Statutes Title 10, Chap. 13. Sec. 35 … Texas Constitution Art. 8, Sec. 2

… House Bill 39, signed June 27, 2006 Senate Bill 3050, signed July 13, 2006

Oklahoma Constitution X-23

Tax and Expenditure Limitations

151

References Anderson, Nathan B. 2006. “Property Tax Limitations: An Interpretative Review.” National Tax Journal 59, no. 3 (September): 685–94. Atkins, Patricia S. and others. 2005. “Intrametropolitan Area Revenue Raising Disparities and Equities.” Working Paper 019. George Washington University, Institute of Public Policy (November). Bae, Suho, and Thomas Gais. 2007. “The Effects of State-Level Tax and Expenditure Limitations on Revenues and Expenditures.” Rockefeller Institute Policy Brief. Albany, N.Y.: Rockefeller Institute of Government. Beito, David T. 1989. Taxpayers in Revolt: Tax Resistance during the Great Depression. University of North Carolina Press. Besley, Timothy, and Michael Smart. 2007. “Fiscal Restraints and Voter Welfare.” Journal of Public Economics 91, nos. 3–4: 755–73. Bird, Richard M. 1993. “Threading the Fiscal Labyrinth: Some Issues in Fiscal Decentralization.” National Tax Journal 46, no. 2: 207–27. Bland, Robert L., and Phanit Laosirirat. 1997. “Tax Limitations to Reduce Municipal Property Taxes: Truth in Taxation in Texas.” Journal of Urban Affairs 19, no. 1: 45–58. Bowman, John H., Susan MacManus, and John L. Mikesell. 1992. “Mobilizing Resources for Public Services: Financing Urban Government.” Journal of Urban Affairs 14, nos. 3–4: 311–35. Bradbury, Katharine L., Karl E Case, and Christopher J. Mayer. 1998. “School Quality and Massachusetts Enrollment Shifts in the Context of Tax Limitations.” Federal Reserve Bank of Boston New England Economic Review (July–August): 3–20. Brennan, Geoffrey, and James Buchanan. 1979. “The Logic of Tax Limits: Alternative Constitutional Constraints on the Power to Tax.” National Tax Journal 32, no. 2 (Supplement): 11–22. Brooks, Leah, and Justin Phillips. 2006. “When to Supplement and When to Supplant? Using Community Development Block Grant Program to Explore the Flypaper Effect.” Working Paper. McGill University, Department of Economics, and Columbia University, Department of Political Science. Brown, Tom. 2000. “Constitutional Tax and Expenditure Limitation in Colorado: The Impact on Municipal Governments.” Public Budgeting and Finance 20, no. 3: 29–50. Brunori, David. 2003. Local Tax Policy: A Federalist Perspective. Washington: Urban Institute. Citrin, Jack. 1984. “Introduction: The Legacy of Proposition 13.” In California and the American Tax Revolt: Proposition 13 Five Years Later, edited by Terry Schwadron, Paul Richter, and Jack Citron, pp. 1–69. University of California Press. Cornia, Gary C., and Lawrence C. Walters. 2005. “Full Disclosure: Unanticipated Improvements in Property Tax Uniformity.” Public Budgeting and Finance 25, no. 2 (June): 106–23. ———. 2006. “Full Disclosure: Controlling Property Tax Increase during Periods of Increasing Housing Values.” National Tax Journal 59, no. 3 (September): 735–49. Courant, Paul N., Edward M. Gramlich, and Daniel L. Rubinfeld. 1980. “Why Voters Support Tax Limitation Amendments: The Michigan Case.” National Tax Journal 33, no. 1 (March): 1–20. Cutler, David M., Douglas W. Elmendorf, and Richard J. Zeckhauer. 1997. “Property Tax Limitations in Retrospect: The Example of Massachusetts.” State Tax Notes 12: 771–76.

152

Brunori, Bell, Cordes, and Yuan

———. 1999. “Restraining the Leviathan: Property Tax Limitation in Massachusetts.” Journal of Public Economics 71, no. 3 (March): 313–34. De Tray, Dennis, and Judith Fernandez. 1986. “Distributional Impacts of the Property Tax Revolt.” National Tax Journal 39, no. 4: 435–50. Downes, Thomas A., Richard F. Dye, and Therese J. McGuire. 1998. “Do Limits Matter? Evidence on the Effects of Tax Limitations on Student Performance.” Journal of Urban Economics 43, no. 3 (May): 401–17. Downes, Thomas A., and David N. Figlio. 1999. “Do Tax and Expenditure Limits Provide a Free Lunch? Evidence on the Link between Limits and Public Sector Service Quality.” National Tax Journal 52, no. 1 (March): 113–28. Dye, Richard F., and Therese J. McGuire. 1997. “The Effect of Property Tax Limitation Measures on Local Government Fiscal Behavior.” Journal of Public Economics 66, no. 3 (December): 469–87. Dye, Richard F., Therese J. McGuire, and Daniel P. McMillen. 2005. “Are Property Tax Limitations More Binding over Time?” National Tax Journal 58, no. 2 (June): 215–25. Elder, Harold W. 1992. “Exploring the Tax Revolt: An Analysis of the Effects of State Tax and Expenditure Limitation Laws.” Public Finance Review 20, no. 1: 47–63. Figlio, David N. 1997. “Did the ‘Tax Revolt’ Reduce School Performance?” Journal of Public Economics 65, no. 3 (September): 245–69. ———. 1998. “Short-Term Effects of a 1990s-Era Property Tax Limit: Panel Evidence on Oregon’s Measure 5.” National Tax Journal 51, no. 1 (March): 55–70. Figlio, David N., and Arthur O’Sullivan. 2001. “The Local Response to Tax Limitation Measures: Do Local Governments Manipulate Voters to Increase Revenues?” Journal of Law and Economics 44, no. 1 (April): 233–57. Figlio, David N., and Kim S. Rueben. 2001. “Tax Limits and the Qualifications of New Teachers.” Journal of Public Economics 80, no. 1: 49–71. Fischel, William. 1989. “Did Serrano Cause Proposition 13?” National Tax Journal 42, no. 4: 465–73. Frug, Gerald E. 1980. “The City as a Legal Concept.” Harvard Law Review 93, no. 6: 1057–154. Galles, Gary M., and Robert L. Sexton. 1998. “A Tale of Two Tax Jurisdictions: The Surprising Effects of California’s Proposition 13 and Massachusetts’ Proposition 21/2.” American Journal of Economics and Sociology 57, no. 2: 123–33. ———. 2000. “Computing the Extent of Circumvention of Proposition 13: A Response.” American Journal of Economics and Sociology 59, no. 1: 133-40. Giertz, J. Fred, and Therese McGuire. 1991. “State and Local Imposed Centralization and Local Fiscal Outcomes.” In Proceedings of the Eighty-Third Annual Conference on Taxation, edited by Frederick Stock, pp. 62–67. Washington: National Tax Association. Glickman, Mark M., and Gary D. Painter. 2004. “Do Tax and Expenditure Limits Lead to State Lotteries? Evidence from the United States: 1970–1992.” Public Finance Review 32, no. 1 (January): 36–64. Gramlich, Edward M. 1993. “A Policymaker’s Guide to Fiscal Decentralization.” National Tax Journal 46, no. 2: 229–35. Haselhoff, Kim DeFronzo. 2002. “Motivations for the San Fernando Valley Succession Movement.” Journal of Urban Affairs 24, no. 4: 425–43. Hoene, Christopher. 2005. Local Budget and Tax Policy in the U.S.: Perceptions of City Officials. Research Report on America’s Cities. Washington: National League of Cities.

Tax and Expenditure Limitations

153

Joyce, Philip G., and Daniel R. Mullins. 1991. “The Changing Fiscal Structure of the State and Local Public Sector: The Impact of Tax and Expenditure Limitations.” Public Administration Review 51, no. 3: 240–54. Ladd, Helen F., and Julie Boatright Wilson. 1982. “Why Voters Support Tax Limitations: Evidence from Massachusetts’ Proposition 2 1/2.” National Tax Journal 35 (June): 121–48. Lo, Clarence Y. H. 1995. Small Property versus Big Government: Social Origins of the Property Tax Revolt. University of California Press. McCabe, Barbara Coyle, and Richard Feiock. 2000. “State Rules and Local Fiscal Choices.” Paper prepared for the 96th Annual Meeting of the American Political Science Association. Washington, August 31–September 4. McGuire, Therese J. 1995. Issues and Challenges in State and Local Finance. Washington: The Finance Project. ———. 1999. “Proposition 13 and Its Offspring: For Good or for Evil?” National Tax Journal 52, no. 1 (March): 129–38. Merriman, David. 1986. “The Distributional Effects of New Jersey’s Tax and Expenditure Limitation.” Land Economics 62, no. 4: 353–61. Mullins, Daniel R., and Philip G. Joyce. 1996. “Tax and Expenditure Limitations and State and Local Fiscal Structure: An Empirical Assessment.” Public Budgeting and Finance 16, no. 1 (Spring): 75–101. Musgrave, Richard. 1983. “Who Should Tax Where and What?” In Tax Assignment in Federal Countries, edited by Charles E. McLure, pp. 2–19. Canberra: Australian National University Press. NCSL (National Conference of State Legislatures). 1997. Critical Issues in State-Local Fiscal Policy: Sorting Out State and Local Responsibilities. Denver. Norris, Donald F. 2001. “Prospects for Regional Governance under the New Regionalism: Economic Imperatives versus Political Impediments.” Journal of Urban Affairs 23, no. 5: 557–71. Oates, Wallace E. 1972. Fiscal Federalism. New York: Harcourt, Brace, Jovanovich. ———. 1991. “The Theory and Rationale of Local Property Taxation.” In State and Local Finance for the 1990s: A Case Study of Arizona, edited by Dana Wolf Naimark and Therese McGuire, pp. 407–24. Arizona State University. O’Sullivan, Arthur. 2000. “Limitations on Local Property Taxation: The U.S. Experience.” State Tax Notes 18, no. 2 (May 15): 1697–713. O’Sullivan, Arthur, Terri A. Sexton, and Steven M. Sheffrin. 1994. “Differential Burdens from the Assessment Provisions of Proposition 13.” National Tax Journal 47, no. 4: 721–29. ———. 1995. Property Taxes and Tax Revolts: The Legacy of Proposition 13. Cambridge University Press. Peterson, Paul E. 1995. The Price of Federalism. Brookings. Poterba, James M., and Kim S. Rueben. 1995. “The Effect of Property-Tax Limits on Wages and Employment in the Local Public Sector.” American Economic Review 85, no. 2 (May): 384–89. Preston, Anne E., and Casey Ichniowski. 1991. “A National Perspective on the Nature and Effects of the Local Property Tax Revolt, 1976–1986.” National Tax Journal 44, no. 2 (June): 123–45.

154

Brunori, Bell, Cordes, and Yuan

Richter, Paul. 1984. “A Look at the Revolt.” In California and the American Tax Revolt: Proposition 13 Five Years Later, edited by Terry Schwadron, Paul Richter, and Jack Citrin, pp. 70–97. University of California Press. Sexton, Terri A., Steven M. Sheffrin, and Arthur O’Sullivan. 1999. “Proposition 13: Unintended Effects and Feasible Reforms.” National Tax Journal 52, no. 1 (March): 99–111. Shadbegian, Ronald J. 1996. “Do Tax and Expenditure Limitations Affect the Size and Growth of State Government?” Contemporary Economic Policy 14, no. 1 (January): 22–35. ———. 1999. “The Effect of Tax and Expenditure Limitations on the Revenue Structure of Local Government, 1962–87.” National Tax Journal 52, no. 2 (June): 221–37. ———. 2003. “Did the Property Tax Revolt Affect Local Public Education? Evidence from Panel Data.” Public Finance Review 31, no. 1 (January): 91–121. Sheffrin, Steven, M. 2005. “Property Tax, Real Property, Residential.” In The Encyclopedia of Taxation and Tax Policy, 2d ed., edited by Joseph J. Cordes, Robert D. Ebel, and Jane G. Gravelle. pp. 317–19. Washington: Urban Institute. Sokolow, Alvin D. 1998. “The Changing Property Tax and State-Local Relations.” Publius 28, no. 1 (Winter): 165–87. ———. 2000. “The Changing Property Tax in the West: State Centralization of Local Finances.” Public Budgeting and Finance 20, no. 1 (Spring): 85–104. Temple, Judy A. 1996. “Community Composition and Voter Support for Tax Limitations: Evidence from Home-Rule Elections.” Southern Economic Journal 62, no. 4 (April): 1002–016. Vigdor, Jacob L. 2004. “Other People’s Taxes: Nonresident Voters and Statewide Limitation of Local Government.” Journal of Law & Economics 47, no. 2 (October): 453–76. Wallin, Bruce A. 2004. “The Tax Revolt in Massachusetts: Revolution and Reason.” Public Budgeting and Finance 24, no. 4 (Winter): 34–50. Waters, Edward C., David W. Holland, and Bruce A. Weber. 1997. “Economic Impacts of a Property Tax Limitation: A Computable General Equilibrium Analysis of Oregon’s Measure 5.” Land Economics 73, no. 1 (February): 72–89. Wial, Howard, 2004. Limited Learning: How School Funding Caps Erode the Quality of Education. Harrisburg, Penn.: The Keystone Research Center (May). Wolman, Harold. 1995. “Local Government Institutions and Democratic Governance.” In Theories of Urban Politics, edited by David Judge, Gerry Stoker, and Harold Wolman, pp. 135–59. Thousand Oaks, Calif.: Sage Publications.

5 Preschool Education and Human Capital Development in Central Cities clive belfield

R

esearch on early childhood education has found that high-quality preschool benefits children.1 They are better prepared for school and their academic performance—both test scores and attainment—is enhanced. These educational advantages lead to a more secure economic future in adulthood. Persons with higher attainment are more likely to be employed and earn more when employed, and they are less likely to be in poor health, on welfare, or involved in the criminal justice system. Taken together, these private benefits become public benefits: local communities benefit when children attend preschool. Taxpayers gain because the need for government expenditures is reduced and tax revenues are increased, and society benefits because fewer “bads,” such as ill health or crime, are produced. In this light, one might view investments in preschool as an economic development strategy for local communities. This chapter critically evaluates the extent to which preschool promotes economic development in cities. The focus is on the impacts of preschool on central cities (rather than on metropolitan areas) compared with those on other localities. Mostly, these comparisons are between central cities and the suburbs

The author is grateful for comments from Howard Wial, Tim Bartik, Anna Santiago, Amy Ellen Schwartz, Leanna Stiefel, Elena Safirova, Marge Turner, Hal Wolman, and participants in the Conference on Urban and Regional Policy and Its Effects. The author also appreciated empirical research assistance from Heather Schwartz. 1. The term preschool is used here to mean any educational program for children aged three to five, including pre-K, center-based programs, and nursery schools (but not kindergarten). Preschool is not distinguished from Head Start, although the latter is considered in the discussion of policy options. Early Head Start is not considered, which serves fewer than 70,000 children, nor are parental education programs in which fewer than one-in-ten parents participate. See Barnett and Belfield (2006).

155

156

Clive Belfield

(rural areas enroll relatively few preschool children largely because of transport costs and low population density). Most debate concerning the low quality of public schooling is pertinent mainly to inner-city school systems that are under one government jurisdiction. Also, given the funding and regulation of education in the United States and the division into small suburban school districts and large urban ones, metropolitan areas are not the most pertinent geographical comparison.2 The chapter begins by describing current preschool enrollments. Next follows a review of the research evidence to establish the full set of economic benefits—private, fiscal, and social—that may be generated by preschool. Policies to expand preschool in terms of enrollment and program quality as well as financing are then considered. These potential expansions allow for speculation as to whether preschool would significantly contribute to the economic development of cities. Broadly, preschool, by raising human capital formation, should yield net economic benefits to cities. This conclusion is compelling if greater investments in preschool are funded at the federal level. Therefore, the discussion focuses on expansions that would have to be funded locally. Under this funding scenario, the conclusions are more cautiously optimistic.

Enrollments in Preschool Preschool participation has grown significantly in recent decades. As of 2005, more than two-thirds of four-year-olds attend some form of preschool. Also, two-fifths of three-year-olds are enrolled, and almost all children now attend kindergarten. Some of these preschool children are served by public programs. Head Start, which is federally funded with eligibility determined by poverty status, serves about 12 percent of all four-year-olds. In addition, state preschool programs serve approximately 17 percent, with enrollment similarly restricted to four-year-olds who are identified as being poor or at risk (with an additional 6 percent of all four-year-olds in special education).3 The largest group, almost 35 percent of all four-year-olds, is in locally funded programs or in private centers. Preschool coverage depends on a medley of programs. Enrollment is determined by confounding factors of family circumstances and the practicalities of policy implementation across states and districts within states. In relation to per2. A significant amount of research uses metropolitan statistical areas (MSAs) as the indicator of urbanicity. Rather than ignore this literature, it is used as appropriate when city data are absent, while acknowledging that there are some salient differences between central city and metropolitan descriptors. 3. In addition, approximately 0.25 million four-year-olds received support from the federal Child Care and Development Fund in 2004. All figures in this paragraph are from Barnett and Belfield (2006).

Preschool Education and Human Capital Development

157

sonal family circumstances, surveys from the National Household Education Surveys Program (NHES) show significant differences in preschool participation by income, parental education, and ethnicity. Participation rates are broadly flat up to the level of median income, and then they increase with higher family incomes. Children whose mothers are high school dropouts participate at a rate of 49 percent, which is significantly below the 65 percent rate for children of high school graduates and the 70 percent rate for college graduates’ children. Preschool participation rates also vary by ethnicity. African American children have the highest rates, followed closely by White and Asian children; Latino children have by far the lowest rates. Preschool policies are typically determined at state level, but districts vary in how they implement these policies. State enrollments vary significantly: Ten states have no program, and three have near-universal programs (Oklahoma, Georgia, and Florida). States also vary in how they utilize federal grants from the Child Care and Development Fund. Also, although Head Start targets children in poverty, many potentially eligible children are not served. Current U.S. public policy targets children from economically and educationally disadvantaged backgrounds for participation in preschool programs. But current programs fail to enroll even half of the children in poverty or half of the children whose mothers are high school dropouts. Publicly funded programs at best are compensating therefore for low family incomes rather than offsetting the advantages of wealthier families. However, state investments in preschool are growing; it is therefore difficult to give a fixed picture of who is being served. New enrollees are probably from the third-highest quartile of family incomes, as these families were unable to afford private preschool but ineligible for means-tested programs.4 Overall, these factors make it hard to identify a strictly city-specific aspect to preschool coverage and to conclude whether there is a greater need for preschool in cities. Although unadjusted preschool enrollment rates are higher in metropolitan areas than in nonmetropolitan areas, data for central cities are not available.5 Moreover, there are important differences in family characteristics and government policies between central cities and suburbs, so it is not possible to generalize from preschool enrollment in metropolitan areas to enrollment in 4. Information for this paragraph is from Barnett and Belfield (2006) and the National Institute for Early Education Research yearbooks (www.nieer.org). 5. The NHES 2005 survey data show center-based preschool rates (public and private) for three- and four-year-olds that are 20 percent higher for families that reside in metropolitan areas. SIPP (Survey of Income and Program Participation) data show a similar pattern. As reported by Blau and Currie (2006, table 1), rates of center-based care are higher for metropolitan residents, but the overall patterns for employed mothers are very different from those of nonemployed mothers. Employed mothers in metropolitan areas rely more on centerbased care or parent care and less on nonrelative care. Nonemployed mothers in metropolitan areas rely more on center-based care and less on relative care. See Kimmel (2006).

158

Clive Belfield

central cities. Many of the African American children eligible for Head Start live in urban areas. Yet many Latinos live in the western states, which (even in large cities) are less likely to offer statewide programs. It is also difficult to identify the counterfactual to preschool. As noted above, one-third of four-year-olds are not in preschool, and yet some of these children will be receiving comparable or superior support for development. Some of the children in centers are deprived of family resources and thus may be at a disadvantage. Crucially, preschool quality is uneven: the programs vary in their goals, resources, standards and regulation, and by length of day and year. Children are often in “blended” or unregulated programs; some receive funding from more than one source. Also, three-year-olds enroll at much lower rates than four-yearolds do.6 Thus, even if enrollment was equally distributed across cities and suburbs, differences in program quality may make this parity moot. Notably, urban K–12 public schools post much lower academic performance than suburban ones do; possibly, a similar disparity exists at the preschool level. However, differences in the observable characteristics of preschool programs can be linked to differences in outcomes. Therefore, it is possible to identify programs that genuinely are higher quality. Teachers with more education and better training provide higher quality preschool. Programs with lower child-tostaff ratios are better. Programs with more intensive learning and of longer duration are better. (However, stricter regulations do not necessarily mean better quality.)7 Potentially, new investments can be made in programs that promote child development. The real extent to which urban children are at a disadvantage relative to nonurban children in terms of preschool provision is therefore unknown. Nonetheless, as shown below, there is reasonable evidence that preschool enrollment in cities is suboptimal. Thus there is some justification for further investment in preschool: either by expanding the numbers served or by upgrading program quality for those already enrolled.

The Economic Benefits of Preschool Preschool programs appear to be strongly beneficial for the individual, in the short term and throughout one’s lifetime. Preschooling also conveys a diverse set of academic and behavioral advantages. In reviewing this evidence, the focus is on the average effect and the differential impact for children who live in cities. The evidence that early education interventions yield immediate academic benefits is near conclusive. Meta-analysis of the research indicates an average ini6. Barnett and Belfield (2006). 7. Full citations are given by Gormley (2007).

Preschool Education and Human Capital Development

159

tial effect on cognitive abilities of about 0.50 standard deviation. Importantly, only center-based preschool programs in which children attend classrooms or individual tutoring sessions improve cognitive development (poorly regulated child care programs are not found to improve child development, and Head Start produces only modest gains).8 “Typical” center-based child care improves cognitive abilities by about 0.10 to 0.33 standard deviation.9 But the ChildParent Centers in Chicago and the universal preschool program in Tulsa—both rigorously evaluated—found initial increases in cognitive and language abilities ranging from 0.38 to 0.79 standard deviation. Both programs were situated in cities, which suggests that the average preschool effect may apply to cities.10 In fact, direct evidence from the Early Childhood Longitudinal Study supports the argument of greater benefits for city residents.11 Table 5-1 reports on effect size gains (improvements measured in standard deviations) from attending a center-based preschool with the population divided according to whether the child resides in an urban, suburban, or rural area. Separate estimations are performed for children when they are in first, third, and fifth grade.12 The goal is not to precisely identify the impact of preschool but simply to see whether there are differential impacts across areas. The results in table 5-1 indicate that there are differential impacts by residence location. For children in urban areas, preschool conveys an effect size gain of approximately 0.08 to 0.10 standard deviation in reading and math; this effect size is statistically significant and is maintained throughout first, third, and fifth grades. Suburban children, in contrast, show smaller academic gains from attending preschool, and these gains are dissipated by fifth grade. For rural children, the contrast is even starker: there are no academic gains from attending preschool. Also, the effect on grade retention is strongest for urban children: attending preschool significantly reduces the probability that these children are retained in grade; the effect is weaker for suburban children and nonexistent for rural children. 8. For Head Start, recent evidence shows fairly small effects, from less than 0.10 to 0.24 standard deviation, for language and cognitive abilities after one year of provision. Antisocial behavior does appear to be reduced, however. See Garces, Thomas, and Currie (2002); Currie (2007). 9. See Barnett and Belfield (2006) on preschool effects and NICHD (2003) for child care effects. 10. See Reynolds and others (2002); Gormley and Gayer (2005). 11. Early Childhood Longitudinal Study (1998), Institute of Educaton Sciences, National Center for Education Statistics, U.S. Department of Education. 12. A full array of control variables is included. Details are available from the author. A similar approach, but not split by area, is reported in Magnuson, Ruhm, and Waldfogel (2007).

160

Clive Belfield

Table 5-1. Effect Size of Attending Preschool, by Area a Grade Reading Grade 1 Grade 3 Grade 5 Math Grade 1 Grade 3 Grade 5 Ever retained in grade By grade 3 By grade 5 N (reading grade 1)

City

Suburban areas

Rural areas

0.097 *** 0.075 ** 0.089 ***

0.035 ** 0.070 ** 0.023

–0.017 0.054 –0.070

0.096 *** 0.085 *** 0.096 ***

0.046 * 0.052 ** 0.052

0.003 0.054 –0.058

–0.37 ** –0.32 * 4,504

–0.39 ** –0.08 4,642

0.12 0.04 2,707

Source: Early Childhood Longitudinal Study Kindergarten Cohort (1998). ***Significant at the 1 percent level; **significant at the 5 percent level; *significant at the 10 percent level. a. Improvements are measured in standard deviations. Ordinary least squares estimations are unweighted. Estimations control for sex, race, weight and height, birth status, region, mother’s age and educational attainment, socioeconomic status, family structure, and activities. Details available from the author.

Other educational benefits derived from preschooling may prove more important than these test score gains. The results from high-quality studies and from Head Start are summarized in table 5-2. These show significant reductions in the rate of placement in special education and in grade retention, much lower high school dropout rates, and much higher rates of high school completion and college progression.13 Consequently, even if the academic advantages of preschool fade out, there are long-term consequences. Plausibly, preschool program interventions may have greater effects on these outcomes in cities simply because more children could benefit. Across the United States, rates of special education and grade retention are higher in cities, but the most striking difference is in high school dropout rates. In urban areas, only 58 percent of public school students graduate on time; the rates for suburban, small town, and rural areas are 73 percent, 69 percent, and 72 percent, respectively.14 Such greater prevalence may make preschool more beneficial. 13. Intensive interventions had twice the effect in reducing grade repetition (in twentyfour studies) and four times the effect in reducing special education placement (in twenty studies) as did Head Start and public school programs (Barnett and Ackerman 2006). For information on many of the benefits, see Barnett and Belfield (2006). 14. For special education, grade retention, and high school completion, see respectively Cullen (2003); Eide and Showalter (2001); Swanson (2004).

Preschool Education and Human Capital Development

161

Table 5-2. Impact of Participation in Early Childhood Programs on Education Percent (unless otherwise noted) Outcome Special education placement Abecedarian ECI Perry Preschool Chicago Child-Parent Centers Head Start Public school and Head Start Retained in grade Abecedarian ECI Perry Preschool Chicago Child-Parent Centers National data (ECLS-K) Public school and Head Start High school dropout likelihood Abecedarian ECI Perry Preschool Chicago Child-Parent Centers High school completion Head Start : White children Head Start: African American children College progression Abecedarian ECI Perry Preschool Head Start : White children Head Start: African American children

Effect –48 –43 –32 –28 –29 –47 –13 –33 Negative effect (reduces) –30 –32 –25 –24 20 percentage points increase No clear effect Three times as likely No clear effect 28 percentage points increase No clear effect

Source: See Barnett and Belfield (2006).

There are also positive behavioral impacts from attending preschool. These impacts can be identified directly. Across studies of early education intervention, intensive research programs, and large-scale public programs including Head Start, short-term gains in effect size average 0.25 to 0.40 for self-esteem, problem behavior, and other social behaviors. These behavioral advantages may serve as a mediating factor in generating advantages during childhood and during adulthood.15 Indeed, the gains from preschool appear strong in adulthood. Table 5-3 summarizes the impacts of high-quality preschool programs and Head Start. Individuals who were enrolled in a preschool program report lower rates of teenage pregnancy, higher personal ratings of well-being (for example, fewer health problems and less drug usage), and lower rates of involvement in the criminal 15. Barnett and Ackerman (2006).

162

Clive Belfield

justice system. Again, although these actions are not specific to residents of cities, their higher incidence in urban areas is well documented. Greater prevalence might mean more powerful benefits from preschool in cities. Finally, there are income advantages from attending preschool. The income gains might be inferred from evidence of improved academic performance— particularly high school graduation—for preschool children. But there is also direct evidence from studies with long-term follow-up of participants. As shown in table 5-4, the average net income gain for preschool children is significant. (Head Start does appear to increase earnings, but only for White children of high school dropouts). The income gain for the model programs is expressed as a present value at the age of four, that is, it represents the immediate income value of preschool. Thus far, the benefits of preschool have been expressed in terms of the private gains to the participants. But these private gains result in public benefits. Indeed, the interest here is not in the private advantages of preschool per se but only in how these private gains translate into savings for taxpayers and for society. It is from the perspectives of these last two entities that public investments should be appraised. Expanding preschool programs is likely to yield sizable fiscal benefits to taxpayers and societal benefits for city residents.16 Under many scenarios, these benefits typically exceed the public costs of providing preschool. For the taxpayer, the fiscal benefits are grouped into four domains. First, there are efficiency gains to the school system. Public expenditures on special education and grade retention are reduced. With more proficient (high-achieving) students, schools can provide education more effectively and efficiently. Higherachieving students reduce teacher turnover and absenteeism and allow for better use of instructional time. Second, there are increases in tax revenues as preschool children grow up to be more productive and earn more when they enter the labor force. Also, preschool helps families meet their child care needs, freeing up parents to participate in the labor market, further increasing the tax base. Third, spending on criminal justice is lower as crime is reduced. Fourth, there are savings to health and welfare budgets. Preschool programs provide important health and welfare services (such as health screening and nutrition guidance). Such preventative treatments are cost effective. Public spending on these welfare programs and on any subsequent curative interventions is therefore offset. The return to a local society or community includes, but is not limited to, the taxpayer benefits. A diverse set of benefits should accrue to all city residents 16. Evidence is given in Belfield (2006, 2007); Belfield and Schwartz (2007); Hawkins and others (2005); Cunha and others (2005); McCarton and others (1997); Pratt and Kay (2006); Schulman and Barnett (2006); Warner (2006). On the social benefits, see Karoly and Bigelow (2005).

Preschool Education and Human Capital Development

163

Table 5-3. Impacts of Early Childhood Interventions on Adolescent and Adult Behaviors Percent (unless otherwise noted) Behavior category

Control or Group receiving early comparison group childhood program

Teenage parenting rates, by program Abecedarian ECI 45 Perry Preschool 37 Chicago Child-Parent Centers 27 Issues of well-being Health problema 29 Drug userb 39 Needed treatment for addictiona 34 Abortiona 38 Abuse and neglect by the age of seventeenc 9 Criminal activity Average number of violent felony assaults committeda 0.37 Juvenile court petitionsb 25 Booked and charged with a crimed

26 26 20 20 18 22 16 6 0.17 16 12 percentage points lower

Source: Barnett and Belfield (2006). a. Perry Preschool Program. b. Abecedarian Early Childhood Intervention. c. Chicago Child-Parent Centers. d. Head Start.

Table 5-4. Net Earnings Gain from Participation in Early Childhood Program Dollars Program Abecedarian ECI Perry Preschool Chicago Child-Parent Centers Head Start

Gaina 35,531 38,892 30,638 No effect

Source: Barnett and Belfield (2006). a. Gain is expressed as a present value at the age of four; that is, it represents the immediate income value of preschool.

who interact with the residents who were preschool children themselves. First, there are economic spillover benefits. There is evidence, for example, that more productive coworkers raise each other’s own productivity and that the economies of regions with higher densities of educated workers grow faster.17 If 17. For a discussion of the issues, see Taylor (1999). For empirical evidence, see Berry and Glaeser (2005); Shapiro (2006); Wheeler (2006). This evidence comes from studies of metro-

164 Clive Belfield preschool-educated children in central cities remain in central cities when they become working adults, or if central cities can attract workers who attended preschool elsewhere, then they may be able to reap these benefits. Second, city residents benefit from a lower crime rate. They are less likely to be victims of crime, they need to spend less on avoiding being the victim of crime, and they will have to pay less for insurance against possible crime. Finally, there are reductions in the economic distortion imposed by collecting tax revenue to pay for public services (such as welfare assistance and policing).18 In summary, the gross individual, fiscal, and societal benefits of preschool are likely to be significant. The net benefits, expressed in present values, should also be positive. Certainly, the net benefits to individuals should be positive if any public subsidies are available. But the net benefits to the taxpayer are also likely to be positive and, as consequence, the net benefits to society should be as well. Full economic evaluations based on follow-up of preschool participants bear this conclusion out: the benefit-cost ratios easily exceed 1. The High/Scope Perry Preschool program has generated public benefits of $12.90 per $1 invested. For the Chicago Child-Parent Center program and the Abecedarian Early Childhood Intervention, the ratios are $6.87 and $2.69, respectively.19 There is a lag between the investment and the returns, but these ratios are based on present value calculations. They are also sufficiently large that they are not reversed under sensitivity testing. However, this evidence is about the average impact of preschool and not the marginal impact of expansion. Yet, it is the latter on which policy decisions should be calibrated. Conventionally, marginal benefits are held to be smaller than the average benefits: children on the margin are anticipated to gain less from preschool (otherwise they would already be enrolled). Also, marginal costs of expansion are assumed to exceed the average costs (because hiring more teachers will require paying them more). Hence, the available evidence may overstate the benefits and understate the costs. Again, the significance of this argument depends on the specifics of the preschool program. For example, ten states do not offer preschool to any children; and so the marginal child in Indiana (a state with no public preschool) might politan areas, not central cities; but to the extent that the economic spillovers arise because of spatial proximity, it would be expected that these effects apply to cities as well. However, recent evidence from metropolitan areas suggests that the spillover effects may be small or hard to identify (Rausch and Negrey 2006). 18. Taxes induce consumers to shift their purchases away from taxed goods and services. When a good or service does not impose a cost on others (for example, causing pollution), this shift distorts the pattern of consumption and production in the economy. See Allgood and Snow (1998). 19. Temple and Reynolds (2007); Belfield and others (2006).

Preschool Education and Human Capital Development

165

benefit considerably more than the average child in Oklahoma (where 70 percent of children are in preschool). Also, marginal costs might be close to average costs under some scenarios. In cities with low enrollment, the average cost of preschool might fall if a district can exploit economies of scale. Of course, if preschool programs are funded at the federal level, then the net benefits for cities will be substantially higher. Plus, given the magnitude by which the anticipated benefits currently exceed the costs, there would need to be a significant divergence between the evidence base and the consequences for the marginal child such that the general policy conclusion would be overturned. Important, as noted above, the benefits of preschool are likely to be stronger if account is made for economic conditions in cities. Public spending on special education, grade retention, and remedial education is higher in cities, as is the dropout rate. Crime and welfare dependency are higher. The densities of cities may intensify human capital externalities that can arise when preschooleducated people work in cities. On this logic, the average benefit of preschool, even as it is suggestive of greater investments in preschool, is probably an understatement for cities.

Preschool and Urban Development Given the promise of preschool as a public investment, the chapter now directly assesses whether preschool can play a meaningful role in urban development, specifically in terms of building human capital and growing a competitive economy. Related benefits may also be considered, such as whether preschool will help create quality neighborhoods for families. To repeat, the discussion focuses primarily on decisions to invest in preschool made by local authorities, using (primarily) local funds.

Building Human Capital and Growing a Competitive Economy Expanding preschool programs will be an effective way to increase the stock of human capital in a city. As noted above, there is compelling evidence that children who attend preschool will accumulate more human capital for adulthood, and in the short run, families will increase their own labor market attachment. Simple prevalence differences should make preschool more beneficial in urban communities. Moreover, in light of the low educational attainment by students in urban public schools and the absence of alternative reforms for improvement, investments in preschool to raise the rate of high school graduation should easily pass cost-benefit tests. Certainly, it will take a long time for the flow of augmented human capital from preschool to reach the labor market (and even longer for that flow to influ-

166

Clive Belfield

ence the human capital stock). But the net present value of preschool investments is sufficiently high that the payoff is likely to be strongly positive. Recent calculations indicate the economic value of more education. In present values, from the age of four the lifetime fiscal cost per high school dropout is $120,600.20 This figure reflects the loss in tax revenues and the increase in spending on health, crime, and welfare as a result of failing to graduate from high school. It is probably an underestimate because it assumes that future cohorts of dropouts will be no worse off than current cohorts. Given the high school dropout rate in cities is at least 30 percent and that most states spend less than $7,000 per child on preschool, there should be high returns to investments in any programs that demonstrably reduce the dropout rate.21 Nevertheless, cities are not typically characterized by their investments in high-quality public education. It is necessary, therefore, to consider the specific characteristics of cities that militate against public preschool provision. First, urban demography is important. Cities are disproportionately composed of single persons and households with no children. This suggests that educational policies are unlikely to have a significant impact on urban development and that cities will not have strong political constituencies in favor of public investments in preschool. Yet, this disproportion should not be overstated. Of the 86.9 million central city households across the United States, only 35 percent are “nonfamily households” (that is, a household of an individual or a shared household); 31 percent include children under the age of eighteen. Just over half of all central city residents aged sixteen and older are married.22 Thus cities do host large numbers of families with school-aged children, even as their representation is greater in the suburbs. Of course, it is not clear whether the demography is driving central city education, that is, young taxpayers place a low value on public schooling, or central city education is driving demography, that is, families leave the city for better public schools outside the city. But structural factors are likely to play a big role in terms of differences in housing stock, amenities, and lifestyles. Preschool may be influential at the margin, but preschool is unlikely to create a significant change in city demography unless it changes the entire education system. A related aspect of urban demography is the relatively high mobility of city residents. Migration is not a concern if the preschool program is funded at the federal level, but it becomes more salient the geographically smaller the funding

20. Levin and others (2007). 21. Spending figures are from the National Institute for Early Education Research yearbooks. 22. These statistics are from the 2005 American Community Survey, U.S. Census Bureau.

Preschool Education and Human Capital Development

167

authority or jurisdiction is. So, if cities invest in education but then experience out-migration, the benefits of the investment are attenuated. The share of central city residents who move to the suburbs each year is less than 6 percent.23 Because mobility rates for families with children are smaller than the average mobility rate across the population by 20 percent, the share of central city families with children who move to the suburbs is less than 5 percent per year. Of course, some families leave the central city for nonmetropolitan areas, and these families are not included in this estimate. Families with children enrolled in preschool may be more or less likely than other families to leave central cities. But the attenuating effect of migration should not be overstated, not least because people move out and back in over time and because many persons may migrate within the same metropolitan area.24 Furthermore, because most mobility is driven by job search and household formation, many migrants intermittently return to, or commute into, cities and metropolitan areas. The only study that has directly accounted for mobility is by Lynn Karoly and James Bigelow.25 Their cost-benefit analysis for California, a state with relatively high mobility, indicates that migration might offset one-quarter of the returns on investment in preschool. This offset is not sufficient to overturn their conclusion concerning the net benefits of preschool at the state level. Conventional measures of urban vitality lend limited support to investments in preschool. Typically, the vitality of an urban economy is measured in terms of population growth, wages, and land prices. For population growth, dramatic 23. This estimate was derived as follows. According to Current Population Survey data, 14 percent of persons move within the United States each year. (This figure is for 2002, which is the most recent date for available data because of revisions by the U.S. Census Bureau to the method for calculating mobility.) About 13 percent of movers leave any central city for the suburbs each year, so the share of the total population that consists of central city residents moving to the suburbs is 13 percent times 14 percent, or about 1.8 percent annually. According to Census 2000, the U.S. metropolitan population in 2000 was 225,981,679, and the total U.S. population was 281,421,906 (see “Geographic Comparison Table: United States and Puerto Rico: 2000” [http://factfinder.census.gov/servlet/GCTTable?_bm=y&-geo_id= 01000US&-_box_head_nbr=GCT-PH1-R&-ds_name=DEC_2000_SF1_U&-format=US10S]), while the suburban population was 140,645,981 (derived from census data in State of the Cities Data Systems [http://socds.huduser.org/Census/totalpop.odb?msacitylist= 0.0*0*1.0&metro=msa]), so the central city population was 85,335,698 (the metropolitan population minus the suburban population). Therefore, the number of central city residents who move to the suburbs each year is about 1.8 percent of 281,421,906, or about 5,065,594, and the share of central city residents who move to the suburbs each year is that number divided by 85,335,698, or slightly less than 6 percent. 24. Also, the migration that could result from expanding the availability of preschool could have a different effect depending on whether one city offers preschool or all cities do. If only one city offers preschool, it may lose out: families may take advantage of preschool and then migrate to another city. The long-term benefits of preschool would then accrue to the destination city with the costs paid by the origin city. 25. Karoly and Bigelow (2005).

168

Clive Belfield

effects from preschool might not appear. Edward Glaeser and Jesse Shapiro described city growth as broadly inert, within the context of population growth and migration during the late twentieth century toward places with good weather and low density.26 Yet, rising educational attainment fuels population growth as migrants move to places with skilled inhabitants. Rising attainment also fuels employment growth.27 However, rising land prices may put a check on education-induced growth in cities and metropolitan areas. Glaeser and Gottlieb estimated that a 10 percent increase in college-educated workers within a metropolitan area generated a 0.7 percent increase in house values and rents, that is, house prices rise three times faster than wages.28 Although the amount of the housing price increase may differ for central cities, this general result may apply to cities. If preschool education increases the share of college-educated workers in a city, then it may increase housing prices there. The increase in housing prices may impede urban economic growth.29 Successful urban industries are likely to be those with relatively high land use productivity (for example, a lawyer’s office rather than a car factory). Speculatively, preschool (or any other educational enterprise) probably has a land use productivity that is below the city-wide average. Preschool children need space to learn as well as to play, and preschool facilities cannot be used intensively. Also, preschool is a service, not an amenity. Unlike parks and restaurants, which are valuable even to those who use them only very infrequently, preschool spaces are only valuable to the families enrolled. Therefore, in terms of opportunity cost, space for a preschool will be disproportionately more expensive in a city, thereby pushing up land prices faster than a new restaurant would. Another check on urban growth may arise through the government spending required to finance preschool. Urban residents without children may not value investments in preschool. Glaeser and Shapiro calculated that cities with relatively 26. Glaeser and Shapiro (2003). One provocative, novel depiction of urban renewal is the emergence of a “creative class” that prospers in urban areas. As described by Richard Florida in Cities and the Creative Class (2004), urban growth is greatest in cities that have amenities and attractions that cater to the creative class, rather than in cities that have low taxes or affordable suburban housing. The demography of the creative class is young, single people and large gay communities, that is, a population that disproportionately is not one of families with children. However, there is considerable debate about the merits of this argument; see Rausch and Negrey (2006). 27. Wheeler (2006); Glaeser and Shapiro (2003). The broad relationship between educational attainment and gross metropolitan product was also identified by Rausch and Negrey (2006). 28. Glaeser and Gottlieb (2006). 29. However, the increase in housing prices will not impede growth if it results entirely from urban amenities that the increase in educational attainment produces, for example, if the increase in educational attainment leads to reduced crime and the reduced crime is fully responsible for the housing price appreciation.

Preschool Education and Human Capital Development

169

high government spending grew more slowly than those with lower levels of public expenditure.30 However, this effect is mainly driven by spending on police and health care. Strictly speaking, if taxpayers believe the evidence above, all urban residents—even childless ones—should anticipate lower spending in the future after investments in preschool. Therefore, this check should therefore not apply. Finally, the source of city vitality might not be productivity gains from a greater human capital stock. Instead, recent research has highlighted the notion of the “consumption city.” Urban resurgence may be a consequence not of rising productivity but rather of changes in preferences for city living and in the amenities and services that cities now offer.31 It is unclear how preschool would fit into this change in the culture of the city and the rise of the consumption city. In one respect, preschool might represent a “consumption good,” attracting families into the cities. The attractiveness of life in a city would diminish if putative parents realized they would have to move out once they had children. Moreover, unlike restaurants and shops, which consumers can patronize over decades, families access preschool only for a limited period. Thus the demand for preschool as a consumption good is limited.

Spillover Benefits Given its diverse impacts, preschool may generate a range of spillover benefits to urban economies. The focus is on two domains pertinent to the choice between city living and suburban living: the quality of the neighborhood and the proportion of middle-class families.32 First, preschool may improve the quality of urban neighborhoods for families. High-quality neighborhoods are those in which crime is low and good quality schools are available.33 30. See Glaeser and Shapiro (2003). 31. Glaeser and Gottlieb (2006). 32. Other consequences, such as the influence of preschool on urban sprawl, are likely to be trivial. Federal policies and property tax parameters do not strongly influence sprawl; see Nechyba and Walsh (2004). If land rents are pushed upward by preschool, families would tolerate longer commuting times, and so sprawl would increase. However, enrollment in preschool will require a commute of its own (and for more members of the family). Reluctant to accept this commuting time, families may locate closer to the preschool. An urban location for a preschool may therefore raise the costs of commuting and so induce families to remain in the city. 33. It is also possible to measure quality in terms of the cultural amenities that cities offer. Indeed, the effect of education on general cultural activity is substantively strong. Glaeser and Shapiro (2003) show that controlling for income and other covariates college graduates are significantly more likely to go out: visit an art gallery or museum, eat out, see a movie, or attend a concert. City residents are more likely to do these things too. However, the extent to which cities will offer attractions that are appropriate for or desired by families, even highly educated ones, is debatable.

170

Clive Belfield

Generally, crime is higher in cities, in part because of lower arrest rates and more opportunities for crime.34 One reason why families with children leave for the suburbs is their lower tolerance for life in high-crime neighborhoods. Over the long term, preschool should reduce the crime rate and so make cities more attractive to families (see table 5-3). Again, the reduction is likely to be mediated through increases in human capital: this mediating effect is identifiable directly (crime is lower the greater is the proportion of the population with a high school diploma) or occurs through improved employment opportunities.35 Also, there is a “school incapacitation effect” on crime: children who are in school commit fewer crimes. Important, almost one-third of crime is committed by juveniles, so it is important to implement crime-prevention strategies early.36 Typically, better quality public schools are found outside of cities. To attract families back to the city, raising the quality of public K–12 schooling is likely to be paramount. From the family’s perspective, life in the city may mean a double-hit of high property prices and low-quality public schools. However, actually improving public schools is a very difficult challenge. Many policies are promoted, but few are supported by solid evidence. Any consensus on how to raise school quality has been limited to generalizations: better teaching, more proficient and motivated students, and more resources. But there is some evidence that preschool might help in these dimensions. First, preschool—by raising children’s school readiness—should be considered as a way to improve the productivity of the most important input into K–12 schooling: the children’s own ability. As shown in tables 5-1 and 5-2, preschool does improve students’ proficiency, and it allows resources to be deployed more efficiently (for example, by reducing the need for special education). An inner-city public school with a high-quality preschool is therefore likely to be more attractive to parents. Families who attend a public preschool may then be more willing to attend that school for elementary education. Preschool may help change the perceptions of urban schools, as children who attended preschool attain higher levels of achievement at all educational levels. Second, preschool may help “grow the middle class.” Recent evidence on the wage structure has highlighted a polarization of the labor market, as the wages of low-skilled workers have flattened and those of high-skilled workers have grown.37 The likely consequence is the erosion of a middle class or equivalently a burgeoning low-income class. Preschool may help families in that individuals with more human capital are less likely to fall into the low-income class.38 34. Glaeser and Sacerdote (1999). 35. As found by Gould, Weinberg, and Mustard (2002). 36. Cohen (2005). 37. Autor, Katz, and Kearney (2006). 38. The middle class might also be attracted by a better functioning city government. Education has a strong influence on general civic engagement, which may be an input into

Preschool Education and Human Capital Development

171

These spillover benefits may be important, both in promoting economic growth and creating livable cities. However, they cannot be easily monetized and are unlikely to be politically attractive.

Costs and Benefits of Expanding Preschool Coverage We now consider the issue of implementing policies that would expand preschool coverage. First, it is necessary to consider who should be served by new preschool programs and what components those programs should include. Each city will face the same challenges—who to enroll and how much to spend—but the precise answers will differ depending on the initial provision of preschool and the economic conditions in the city. Therefore, a specific illustration is given: the economic impacts of expanding preschool in Washington, D.C. This case study shows how preschool is likely to yield positive gains for taxpayers, even in a city where programs are already available for many children.39 Finally, issues of funding for preschool and of which level of government might be responsible are considered.

Policy Concerns in Expanding Preschool The returns to preschool are highest for low-income families, and these are predominantly located in cities. Whereas 7 percent of suburban residents are poor, 17 percent of inner-city residents are poor. Of urban households with children aged seventeen and younger, 26 percent are headed by a nonmarried female.40 A program targeted to these families would have the greatest return. For example, if a preschool program costing $10,000 per student could target perfectly students on the margin of high school graduation, it would recoup for taxpayers more than $10 for every $1 invested. Yet, recent policy proposals are for expansion of universal preschool, covering all children regardless of need. Political support for universal programs is greater, and there are practical challenges in accurately targeting programs to the disadvantaged. A universal program will still convey positive net benefits to city residents, but it may also subsidize wealthy families who already access private pregovernance. College graduates are more likely to work on a community project, write to a newspaper, contact a public official, and be a registered voter (Glaeser and Shapiro, 2003, Table 5). College education raises the likelihood of voting by 21 to 30 percentage points, and each year of schooling raises voting rates by almost 4 percentage points (Dee 2004). 39. Direct evidence is in Belfield and Winters (2005) who compared the returns from investments in preschool in Wisconsin with returns from such investments targeted to Milwaukee. They found greater per child benefits in the city compared with those for the state as a whole. 40. Figures from the United States Census 2000.

172

Clive Belfield

school programs. However, if the universal program included some amount of cross-subsidy to city residents, the case for investment would be greater. Differences in the quality of preschool should also be taken into account. It is important for the preschool programs to be high quality. Quality characteristics include well-qualified, well-paid teachers and small group sizes. Given the much-noted inability of urban schools to ensure high-quality education, it is questionable whether public preschools will be able to meet these high standards. For example, preschools must recruit teachers from the same labor market as urban public schools do. Moreover, public schools receive higher per child subventions than preschools do, which should allow public schools to hire more of the best-qualified teachers. If the urban labor market has a limited number of qualified educators, the preschools will be unable to obtain qualified staff. Furthermore, some evidence points to fade-out of the benefits from preschool and that this fade-out is attributable to the lower-quality elementary schooling that some preschool children receive. Thus it is important that high-quality preschool be supplemented by a reasonable standard of K–12 schooling. Finally, as noted above, most children in cities already enroll in preschool of some kind. Therefore, expanding preschool coverage would require upgrades to the existing programs as much as increasing the availability of new places in existing programs. However, there is limited information on how much to upgrade preschool programs such that they yield positive outcomes or indeed on whether upgrades are worth the additional cost. Janet Currie and Matthew Neidell have found that increased spending on Head Start does appear to enhance outcomes.41 Also, another option is to upgrade state programs (most of which are funded at rates below Head Start). Here the challenge is to get sufficient resources for high-quality targeted programs. One option is to expand Head Start and state programs to allow all children to enroll for two years. This would generate stronger effects. At present, most children attend such programs only for one year. However, the specific inputs needed to upgrade programs are not easily identifiable.

Case Study: Expanding Preschool in Washington, D.C. A case study analysis for Washington, D.C., shows the extent of the economic benefits for cities from investing in high-quality preschool. It also illustrates the complexity of assessing the marginal benefits from preschool in cities where some provision already exists.42 41. Currie and Neidell (2007). 42. This section summarizes work by Belfield (2006). Full sources are given in that document.

Preschool Education and Human Capital Development

173

Many children in Washington, D.C., grow up under severe economic conditions: 23 percent are in extreme poverty, 55 percent are in low-income families, and 27 percent of children live in homes in which the household head did not finish high school (nationally, the figures are 8 percent, 39 percent, and 17 percent, respectively).43 Children in the District of Columbia Public School (DCPS) system are predominantly African American or Hispanic, 12 percent belong to a language minority group, an additional 8 percent are classified as limited English proficient, and more than 60 percent qualify for reduced-price lunches. Enhanced preschooling is one way to improve opportunities for these children. As in most cities, Washington, D.C., already has a sizeable preschool sector. Of the city’s 6,600 four-year-olds, 69 percent are served by public programs (Head Start, DCPS preschool, and the new Pre-K Incentive Program). These programs are not cheap: DCPS funding is $8,100 per child, Head Start is $10,900 per child, and the Pre-K Incentive Program is $12,500 per child. Child care programs offer some additional support, but resources per child are considerably less, at $4,800 to $6,500 per child. Plus, the District of Columbia (the District) faces considerable challenges in providing high-quality preschool. Programs are often imperfectly targeted because families move in and out of poverty. (Head Start covers less than 20 percent of district children even as 55 percent grow up in low-income households.)44 District funding has failed to keep up with inflation, and federal funding has declined without corresponding increases in local aid. And organizational changes are needed: the provision of preschool programs in the District has been described as “patchwork” and “fragmented,” with six city agencies offering services through 400 separate providers. Accountability is therefore weak. We assume a policy of expanding preschool in the District to cover children aged three and four who currently either receive no publicly funded preschool or are in programs that are not adequately funded. The policy would ensure high-quality preschool on a voluntary basis, which is typically the policy goal of preschool advocacy groups. The economic benefits of such a proposed policy are summarized in table 5-5. In terms of enrollment, the main change would be to add 1,000 new preschool places as part of the existing Pre-K Incentive Program. Enrollment in Head Start and DCPS preschool would not be altered, but these programs would be allocated more funding. This proposal would require a significant 43. Data on family circumstances are for 2003 and are from the Kids Count initiative of the Annie E. Casey Foundation (www.aecf.org/kidscount). Percentage of children in poverty is the share of children under the age of eighteen who live in families with incomes below the U.S. poverty threshold, which is a series of real income thresholds based on family characteristics. Low-income families are those with incomes below 200 percent of the poverty level. 44. Lyons and others (2002).

174

Clive Belfield

additional investment. High-quality preschool would likely cost $13,000 per child, and upgrades to existing programs would likely require 30 percent more than is currently spent. In total, present value extra spending would amount to $58.5 million for each cohort of children. Although substantial, this is approximately 4 percent of spending by the District during the same period. This investment would yield benefits in the domains identified in tables 5-2 and 5-3. The focus is only on the gains to the taxpayer in the form of reduced spending on educational services, on health and welfare services, and on crime, as well as increased tax revenues. The benefits are assumed to be only a fraction of those reported above for model programs. Given the local circumstances in Washington, D.C., the savings are likely to be considerable.45 The District has a very high rate of special education placement (18.6 percent compared with the national average of 13.2 percent) and a very high grade retention rate. Teaching conditions in DCPS are less than ideal, with high rates of teacher turnover and absenteeism. DCPS expenditures for school health, intervention services, child and family services, and mental health total more than $20 million annually. The high school dropout rate is high, with four out of ten public school students failing to graduate on time. Finally, the rate of violent crime is more than three times the national average. Overall, the anticipated benefits to the taxpayer in terms of additional revenues and lower expenditures are estimated to be $81.5 million in present values. When one weighs the investment costs relative to the public benefits, the net result is a present value cost saving of $23 million. Expressed as a ratio, the investment would yield $1.39 for every $1 invested. There a number of ways to interpret this cost-benefit ratio. The first is simply to note that it is clearly greater than 1, which is a critical value for investment appraisal. The second is to compare it with alternative investments. Clearly, this return is below that found for the model preschool programs. However, these programs were targeted to at-risk groups who did not have any other options available. Also, this return of 1.4 to 1 compares very favorably with those from other educational reforms (for example, the Job Corps program). In comparison with educational reforms for youth at later ages, and in comparison with other social reforms not provided through the education system, preschool programs appear very cost effective.46 Two other conclusions should be drawn from this economic evaluation (and others like it).47 First, the benefits of preschool are spread across government 45. District information is from the DCPS webpage “Just the Facts” (www.k12.dc.us/dcps/ offices/facts1.html#1); Annie E. Casey Foundation Kids Count website (www.aecf.org/ MajorInitiatives/KIDSCOUNT.aspx); D.C. Kids Count Collaborative (www.neighborhoodinfodc.org/pdfs/900910_every_kid.pdf). 46. Carneiro and Heckman (2003). 47. See Karoly and Bigelow (2005).

Preschool Education and Human Capital Development

175

Table 5-4. Present Value Fiscal Impacts of Investments in Preschool in Washington, D.C.a Millions of dollars (except where noted) Impact of program expansion for all three- and four-year-olds Additional pre-K investment [C] (cost for 1,000 new places and upgrades of existing programs and Head Start) School system cost savings Increases in tax revenues Cost savings to health and welfare systems Cost savings to the criminal justice system Total fiscal cost savings and increased revenues [B] Net fiscal savings [B – C] Benefit-cost ratio [B/C]

58.51

29.88 27.07 6.93 17.62 81.50 22.98 1.39

Source: Belfield (2006). a. Economic values are in 2005 dollars. Amounts are present values based on a 3.5 percent discount rate. School system cost savings include special education, grade retention, teacher absenteeism, teacher turnover, and school safety.

agencies. The school system recoups only half of the investment cost; large fractions are offset through tax revenues accruable to the federal government. This dilutes the incentives for public investments. Second, the payback from preschool is spread over several decades, as the children grow up and become productive citizens. In this case study, less than 10 percent of the benefits are recouped within the first four years, and only half of the benefits are recouped by the time the preschoolers graduate from high school. Hence, investments in preschool depend on the interest rate for financing.

Funding for Preschool Programs Inevitably, funding availability will constrain the optimal investment in preschool. Although general political support for preschool is strong and the evidence is sufficiently compelling, it is necessary to find sufficient public funds. There are a number of ways to finance investments in preschool.48 Each has its drawbacks. One is resource redistribution, that is, spending less in other areas to fund preschool.49 However, choosing the area for efficiency gains may be politically sensitive even as a number of economists have argued that preschool 48. Scrivner and Wolfe (2003). 49. Given that the investment is a relatively small proportion of total education expenditures, this approach is feasible to some extent. (When the context is not solely education spending but all state spending, then it appears even more feasible.)

176

Clive Belfield

spending yields higher returns than spending on higher education does, and it has been found to be more equitable.50 Yet, funding for higher education also has a strong constituency composed both of students and university employees. Similarly, preschool yields a higher return relative to industry-based development projects, but these projects too have their supporters.51 A second approach is to raise extra government revenues (or borrowing), possibly with earmarked revenues. This approach is used in some states, and it may be the most feasible option. Rather than supersede other priorities for funding, earmarked programs are easier to introduce de novo. However, earmarking means that the preschool program will depend on how much revenue is allocated. These allocations may be volatile over time. A third option is to use federal funds. One incremental change would be to use Child Care and Development Fund (CCDF) moneys, along with Temporary Assistance for Needy Families funding, to improve preschool quality. This is a possibility for urban districts with high numbers of eligible families.52 One possible option at the federal level is to transform Head Start into a large-scale preschool program for many more children (but still keep it as a means-tested program). Currently, Head Start serves far fewer students than are eligible, and many of these reside in cities. Increased funding would allow it to reach many more children. However, Head Start is a politically sensitive program. Although it is broadly effective, the gains are by no means dramatic, and it has been criticized on a number of grounds (for example, for its bureaucracy and low productivity). One policy solution is to transform Head Start into a preschool voucher program. This solution would appeal to those in favor of a greater role of the federal government in education and to those in favor of greater choice and competition in the provision of education. However, as noted above, the main obstacle is that the burden of funding and the gains from preschool are not balanced appropriately. Presently, the federal government receives the largest share of benefits (via higher income tax payments), but state and local governments are responsible for the largest share of funding for education. From the perspective of city treasuries, preschool programs are therefore more of a risk. Given their electorate, city governments may therefore have insufficient incentive to invest optimally in preschool.

50. Carneiro and Heckman (2003); Restuccia and Urrutia (2004). 51. For a full discussion, see Bartik (2006); Greenberg and Schumacher (2003). It is also possible that the creation of a preschool sector will induce a multiplier effect on economic output. However, there is not much rigorous evidence on this, and there is no strong evidence that the multiplier effect of preschool exceeds that of other industries or of lower taxes. 52. Greenberg and Schumacher (2003).

Preschool Education and Human Capital Development

177

Conclusion If preschool raises educational attainment, urban areas may reap substantial benefits across each of the domains considered above. The magnitude of the impact will of course depend on the density of families with preschool, but the direction of change will not. Indeed, preschool appears to be one of the few policy interventions on which there is very little disagreement. Even if the fiscal benefits to the taxpayer from statewide programs may not be as large as those found in model programs, the private benefits appear to be economically meaningful, particularly for low-income families. The argument that these families are likely to face the largest constraints on optimal investment in their children is also plausible and suggests support for preschool on the grounds of efficiency and equity. The fact that low-income families disproportionately live in urban areas and that urban public schools have very high dropout rates means that preschool is a promising way to boost urban development. For adults, the benefits of education—seen in higher incomes and staying off welfare and out of the criminal justice system—are also magnified in urban areas. Thus, from the perspective of what families need and which families will benefit most, preschool appears to be an attractive public investment. The private benefits to the preschool educated will yield substantial public benefits to all urban residents. However, there are some important cautions, particularly if city residents must fully fund an expansion of preschool. Strictly, the available evidence relates to average impacts, and not to the marginal impacts, of expanding preschool. This distinction is particularly important if cities already offer some preschool. Also, city populations are often mobile, such that the benefits of preschool might be lost if families migrate out to the suburbs and voters are less likely to support long-term investments. Finally, the pressure on land use in cities may constrain investments in preschool.

178

Clive Belfield

References Allgood, Sam, and Arthur Snow. 1998. “The Marginal Cost of Raising Tax Revenue and Redistributing Income.” Journal of Political Economy 106, no. 6: 1246–273. Autor, David H., Lawrence F. Katz, and Melissa S. Kearney. 2006. “The Polarization of the U.S. Labor Market.” Working Paper. Cambridge, Mass.: National Bureau of Economic Research. Barnett, W. Steven, and Debra J. Ackerman. 2006. “Costs, Benefits, and Long-term Effects of Early Care and Education Programs: Recommendations and Cautions for Community Developers.” Journal of the Community Development Society 37, no. 2: 86–100. Barnett, W. Steven, and Clive R. Belfield. 2006. “Early Childhood Development and Social Mobility.” The Future of Children 16, no. 2: 73–98. Bartik, Timothy J. 2006. Taking Preschool Seriously as an Economic Development Program: Effects on Jobs and Earnings of State Residents Compared to Traditional Economic Development Programs. Report, revised. Kalamazoo, Mich.: W.E. Upjohn Institute for Employment Research (March 30) (www.upjohninst.org/preschool/Full_report.pdf ). Belfield, Clive R. 2006. “The Fiscal Impacts of Universal Pre-K: Case Study Analysis for Three States.” Invest in Kids Working Paper 6. Washington: Committee for Economic Development, Invest in Kids Working Group (July) (www.ced.org). ———, ed. 2007. “The Economics of Early Childhood Education.” Economics of Education Review 26, no. 1 (special issue, February). Belfield, Clive, and Heather Schwartz. 2007. “The Medium-term Effects of Preschool: Evidence Using the ECLS-K Data.” Working Paper. Rutgers, N.J.: National Institute for Early Education Research (www.nieer.org). Belfield, Clive R., and Dennis K. Winters. 2005. “An Economic Analysis of Four-Year-Old Kindergarten in Wisconsin: Returns to the Education System.” Pre-K Now Research Paper, Trust for Early Education. Washington: Pre-K Now (www.preknow.org). Belfield, Clive R., and others. 2006. “The High/Scope Perry Preschool Program: CostBenefit Analysis Using Data from the Age-40 Follow-Up.” Journal of Human Resources 41, no. 1: 162–90. Berry, Christopher R., and Edward L.Glaeser. 2005. “The Divergence of Human Capital Levels across Cities.” Papers in Regional Science 84, no. 3: 407–44. Blau, David, and Janet Currie. 2006. “Who’s Minding the Kids? Preschool, Day Care, and After School Care.” In Handbook of the Economics of Education, edited by Finis Welch and Eric Hanushek, pp. 539–87. New York: North Holland. Carneiro, Pedro, and James J. Heckman. 2003. “Human Capital Policy.” In Inequality in America: What Role for Human Capital Policies? edited by James J. Heckman and Alan B. Krueger, pp.77–240. MIT Press. Cohen, Mark A. 2005. The Costs of Crime and Justice. New York: Routledge. Cullen, Julie Berry. 2003. “The Impact of Fiscal Incentives on Student Disability Rates.” Journal of Public Economics 87, nos. 7–8: 1557–589. Cunha, Flavio, and others. 2005. “Interpreting the Evidence on Life Cycle Skill Formation.” Working Paper w11331. Cambridge, Mass.: National Bureau of Economic Research. Currie, Janet. 2007. “How Should We Interpret the Evidence about Head Start?” Journal of Policy Analysis and Management 26, no. 4: 681–83.

Preschool Education and Human Capital Development

179

Currie, Janet, and Matthew Neidell. 2007. “Getting Inside the ‘Black Box’ of Head Start Quality: What Matters and What Doesn’t.” Economics of Education Review 26, no. 1: 83–99. Dee, Thomas S. 2004. “Are There Civic Returns to Education?” Journal of Public Economics 88, nos. 9–10: 1697–720. Eide, Eric R., and Mark H. Showalter. 2001. “The Effect of Grade Retention on Educational and Labor Market Outcomes.” Economics of Education Review 20, no. 6: 563–76. Florida, Richard. 2004. Cities and the Creative Class. New York: Basic Books. Garces, Eliana, Duncan Thomas, and Janet Currie. 2002. “Longer-term Effects of Head Start.” American Economic Review 92, no. 4: 999–1012. Glaeser, Edward, and Jonah Gottlieb. 2006. “Urban Resurgence and the Consumer City.” Urban Studies 43, no. 8: 1275–299. Glaeser, Edward L., and Bruce Sacerdote. 1999. “Why Is There More Crime in Cities?” Journal of Political Economy 107, no. 6: S225–S258. Glaeser, Edward L., and Jesse M. Shapiro. 2003. “Urban Growth in the 1990s: Is City Living Back?” Journal of Regional Science 43, no. 1: 139–65. Gormley, William Jr. 2007. “Early Childhood Care and Education: Lessons and Puzzles.” Journal of Policy Analysis and Management 26, no. 3: 633–71. Gormley, William T. Jr., and Ted Gayer. 2005. “Promoting School Readiness in Oklahoma: An Evaluation of Tulsa’s Pre-K Program.” Journal of Human Resources 40, no. 3: 533–58. Gould, Eric D., Bruce A. Weinberg, and David B. Mustard. 2002. “Crime Rates and Local Labor Market Opportunities in the United States: 1979–1997.” Review of Economics and Statistics 84, no. 1: 45–61. Greenberg, Mark, and Rachel Schumacher. 2003. “Financing Universal Pre-Kindergarten: Possibilities and Technical Issues for States in Using Funds under the Child Care and Development Fund and Temporary Assistance for Needy Families Block Grant.” Report, revised. Washington: Center for Law and Social Policy (May) (www.clasp.org). Hawkins, J. David, and others. 2005. “Promoting Positive Adult Functioning through Social Development Intervention in Childhood.” Archives of Pediatrics and Adolescent Medicine 159: 25–31. Karoly, Lynn A., and James H. Bigelow. 2005. The Economics of Investing in Universal Preschool Education in California. Pittsburgh: RAND Corporation. Kimmel, Jean. 2006. “Child Care, Female Employment, and Economic Growth.” Journal of the Community Development Society 37, no. 2: 71–85. Levin, Henry M., and others. 2007. “The Public Returns to Public Educational Investments in African American Males.” Economics of Education Review 26, no. 6 (December): 699–708. Lyons, Deborah, and others. 2002. UDC/CARUP 2002 Head Start Community Needs Assessment. University of the District of Columbia, Center for Applied Research and Urban Policy. Magnuson, Katherine A., Christopher Ruhm, and Jane Waldfogel. 2007. “Does Prekindergarten Improve School Preparation and Performance?” Economics of Education Review 26, no. 1: 33–51. McCarton, Cecelia, and others. 1997. “Results at Age 8 Years of Early Intervention for Low Birth-weight Premature Infants: The Infant Health and Development Program.” Journal of the American Medical Association 277, no. 2: 126–32.

180

Clive Belfield

Nechyba, Thomas J., and Randall P. Walsh. 2004. “Urban Sprawl.” Journal of Economic Perspectives 18, no. 4: 177–200. NICHD (National Institute of Child Health and Human Development) Early Child Care Research Network, and Greg J. Duncan. 2003. “Modeling the Impacts of Child Care Quality on Children’s Preschool Cognitive Development.” Child Development 74, no. 5: 1454–475. Pratt, James E., and David L. Kay. 2006. “Beyond Looking Backward: Is Child Care a Key Economic Sector?” Journal of the Community Development Society 37, no. 2: 23–37. Rausch, Stephen, and Cynthia Negrey. 2006. “Does the Creative Engine Run? A Consideration of the Effect of Creative Class on Economic Strength and Growth.” Journal of Urban Affairs 28, no. 5: 473–89. Restuccia, Diego, and Carlos Urrutia. 2004. “Intergenerational Persistence of Earnings: The Role of Early and College Education.” American Economic Review 94, no. 4: 1354–378. Reynolds, Arthur J., and others. 2002. “Age 21 Cost-Benefit Analysis of the Title I Chicago Child-Parent Centers.” Educational Evaluation and Policy Analysis 24, no. 4: 267–303. Schulman, Karen L., and W. Steven Barnett. 2006. “What Impacts Does Preschool Have on Personal Responsibility and Related Social Behavior?” Working Paper. Rutgers, N.J.: National Institute for Early Education Research (www.nieer.org). Scrivner, Stewart, and Barbara Wolfe. 2003. “Universal Preschool: Much to Gain but Who Will Pay?” Discussion Paper 1271-03. University of Wisconsin–Madison, Institute for Research on Poverty. Shapiro, Jesse M. 2006. “Smart Cities: Quality of Life, Productivity, and the Growth Effects of Human Capital.” Review of Economics and Statistics 88, no. 2: 324–35. Swanson, Christopher B. 2004. “Who Graduates? Who Doesn’t? A Statistical Portrait of Public High School Graduation, Class of 2001.” Working Paper. Washington: Urban Institute, Education Policy Center. Taylor, Lori L. 1999. “Government’s Role in Primary and Secondary Education.” Economic Review (First Quarter): 15–24 (Federal Reserve Bank of Dallas). Temple, Judy A., and Arthur J. Reynolds. 2007. “Benefits and Costs of Investments in Preschool Education: Evidence from the Child-Parent Centers and Related Programs.” Economics of Education Review 26, no. 1: 126–44. Warner, Mildred E. 2006. “Putting Child Care in the Regional Economy: Empirical and Conceptual Challenges and Economic Development Prospects.” Journal of the Community Development Society 37, no. 2: 7–22. Wheeler, Christopher H. 2006. “Human Capital Growth in a Cross Section of U.S. Metropolitan Areas.” Review 88, no. 2 (March-April): 113–32 (Federal Reserve Bank of St. Louis).

6 Can Economically Integrated Neighborhoods Improve Children’s Educational Outcomes? ingrid gould ellen, amy ellen schwartz, and leanna stiefel

M

any state and local policymakers aspire to integrate U.S. elementary and secondary public schools to improve children’s educational outcomes. Historically these policymakers have focused mostly on racial integration, but following the Supreme Court’s 2007 decision in the pair of cases Parents Involved v. Seattle School District No. 1 and Meredith v. Jefferson County Board of Education, which invalidated race-conscious assignment plans in Louisville and Seattle, socioeconomic integration is likely to present a more promising alternative.1 Socioeconomic integration, it is hoped, will produce a more equitable distribution of resources, greater equality in academic performance and subsequent labor market outcomes, and an enriched educational experience for all groups. The first school racial integration efforts took place within school districts following the Supreme Court decision Brown v. Board of Education in 1954. These efforts often involved court mandates and sometimes required busing children out of their neighborhoods to schools in other neighborhoods. Although these efforts had some success in the South, within-district integration strategies in the Northeast and Midwest had only a limited impact given the We gratefully acknowledge the excellent and patient assistance of Colin Chellman, Ioan Voicu, and Ryan Downer, without whom this chapter would not have been possible. 1. Kahlenberg (2007). Given the correlation between income and race of students, an economically integrated school system is likely to be a more racially integrated one as well. However, evidence by Reardon, Yun and Kurlaender (2006), as cited in Linn and Welner (2007, p. 46), indicates that in large, diverse urban school districts income and race are not correlated highly enough to serve as indicators for one another.

181

182

Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel

homogeneity of school districts in these regions.2 Unfortunately, cross-district efforts never got off the ground in large part because of the Supreme Court’s 1974 decision in Milliken v. Bradley, which ruled interdistrict remedies impermissible without clear evidence of an interdistrict violation. Meanwhile, many desegregation orders have been lifted recently in districts in the South.3 In response to the perceived lack of progress, some analysts and advocates have begun to focus on integrating neighborhoods as a means to integrate schools. Authors of one advocacy report state bluntly, “The most effective mechanism for desegregating . . . schools would be to desegregate . . . neighborhoods.”4 But there are serious challenges here as well, especially if, as some researchers have suggested, neighborhood segregation is “caused” by a desire for school segregation. As an example, Nancy Denton wrote, “Neighborhood segregation has become the contemporary way of keeping schools segregated.”5 If true, then attempts to integrate schools by reducing neighborhood segregation may be frustrated because parents in mixed neighborhoods eschew the neighborhood public school (for private or parochial schools or even just public schools in other neighborhoods) or because families with children simply avoid integrated neighborhoods altogether. The implication is that the racial and economic composition of public schools will not necessarily mirror the composition of their neighborhoods. In this chapter, our goal is to examine what we know about the links between the economic integration of neighborhoods and the economic integration of schools and the connection in turn between the economic integration of neighborhoods or schools and educational outcomes. Will integrated neighborhoods lead to integrated schools? Will integrated schools in turn boost student educational outcomes and children’s life chances? Will neighborhood integration alone help to improve educational performance even if schools remain segregated? Given the enormity of the topic and limitations on space, we leave for another paper a detailed discussion of policies that might be pursued to bring about economic integration of neighborhoods, and instead we focus here on the possible benefits that such integration could bring to children. In addition, because there is so little empirical research on socioeconomic integration compared with that on racial integration, we draw on the racial integration literature when we think it is applicable. 2. Whether the integration efforts themselves stimulated white flight and led to this homogeneity is the subject of debate. See Clotfelter (2001). 3. Orfield (2005). 4. Powell (2004). 5. Denton (2001, p. 100).

Economically Integrated Neighborhoods

183

We begin with a discussion of how school or neighborhood integration or both could affect students’ education outcomes, while the second section reviews what we know about the links between school and neighborhood integration and the benefits of each to children. The third section offers some new evidence on patterns of income composition in neighborhoods and schools in New York City. Finally, the last section discusses the implications of what we know for policy.

Effects of Economic Integration on School Children Integrating neighborhoods along economic lines could benefit children in two ways. First, integrated neighborhoods might produce integrated schools, which might, in turn, benefit students. Second, integrated neighborhoods might confer some advantages upon the children living there, independent of the effects on the local schools. We discuss each of these possibilities in turn below.

Should We Expect Integrated Neighborhoods to Lead to Integrated Schools? In the typical school district in the United States, students are assigned to attend an elementary school (and often middle and high schools) according to their residence in a school enrollment zone (commonly known as attendance zones or catchment areas). Enrollment zones typically are geographically proximate to the school, with boundaries defined to allow children to travel to school within a reasonable amount of time. Some cross-zone enrollment may be permitted (or overlooked), but schools are typically required to enroll any child who lives within their catchment area, and most children attend their “zoned” school. The implication is that, under commonly used school assignment policies, the composition of the student body in an elementary school will reflect, in large part, the composition of the local neighborhood. Neighborhood integration may not be sufficient to guarantee the integration of schools, however. In fact, there are many reasons why the composition of the school may not mirror the composition of the neighborhood. Perhaps most important is that schools draw only from the population of children (typically between the ages of 5 to 18), and the composition of families with school-aged children in a community may differ significantly from that of the broader population. Perhaps the middle-class or affluent families living in integrated neighborhoods tend to move to the suburbs as their children reach school age in search of larger living quarters (or homogeneous schools). Alternatively, if the housing stock is old relative to other neighborhoods, middle-income households with grown children may continue to live in the neighborhood even while

184 Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel poorer households with children move in, attracted by the affordability of the older housing. The link between residential and school integration may also be weakened by private school attendance. If higher-income parents choose private schools rather than public schools, then schools will be less integrated than their neighborhoods. A final issue is the existence of school attendance policies that allow some level of choice among public schools. In the presence of such choice, the affluent households living in economically integrated neighborhoods may opt to send their children to schools in other neighborhoods (magnet schools, for example). Ultimately, then, although we expect the composition of schools to reflect the composition of their neighborhoods, the degree of similarity is not clear. The next section explores empirical evidence on the link between the integration of neighborhoods and schools. It is worth underscoring that integrated schools do not guarantee integrated classrooms (any more than integrated neighborhoods guarantee integrated schools). It is possible that students do not interact very much with others of different income levels even when they attend the same schools. First, the poor and more affluent students may be assigned to different classrooms, reflecting differences in educational needs (say limited English skills) or differences in course or curricular preferences (AP English as opposed to accounting).6 In other words, there may be significant segregation within schools.7 Moreover, even if students from different groups share the same classrooms, they may interact with one another only minimally, absorbed in their own social networks, sitting at different tables at the cafeteria. As compelling as these arguments may be, the level of social interaction between children from different backgrounds has not been assessed in any large-scale study of neighborhoods or schools.

Why Would the Economic Integration of Schools Affect Educational Outcomes and Life Chances of Children? Most of the discussion about the economic integration of schools focuses on the benefits it has for children and, particularly, on the ways in which integration might increase children’s academic achievement. These benefits are believed to reflect some combination of peer effects, access to social networks, enhanced parental involvement, and additional school resources, including higher-quality teachers.8 Peer effects are commonly emphasized. The notion is that poor children benefit from the influence of higher-income peers, who are, on average, better pre6. See Oakes (1985) on tracking. 7. See Conger (2005). 8. Kahlenberg (2000, 2002).

Economically Integrated Neighborhoods

185

pared to succeed in school as a result of their family background. Middleincome parents, who typically not only have higher income but are also more highly educated, are more likely to read to their children, make educational resources such as books and computers available, provide more homework help, and create greater opportunities for learning outside of school. Better prepared peers may help poor children by working with them in groups or serving as role models or providing information about activities and opportunities of which poor children might be otherwise unaware. Better prepared peers may also increase the level and pace of learning in the classroom. A second way that exposure to higher-income peers may benefit lowerincome students is through social networks. Specifically, the families of middleincome classmates may provide access to networks that prove to be valuable for finding jobs, internships, and other training opportunities.9 The families of middle-income peers may also influence poor children by familiarizing them with middle-class occupations and careers and teaching them how to interact with people from higher-income backgrounds. A third reason that sharing schools with higher-income peers can be beneficial relates to parental involvement. Affluent parents are more likely to be involved directly in their children’s schools through providing in-school services and advocating for critical resources.10 School district leaders simply may be also more responsive to higher-income parents. Such care and attention is likely to benefit the poor as well as affluent children in the school. Finally, poor children might benefit from integrated schools because integrated schools are better able to attract and retain highly educated and experienced teachers and, perhaps, other resources as well (better college counseling, for example).11 It is commonly known that teachers respond to the composition of the student body in choosing where to work and where to transfer.12 Schools that educate predominantly poor children experience more teacher turnover and have difficulty attracting more experienced teachers, who view teaching in environments dominated by poor students with poorly educated parents as a significant challenge. Such problems as hunger, poor health, low attendance rates, and 9. Wells and Crain (1994). 10. Kahlenberg (2000, p. 4). 11. Notice that some of these arguments suggest that what matters most is the composition of the school attended by the student, which determines the peer effects and access to social networks, and so on. Regarding the attraction of highly educated and experienced teachers, the relationship between the composition of the school and the composition of the district (or state) may be what is key. In choosing schools, teachers will likely consider the options within a geographic area, in which case relative poverty (or relative integration) may be more important than the absolute level. See Boyd and others (2005). 12. Lankford, Loeb, and Wyckoff (2002).

186

Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel

lack of space at home to work on assignments have all been cited in qualitative work and case studies as common challenges in high-poverty classrooms.13 Moreover, there is some evidence that the expectations of teachers are lower, or at least different, when classrooms are composed of predominantly poor (often minority) students, and as a consequence, neither the curriculum nor the feedback is as challenging.14 Even the very same teachers may be more effective in more economically integrated schools than they are in schools where the students are predominantly poor. That said, integrated schools may not deliver all of the hoped-for benefits to poor children. If middle-class students draw more attention from the teachers because of, perhaps, more effective parent advocacy or maybe their own superior “navigation” or self-efficacy, then poor students may receive relatively less attention and fewer resources (although still perhaps more than they would receive in segregated schools). For similar reasons, middle-class students may dominate the school’s leadership positions such as class president or newspaper editor, leaving poor students with fewer such opportunities. Finally, we should note that some researchers go further, arguing that poor children might be hurt by attending more integrated schools because they will feel relatively deprived or that, in the case of minority children who compose too small a proportion of the student population, they may experience negative effects of tokenism or stereotype threat.15 While the bulk of the literature focuses on the likely benefits (or costs) of integration for lower-income children, some researchers make the stronger claim that higher-income children can benefit from integration as well.16 The argument is that attending integrated schools teaches students how to interact with people from different backgrounds, exposes them to a richer and more diverse set of perspectives, and breeds a fundamental tolerance for others that will be carried through to future work and social interactions.

Why Would the Economic Integration of Neighborhoods Affect Educational Outcomes and Life Chances of Children? Even if integrating neighborhoods failed to dramatically change the composition of local schools, it is still possible that lower-income children would benefit 13. Kozol (1991). Camille Holt (2005) indicates some of these conditions in an article that advocates a particular approach, based on the idea of a culture of poverty, to solving the problems. Although the approach is controversial (see Bomer and others 2008), the descriptions of classroom challenges are typical. 14. Ferguson (1998). 15. Regarding the feeling of deprivation, see Jencks and Mayer (1990). Regarding tokenism or stereotype threats, see Linn and Welner (2007, p. 48). There is little work exploring whether poor children attending schools composed largely of higher-income students tend to experience similar sorts of stereotype threats. 16. Powell (2001).

Economically Integrated Neighborhoods

187

from the experience of living in a more economically integrated environment. They may be influenced by the study habits of their neighbors, for instance, even if they do not attend the same school. They may also benefit from the presence of neighborhood-based resources, such as afterschool programs, and they may be better able to focus on school and homework in areas with lower poverty and reduced crime. Finally, they may be motivated to pursue higher levels of academic performance so that they can have the careers they see adults in their neighborhood experience. It is also possible that lower-income adults benefit from the experience of living in a more economically integrated neighborhood and become more effective parents. The more affluent neighborhoods may be quieter and safer, for example; and as a result parents may feel calmer; feel less anxious about immediate risks; and ultimately have more time, energy, and patience for their children. It is also possible that parents gain critical information from their neighbors; if so, then living near parents with greater resources and education may be beneficial.

Existing Literature In this section, we summarize the existing empirical evidence on the association between neighborhood and school integration and the benefits of each to children’s educational outcomes. Since the literature on economic integration is thin, we draw, where relevant, upon the literature on racial integration of neighborhoods and schools.

Evidence on Integration of Neighborhoods and Schools Not many researchers have studied the economic integration of either neighborhoods or schools, much less the connections between them. A few researchers have attempted to measure changes in economic segregation across neighborhoods and generally have found that economic segregation increased between 1970 and 1990 and then decreased during the 1990s.17 None of these studies address connections to school segregation.18 Considerably more research examines the racial segregation of neighborhoods and schools, and we believe this literature sheds some light on the question of whether economically integrated neighborhoods are likely to lead to economically integrated schools. In part, this derives from the correlation (albeit modest) between neighborhoods’ economic and racial and ethnic mix. Margery Turner and Julie Fenderson, for instance, show a clear link between the eco17. Fischer (2003); Jargowsky (1996); Watson (2006); Wilson (1987). 18. Results of these studies differ slightly in part because of their different choices regarding the measurement of “economic segregation.” See Jargowsky (1996) for more discussion.

188

Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel

nomic mix of a neighborhood and its racial and ethnic composition among very-low-income and very-high-income neighborhoods.19 Moreover, the same set of pathways that links the racial composition of schools and neighborhoods also links the economic composition of both. Thus there are four strands of literature on racial segregation that are relevant. One strand of the literature examines the relationship between the segregation of children and the segregation of adults. The growing number of studies that document that children are more residentially segregated by race than are adults suggests that integrating schools may be a more difficult proposition than would integrating neighborhoods. For example, John Logan and others found that on average in 2000 Black, Hispanic, and Asian children living in metropolitan areas were more residentially segregated from White children than their adult counterparts were from White adults in these same metropolitan areas.20 Similarly, in a study of the Boston metropolitan area in particular, Logan, Deirdre Oakley, and Jacob Stowell found again that minority-White segregation indexes were higher for the population under the age of 18 than they were for the population as a whole.21 Finally, Ingrid Gould Ellen found that White households without children (or with children in private schools) were significantly more likely to live in racially integrated neighborhoods.22 Thus, if economic integration follows the patterns observed for racial segregation, creating economically integrated schools will be more difficult than creating economically integrated neighborhoods. A second strand of the literature examines the link between neighborhood and school composition. Here, a few studies have suggested that private schools may weaken significantly the link between the integration of neighborhoods and the integration of schools. Logan, Oakley, and Stowell showed that in the city of Boston, while 25.4 percent of the under-18 population in the city was White in 2000, only 13.6 percent of public elementary students were White (in the 1999–2000 school year), which suggests that a disproportionate number of White children living in Boston attend private (including sectarian) schools.23 Again, it seems likely that a similar disparity exists between the share of children in a city’s public elementary school who are poor and the share of children who are poor in the city as a whole. A third strand attempts to decompose school segregation into a withinschool district component and a between-school district component, arguing 19. Turner and Fenderson (2006). 20. Logan and others (2001). 21. Logan, Oakley, and Stowell (2003). 22. Ellen (2000). 23. Logan, Oakley, and Stowell (2003).

Economically Integrated Neighborhoods

189

that segregation between districts is ultimately driven by residential segregation and therefore beyond the powers of school districts to combat it.24 All of the studies taking this approach find more segregation between districts than within districts, implying that separation of families by race across jurisdictions drives much segregation. For example Sean Reardon, John Yun, and Tamela Eitle stated, “If we eliminated all within-district [racial] segregation in every district in each metropolitan area [in the United States], we would reduce the total segregation of metropolitan areas, on average, by only 32%. The remaining twothirds of segregation is due to between-district segregation resulting largely from residential patterns.”25 Or as Steven Rivkin stated, “Extensive geographic separation blunts the impact of districts’ attendance policies.”26 Interesting and consistent with the finding that a large portion of school segregation is driven by differences in composition between school districts, these studies also found that school segregation is lower in the South and the West, where school districts tend to be larger and extend throughout entire counties. A fourth, and perhaps the most directly relevant, strand focuses explicitly on the relationship between school and neighborhood segregation across cities and metropolitan areas. Jordan Rickles and Paul Ong, for example, found a very strong correlation between the level of minority-White residential segregation in a metropolitan area in 2000 and the degree of segregation of that area’s public primary schools in the 1998–99 school years.27 There is variation in school segregation across metropolitan areas, however, that is unexplained by residential segregation. In two metropolitan areas with identical levels of residential segregation, the schools tend to be somewhat less segregated in metropolitan areas with larger schools and with a larger percentage of the students in the largest district (presumably the central city). This suggests that policies aimed at schools can help to reduce school segregation independent of residential patterns. Reardon and Yun explored whether the increase in Black-White school segregation that took place in the South during the 1990s was the result of increased residential segregation.28 They found just the opposite. They showed that in 1990, on average, the public schools in metropolitan counties in the South were 40 percent less segregated than residential neighborhoods in the same counties; by 2000 the public schools in Southern counties were only 27 percent less segre24. Rivkin (1994); Reardon and Yun (2001); Reardon, Yun, and Eitle (2000). 25. Reardon, Yun, and Eitle (2000, p. 362). 26. Rivkin (1994, p. 285). 27. Rickles and others (2001). They defined primary schools as schools in which the lowest grade is between pre-K and third grade and the highest grade is no higher than eighth. They used the dissimilarity index (see footnote 30) to measure pairwise segregation for Blacks and Whites, Hispanics and Whites, and Asians and Whites. 28. Reardon and Yun (2005).

190

Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel

gated than their local housing markets. Between 1990 and 2000, in other words, many counties had become less residentially segregated, while their schools had become more segregated. Their conclusion is that schools are becoming more segregated not because of changing residential patterns but rather as a result of a retreat from proactive school desegregation efforts. Again, although all of the studies in this area of inquiry have, to date, focused on the connections between racial segregation in neighborhoods and schools, it seems likely that we would find similar connections between economic segregation in neighborhoods and schools.

Does Economic Integration of Schools Affect Children’s Educational Outcomes? There are several challenges to providing empirical evidence on the connection between economic integration of schools and student outcomes. The first challenge involves how to define an economically integrated school. Which groups need to be represented in a school for it to qualify as integrated? Does a neighborhood or school need to be representative of the full income distribution to qualify? Defining economic integration based upon the representation of middle-class and poor students and disregarding the representation of the rich may be pragmatic and realistic, given the composition of U.S. public schools. Additionally, school districts typically only collect data distinguishing students eligible for the federal free or reduced-price lunch program. (It is sometimes possible to further distinguish those eligible for reduced-price lunches, which requires family income less than 185 percent of the federal poverty threshold, from those eligible for free lunches, requiring family income less than 130 percent of the federal poverty threshold.)29 But ignoring the isolation of the rich is far from ideal. Another issue involves defining the appropriate benchmark for an integrated school. What standard or ideal should one use in deciding whether a school or neighborhood is relatively integrated? Should the composition of a neighborhood or school be compared with the composition of the population in the school district, the city, the metropolitan area, or the nation as a whole?30 That is, is an integrated school one that is representative of the population living in that city or in the country as a whole? In studies of neighborhood integration, researchers use both national benchmarks and city-level benchmarks.31 Both 29. U.S. Department of Education (2008). 30. The dissimilarity index, the most popular measure of segregation, implicitly considers integration as a neighborhood that exactly matches the composition of the larger system, since the extent of segregation is measured as the divergence from a perfectly even distribution of residents across neighborhoods or schools within the city or metropolitan area. 31. See Ellen (2000); Smith (1998); Turner and Fenderson (2006).

Economically Integrated Neighborhoods

191

approaches have advantages and disadvantages. On the one hand, measuring integration relative to the composition of the city or metropolitan area in which a school is located is attractive because of its relevance to policymaking. How can we expect a school district to have integrated schools when the district population is homogenous? On the other hand, it is hard to imagine labeling a school “economically integrated” when 99 percent of students are poor, even if the city population as a whole is 99 percent poor. A second key challenge for empirical work is establishing causality. In general, children attending schools with higher-income peers tend to perform better on tests than do those attending schools with lower-income students, even after controlling for individual income. But this evidence is, as are many of the arguments above, derived by observing differences between schools with different student bodies rather than by observing the specific impact of policies or experiments that have changed economic integration. Thus, to the extent that parents and students choosing integrated schools differ in unobserved ways from those that do not, the impact of school integration policies may be considerably different from what previous experience suggests. What then does the evidence say about integration of schools and educational achievement? Several recent studies explore the correlation between a school’s socioeconomic composition and student performance. Using a large national database, one recent study found that a school’s socioeconomic composition (but not its racial composition) was significantly correlated with student learning during high school and, of note, that this effect appears to be correlated with concrete school practices rather than with school resources or peers per se. School practices that emerge as relevant include teachers’ expectations, hours of homework completed, the average numbers of advanced courses taken in the school, and the percentage of students who said they felt unsafe.32 A second correlational study based on another national database found that racial (but not socioeconomic) composition matters to reading performance between kindergarten and third grade.33 Finally, a third study, which reviews evidence presented to the Supreme Court in its 2007 race-conscious assignment cases, concluded that Black students’ academic performance is enhanced in less segregated schools, especially in the early years.34 Since all three of these studies use correlational research methods, it is unclear whether segregation causes improved outcomes or whether, for example, motivated or prepared low-income or Black students attend more integrated schools. 32. Rumberger and Palardy (2005, p. 2016). 33. Kainz and Vernon-Feagans (2007). 34. Linn and Welner (2007).

192

Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel

This issue of causality is particularly important in studies of peer effects, in which evidence on “pure” peer effects (that is, the direct effects that peers have upon a student’s outcomes that do not work indirectly through resources, school practices, or location decisions of families with similar characteristics) is hard to establish. Some recent studies, however, use what are called natural experiments that rely upon changes in classmates from one year to the next resulting from exogenous events, such as military draft requirements for young men, to find some credibly causal evidence that Black students benefit from increased contact with White students.35 Other credible research on peer effects has been mixed in its conclusions: some finding positive effects of boys’ achievements on girls’ achievements in high schools and some finding no effects for college students.36

Does Economic Integration of Neighborhoods Affect Children’s Educational Outcomes? There is a considerable amount of research exploring how a child’s neighborhood shapes his or her academic performance.37 Like the research on schools, much of this research is nonexperimental and simply compares the academic outcomes of children living in poor neighborhoods with those of children living in more affluent communities. These studies consistently find that children growing up in more affluent neighborhoods perform better. For example, Susan Mayer, in a study of the association between economic segregation and education inequality with respect to years of schooling, found that increased economic segregation between neighborhoods during the period from 1970 to 1990 increased the educational attainment of high-income students and reduced that of low-income students.38 Although many of these studies include a significant number of controls for a student’s family background, it is still likely that families living in more affluent neighborhoods differ in unmeasured ways from those in poor neighborhoods. Research from the Moving to Opportunity (MTO) demonstration program is more telling. The MTO program offered selected public housing tenants living in high-poverty neighborhoods the chance to move to low-poverty areas. Program participants were randomly sorted into three groups: the treatment group, who were given rental housing vouchers but were required to use them in a low-poverty neighborhood; the comparison group, who were given unrestricted rental housing vouchers; and the control group, who were offered no 35. Hanushek and others (2003); Hoxby (2000). 36. For examples of findings of positive effects, see Cipollone and Rosolia (2007); for findings of no effects, see Sacerdote (2001). 37. Ellen and Turner (1997). 38. Mayer (2002).

Economically Integrated Neighborhoods

193

additional housing assistance. Researchers then tracked these families and their children. Because families were randomly assigned to the three groups, differences in academic outcomes across groups could be attributed to group assignment rather than to family characteristics. The analysis of outcomes so far provides little support for the idea that economic integration will improve academic performance of students from poor families. After four to seven years, children in the treatment group appear, on average, to have experienced no improvement in reading scores, math scores, disruptive behavior, or level of school engagement.39 Part of the reason for the lack of results may be that many families in the treatment group have moved back to their original neighborhoods. In addition, many children in the treatment and comparison groups failed to switch schools, even when their families moved to new neighborhoods. It is of course possible that any gains in neighborhood or school environment were offset by the disruption of the move. But overall, these results suggest that neighborhood integration in itself may have only a limited direct effect on children’s academic outcomes, over and above the economic integration of schools. On the other hand, a recent study by David Card and Jesse Rothstein concluded that neighborhood racial segregation is linked to greater Black-White high school test score gaps (using SAT scores) but that school racial segregation has no effect on the size of racial gaps.40 Their analysis suggests that the relationships are causal ones—that is, racial segregation actually shapes the test score gaps. The authors found some evidence that more racially integrated schools experience considerable within-school segregation, which could account for the lack of a school integration effect.

Preliminary Evidence from New York City Neighborhoods and Schools In this section, we offer some preliminary evidence about the relationship between neighborhood and school economic composition in New York City. Although we do not present empirical work on academic performance itself, our intention is to gain insight into how tightly neighborhood and school composition are tied together in a large city. New York City hosts the largest school system in the country, educating more than 1 million children.41 Between 1990 and 2000, the system’s enrollment increased by more than 15 percent and became considerably poorer, with an 39. Sanbonmatsu and others (2006). 40. Card and Rothstein (2006). 41. National Center for Education Statistics (2006).

194 Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel increasing share of its students eligible for free or reduced-price lunch (59.3 to 81.3 percent over all grades).42 The racial and ethnic composition of students also shifted over the decade, with the share of Hispanics and Asians increasing and the proportion of Blacks and Whites decreasing. The proportion of immigrant children also increased. These changes, together with the city’s large size, make New York City a fertile laboratory for the study of the relationship between neighborhood and school composition. Our aim is to examine the extent to which the income and poverty levels of the population living in school attendance zones relate to the poverty rates of the elementary schools attached to those zones. We restrict our analysis to elementary schools, which allows us to link schools to their neighborhoods more clearly since they generally draw their students from neighborhood-based attendance zones.43 We focus on poverty rates (for which we use as a proxy the percentage of students receiving free lunch at the school level) because that is the best measure available to us of the economic composition of a school’s student body. With the cooperation of the New York City Board of Education, we compiled a data set that includes information for 1990 and 2000 about the full set of elementary public schools in New York City. Our sample includes 627 schools in 1990 and 712 schools in 2000. The number of zones served by these elementary schools ranges from 593 in 1990 to 611 in 2000.44 To describe the socioeconomic characteristics of the population living in elementary school zones, we use 1990 and 2000 census tract–level information from the Neighborhood Change Database (NCDB), which was constructed by Geolytics, in partnership with the Urban Institute.45 Table 6-1 provides a description of the elementary schools and their surrounding neighborhoods. (Note that this table shows averages for schools; that 42. Source for school year 1989–90: State of Learning: Statistical Profiles of Public School Districts (January 1991); source for school year 1999–2000: State of Learning: Statistical Profiles of Public School Districts (June 2001). See University of the State of New York, State Education Department, Albany (2001) (www.emsc.nysed.gov/irts/655report/2001/ [March 8th, 2007]). 43. We define elementary schools as schools in which the grade level is fourth or lower. 44. The number of school zones is smaller than the number of schools because two or more schools may serve the same zone. The number of zones served by multiple schools ranges from 32 in 1990 to 83 in 2000. Zone boundaries were obtained from the New York City Department of Education (DOE), and they pertain to 2003. Schools in our sample that were not assigned a zone by the DOE in 2003 (for example, school closings in years before 2003) were assigned by the authors to the zone in which they were physically located. 45. The tract-level data are used to compute school zone weighted means of the available variables, with weights given by number of residential units from the zone that fall within a given tract. The number of residential units is computed as of 2004, based on property-level information provided in the 2004 PLUTO database maintained by the New York City Department of City Planning.

Economically Integrated Neighborhoods

195

Table 6-1. Average Demographic Characteristics, New York City Elementary Schools and their Attendance Zones, 1990 and 2000a 1990

2000

Characteristic

Mean

Standard deviation

Percentage eligible for free lunch

61.4

28.9

75.0

23.8

40,151 32,500 34,623 27,214 23.3 18.6 25.3 8.9 30.2

22,271 16,502 15,927 12,532 16.1 14.1 7.6 3.0 20.6

55,062 41,982 47,873 35,427 25.0 20.0 26.2 7.0 30.9

34,236 23,541 22,588 16,275 14.0 12.9 7.4 2.0 17.4

Attendance Zones Average family incomeb Median family incomeb Average household incomeb Median household incomeb Percentage in poverty Percentage with public assistance Percentage of population under 18 Percentage of population under 5 Percentage of children in poverty Number of schools

627

Mean Standard deviation

712

Source: Authors’ calculations. a. The year 1990 denotes the 1989 to 1990 academic year, and 2000 denotes the 1999 to 2000 academic year. Note that this table shows averages for schools; that is, it describes the average school rather than averages for students, which would describe the characteristics of the whole student population. b. In dollars.

is, it describes the average school rather than averages for students, which would describe the characteristics of the whole student population.) During this period, the percentage of students eligible for free lunch in the average school (which officially corresponds to the share of students in the school living in households with incomes less than 130 percent of the federal poverty line) increased from 61.4 percent to 75 percent.46 The poverty rates increased much less in the neighborhoods defined by school attendance zones. On average, the poverty rate in the zones increased from 23.3 percent to 25 percent, the percentage receiving public assistance rose from 18.6 percent to 20 percent, and child poverty rose from 30.2 percent to 30.9 percent. 46. Although this shift likely reflects some real change in the population of students, it is important to understand the limitations of these data. School lunch eligibility during this period was based on forms distributed by schools to parents, so the number of eligible students depends on the responsiveness of parents as well as on the distribution of income. Also, schools that have high rates of free-lunch eligibility or draw from high-poverty areas may not be required to submit forms and are instead effectively treated as 100 percent poor. Therefore, the increase in measured poverty may be a reflection of changes in the aggressiveness of schools in collecting forms or of changes in the processes by which the district designates a school as “universally eligible for food programs.”

196 Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel Table 6-2. Exposure and Isolation Indexes, New York City Elementary Students and Neighborhood Children, 1990 and 2000 Percent Measure School-based Isolation indexa Poor students Nonpoor students Dissimilarity between poor and nonpoor studentsb Neighborhood-based Isolation indexa Poor children Nonpoor children Dissimilarity between poor and nonpoor childrenb Number of schools

1990

2000

75.5 57.7 50.4

80.7 38.1 39.1

48.0 70.9 40.3

43.8 69.6 33.0

627

712

Source: Authors’ calculations. a. The isolation index provides a measure of contact with one’s own group. It ranges from nearly zero to 1, with an isolation index of 1 indicating that every student in a particular group attends a school in which every student belongs to the same group. b. The dissimilarity index describes the unevenness of the population distribution. It ranges from 0 to 1 and can be interpreted as the share of the subgroup population who would have to move to a different school (or neighborhood) to achieve perfect integration.

In table 6-2, we compare the segregation of poor and nonpoor students with poor and nonpoor neighborhood residents under the age of 18 in 1990 and 2000. We use two measures: the isolation index and the dissimilarity index. The dissimilarity index describes the unevenness of the population distribution. It ranges from 0 to 1 and can be interpreted as the share of the subgroup population who would have to move to a different school (or neighborhood) to achieve perfect integration. The isolation index meanwhile provides a measure of contact with one’s own group. It ranges from nearly zero to 1, with an isolation index of 1 indicating that every student in a particular group attends a school in which every student belongs to the same group. In contrast to the dissimilarity index, the isolation index is clearly affected by changes in the overall composition of the population. The changes in the dissimilarity index suggest that poor children became less segregated across both New York City’s neighborhoods and schools between 1990 and 2000. In 1990 just over half of the poor students in the New York City schools would have had to switch to another school to achieve perfect integration of poor students. By 2000 poor students were distributed more evenly across schools, and that proportion fell to 39 percent.

Economically Integrated Neighborhoods

197

Despite this decline in segregation, the typical poor student in 2000 attended a school with a higher poverty rate than the typical poor student in 1990 did, as the isolation index rose from 75.5 to 80.7 percent. This greater isolation was driven by the increase in the proportion of poor students in the system as a whole. Indeed, the results show that nonpoor students were also attending schools with greater poverty rates by 2000. At the neighborhood level, poverty rates were fairly constant, so we see a decline in neighborhood isolation for poor children, driven by the decline in segregation.

How Are the Economic Compositions of New York City Schools and Neighborhoods Related? As described earlier, the economic composition of the school is likely to match most closely the economic composition of the neighborhood’s school-aged children. Figures 6-1 and 6-2 display scatterplots of the relationship between the percentage of students who are free-lunch eligible in elementary schools and the percentage of children in the school attendance zone who are poor. The figures indicate that school poverty generally tends to be higher than the poverty level in the corresponding zone (in part because school poverty is measured by the proportion of children living in households with an income less than 130 percent of the federal poverty level and thus captures a somewhat broader set of lower-income families). These figures also clearly show a positive relationship between the two measures of poverty in both 1990 and 2000. Table 6-3 displays the coefficients from a simple regression model that relates school poverty (percentage of students eligible for free lunch) to the childhood poverty rate in the neighborhood (census-based federal poverty level) and child poverty squared, estimated separately for 1990 and 2000. As expected, the coefficient of the zone poverty rate is large and positive, suggesting that higher child poverty in the attendance zone is associated with higher levels of poverty in the local school. The coefficient on the quadratic term is negative, suggesting that the impact of increases in child poverty in the neighborhood on school poverty is smaller in high-poverty zones than it is in low-poverty zones. Put differently, a 1 percentage point increase in poverty in a low-poverty zone translates into a larger increase in the poverty of the school than does a similar increase in a highpoverty area. Judging from the R squared value, which captures the percentage of variation in school poverty explained by the model, the relationship becomes somewhat stronger during the decade. Childhood poverty in the neighborhood and poverty squared explain 57 percent of the variation in school poverty in 1990 and 66 percent in 2000. Clearly, the economic compositions of neighborhoods and schools are linked, although it is not a perfect fit. The composition of a

198

Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel

Figure 6-1. Relationship between Free-Lunch Eligibility in Elementary Schools and Children in the School Attendance Zone Who Are Poor, 1990 School poverty (percent)

90 80 70 60 50 40 30 20 10

10

20

30

40

50

60

70

80

90

Neighborhood child poverty (percent) Source: Authors’ calculations School poverty = percentage of students eligible for free lunch.

New York City public school is clearly related to, and affected by, the composition of the children living in its attendance zone, but there is quite a bit of variation in the school poverty rate that is left unexplained. A more critical question, perhaps, is how changes in the composition of a neighborhood translate into changes in the composition of its schools. To examine changes in school and neighborhood composition, we pool the observations for 1990 and 2000 and estimate models that link the change in poverty in a school to the change in poverty in its attendance zone.47 The results 47. Only 601 schools were operating in both periods and are included in the pooled

Economically Integrated Neighborhoods

199

Figure 6-2. Relationship between Free-Lunch Eligibility in Elementary Schools and Children in the School Attendance Zone Who Are Poor, 2000 School poverty (percent)

90 80 70 60 50 40 30 20 10

10

20

30

40

50

60

70

80

90

Neighborhood child poverty (percent) Source: Authors’ calculations School poverty = percentage of students eligible for free lunch.

suggest a positive relationship between neighborhood poverty and school poverty, and again, there appears to be a nonlinear relationship. Note that the percentage of variation in the change in school poverty explained by this simple model is very low (roughly 6 percent). Although typical for change models, this makes clear that are there many other factors that explain changes in school composition beyond those we have examined. In sum, our evidence from New York City suggests that although the neighborhood economic composition is clearly correlated with the school economic regressions. Note that change models such as these are identified only by the variation in changes over time, not by the cross-sectional variation in the previous regressions.

200 Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel Table 6-3. Regression Results, Percentage of Students Eligible for Free Lunch, 1990, 2000, and Pooled a 1990 Variable

Parameter Standard estimate error

Intercept 15.23 2.08 Percentage child poverty 2.53 0.15 (Percentage child poverty)2 –0.02 0.00 R2 0.57 No. of schools

627

2000

Change 1990 to 2000

Parameter Standard Parameter Standard estimate error estimate error 21.21 2.99 –0.03

1.74 0.12 0.00

12.32 1.17 –0.01

0.78 0.24 0.00

0.66

0.06

712

601

Source: Authors' calculations. a. All coefficients are statistically significant at the five percent level of confidence.

composition, neither the poverty level in a neighborhood nor recent changes in that poverty level succeed in explaining all the variation in school poverty rates. As described earlier, there are several possible explanations. One reason that school composition does not precisely match the composition of the students living in the schools’ attendance areas is that a significant number of families in New York City send their children to private schools. In the year 2000, 18.4 percent of elementary and secondary students in the city were enrolled in private schools. In 1990 the proportion was somewhat higher at 20.9 percent.48 Naturally, the students attending private school are not a random draw of the city’s children. Our analysis shows that public school students in New York City live in households that have significantly lower incomes than those of the households of the total population of school-aged children (ages 5 to 17) living in the city. Specifically, while 50 percent of the city’s school-aged children were living in households with incomes below 185 percent of the poverty line, a full 81.3 percent of the city’s public school students were living in households with incomes under this level, according to the data on eligibility for free or reduced-price lunch.49 Second, school choice may be important, allowing children to attend schools outside of their neighborhoods. Third, it is possible that some of the mismatch is due to differences in the definition of poverty. School-based measures of poverty use a standard of 130 percent of the poverty line, while our censusbased numbers capture the percentage below the poverty line. Thus some of the divergence might be explained by differences in these measures, in particular the 48. U.S. Census Bureau (2007). 49. U.S. Census Bureau (2007).

Economically Integrated Neighborhoods

201

difference in the 30 percent above the Census-based measure of poverty but below the school-based measure.

Conclusions and Policy Implications Both our preliminary work in New York City and earlier research suggest that the composition of schools is linked to the composition of neighborhoods. Put simply, in cities and metropolitan areas that are more economically integrated, the schools tend to be more integrated as well. The match is, however, far from exact, and actual affirmative policies designed to integrate neighborhoods might have very different—and unanticipated—effects on schools. As an example, efforts to integrate neighborhoods might lead to accelerated withdrawal or flight of middle-income families to suburbs or private schools, thereby decreasing the representation of the middle class in the school district as a whole, which may carry undesirable implications. There is very little existing research that actually examines the impacts of policies designed to further neighborhood economic integration. Little existing work, for instance, examines the impacts on schools of efforts such as HOPE VI, which is designed to create more economically integrated public housing developments. Similarly, little work explores how encouraging mixed-income, private housing and neighborhoods through inclusionary zoning or other regulatory incentives and subsidy programs shapes the composition of local schools. We hope that future work will examine these important questions. It is also worth considering policies that directly target economic integration of the schools, even in the absence of neighborhood integration. In general, larger schools draw from a larger community, increasing the likelihood that some heterogeneity will emerge. In many districts in the United States, a single high school serves the entire district, and its composition mirrors the composition of the district as a whole, while the more numerous elementary schools are more segregated. In addition, larger districts (measured, say, by the fraction of the metropolitan statistical area enrollment) appear to provide more opportunities to integrate schools within that district. Thus the South and the West, where county-wide districts are common, have more opportunities for intradistrict integration than do the Northeast and Midwest, where districts are smaller and often described as balkanized. Similarly, all else equal, the greater the amount of choice among public school options, the looser is the link between residential location and school attendance, and the looser is the link between the neighborhood composition and the school composition. Greater choice can yield more or less integrated

202

Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel

schools, depending upon the choices made by parents and students. Magnet schools, for example, might be effective in integrating schools if they draw students from different income groups living in different neighborhoods. If the choices include cross-district enrollment, such as in Minneapolis, then opportunities increase even more.50 But Robert Linn and Kevin Welner, in a review of some recent evidence, concluded that without some race-conscious assignment characteristics, choice plans will not increase racial integration in practice.51 The same may be true for socioeconomic integration of schools. It is also important to note that some of the benefits of school integration can potentially be achieved without actual integration of schools, to the extent that these benefits are not based on pure peer effects but rather on associated resource availability or school practices. While evidence on pure peer effects in general is mixed, and evidence on the importance of the income of one’s peers is almost nonexistent, there is evidence suggesting that teacher expectations and experience are lower in schools with larger concentrations of poor children. Thus efforts to change the distribution of teachers could bring about some of the benefits associated with integrated schools. In the end, the evidence—both in New York City and around the country— seems to suggest that our neighborhoods are more economically integrated than are our schools, and increases in the economic integration of neighborhoods may not lead to similarly large increases in the economic integration of schools. Of course, neighborhood integration may bring other important benefits to children more directly such as improved educational outcomes, separate and apart from any effects due to integrating schools. If, however, creating and sustaining economically integrated schools is important in and of itself, then policymakers and educators may well have to look beyond neighborhood integration policies to consider school-based integration policies that can directly shape the economic composition of schools.

50. Katherine Kersten, “Black Flight: The Exodus to Charter Schools,” Wall Street Journal, editorial page, March 2, 2006 (www.opinionjournal.com/cc/?id=110008032 [March 14, 2006]). 51. Linn and Welner (2007).

Economically Integrated Neighborhoods

203

References Bomer, Randy, and others. 2008. “Miseducating Teachers about the Poor: A Critical Analysis of Ruby Payne’s Claims about Poverty.” Teachers College Record 110, no. 12. Boyd, Donald, and others. 2005. “The Draw of Home: How Teachers’ Preferences for Proximity Disadvantage Urban Schools.” Journal of Policy Analysis and Management 24, no. 1: 113–32. Card, David, and Jesse Rothstein. 2006. “Racial Segregation and the Black-White Test Score Gap.” Working Paper 12078. Cambridge, Mass.: National Bureau of Economic Research. Cipollone, Piero and Alfonso Rosolia. 2007. “Social Interaction in High School: Lessons from an Earthquake.” American Economic Review 97, no. 3: 948–65. Clotfelter, Charles T. 2001. “Are Whites Still Fleeing? Racial Patterns and Enrollment Shifts in Urban Public Schools, 1987-1996.” Journal of Policy Analysis and Management 20, no. 2: 199–221. Conger, Dylan. 2005. “Within-School Segregation in an Urban School District.” Educational Evaluation and Policy Analysis 27, no. 3: 225–44. Denton, Nancy A. 2001. “The Persistence of Segregation: Links between Residential Segregation and School Segregation.” In In Pursuit of a Dream Deferred: Linking Housing and Education Policy, edited by John A. Powell, Gavin Kearney, and Vina Kay, pp. 89–119. New York: Peter Lang Publishing. Ellen, Ingrid Gould. 2000. Sharing America’s Neighborhoods: The Prospects for Stable, Racial Integration. Harvard University Press. Ellen, Ingrid Gould, and Margery Austin Turner. 1997. “Does Neighborhood Matter? Assessing Recent Evidence.” Housing Policy Debate 8, no. 4: 833–66. Ferguson, Ronald F. 1998. “Teachers’ Perceptions and Expectations and the Black-White Test Score Gap.” In The Black-White Test Score Gap, edited by Christopher Jencks and Meredith Phillips, pp. 273–317. Brookings. Fischer, Mary J. 2003. “The Relative Importance of Income and Race in Determining Residential Outcomes in U.S. Urban Areas, 1970-2000.” Urban Affairs Review 38, no. 5: 669–96. Hanushek, Eric A., and others. 2003. “Does Peer Ability Affect Student Achievement?” Journal of Applied Econometrics 18, no. 5: 527–44. Holt, Camile B. 2005. “Preparing Teachers for Children in Poverty: The Nashville District Picks Up the Mantle for Qualified Instruction in High-Needs Schools.” School Administrator (December) (http://findarticles.com/p/articles/mi_m0JSD/is_11_62/ai_n15970230 [September 18, 2007]). Hoxby, Caroline M. 2000. “Peer Effects in the Classroom: Learning from Gender and Race Variation.” Working Paper 7867. Cambridge, Mass.: National Bureau of Economic Research. Jargowsky, Paul A. 1996. “Take the Money and Run: Economic Segregation in U.S. Metropolitan Areas.” American Sociological Review 61, no. 6: 984–98. Jencks, Christopher, and Susan E. Mayer. 1990. “The Social Consequences of Growing up in a Poor Neighborhood.” In Inner-City Poverty in the United States, edited by Laurence E. Lynn and Michael G. H. McGeary, pp. 111–86. Washington: National Academy Press. Kahlenberg, Richard D. 2000. “Economic School Integration.” Idea Brief 2. Washington: The Century Foundation (February).

204

Ingrid Gould Ellen, Amy Ellen Schwartz, and Leanna Stiefel

———. 2002. “Economic School Integration: An Update.” Issue Brief Series. Washington: The Century Foundation (September). ———. 2007. Rescuing Brown v. Board of Education: Profiles of Twelve School Districts Pursuing Socioeconomic School Integration. Washington: The Century Foundation (June). Kainz, Kirsten, and Lynne Vernon-Feagans. 2007. “The Ecology of Early Reading Development for Children in Poverty.” Elementary School Journal 107, no. 5: 407–27. Kozol, Jonathan. 1991. Savage Inequalities. New York: Crown Publishers. Lankford, Hamilton, Susanna Loeb, and James Wyckoff. 2002. “Teacher Sorting and The Plight of Urban Schools: A Descriptive Analysis.” Educational Evaluation and Policy Analysis 24, no. 1: 37–62. Linn, Robert L., and Kevin G. Welner, eds. 2007. Race-Conscious Policies for Assigning Students to Schools: Social Science Research and the Supreme Court Cases. Washington: National Academy of Education, Committee on Social Science Research Evidence on Racial Diversity in Schools. Logan, John R., Deirdre Oakley, and Jacob Stowell. 2003. Segregation in Neighborhoods and Schools: Impacts on Minority Children in the Boston Region. New York: University at Albany, Lewis Mumford Center. Logan, John R., and others. 2001. Separating the Children. New York: University at Albany, Lewis Mumford Center. Mayer, Susan E. 2002. “How Economic Segregation Affects Children’s Educational Attainment.” Social Forces 81, no. 1: 153–76. National Center for Education Statistics. 2006. “Characteristics of the 100 Largest Public Elementary and Secondary School Districts in the United States: 2003–04.” Washington: U.S. Department of Education, Institute of Education Sciences (http://nces.ed.gov/ pubs2006/100_largest/tables.asp [March 20, 2008]). Oakes, Jeannie. 1985. Keeping Track: How Schools Structure Inequality. Yale University Press. Orfield, Gary. 2005. “Introduction.” In School Resegregation: Must the South Turn Back? edited by John Charles Boger and Gary Orfield, pp. 1–25. University of North Carolina Press. Powell, John A. 2001. “Living and Learning: Linking Housing and Education.” In In Pursuit of a Dream Deferred: Linking Housing and Education Policy, edited by John A. Powell, Gavin Kearney, and Vina Kay pp. 15–48. New York: Peter Lang Publishing. ———. 2004. “Dismantling Racial Isolation in Public School Education.” Prepared for the Brown v. Board of Education: The Unfinished Agenda Conference, sponsored by ERASE Racism (Education Research Advocacy Support to Eliminate Racism). March 29. Reardon, Sean F., and John T. Yun. 2001. “Suburban Racial Change and Suburban School Segregation, 1987–95.” Sociology of Education 74, no. 2 (April): 79–101. ———. 2005. “Integrating Neighborhoods, Segregating Schools: The Retreat from School Desegregation in the South, 1990-2000.” In School Resegregation: Must the South Turn Back? edited by John Charles Boger and Gary Orfield, pp. 51–69. University of North Carolina Press. Reardon, Sean F., John T. Yun, and Michal Kurlaender. 2006. “Implications of Income-Based School Assignment Policies for Racial School Segregation.” Educational Evaluation and Policy Analysis 28, no. 1: 49–75. Reardon, Sean F., John T. Yun, and Tamela McNulty Eitle. 2000. “The Changing Structure of School Segregation: Measurement and Evidence of Multiracial Metropolitan-Area Segregation, 1989–1995.” Demography 37, no. 3: 351–64.

Economically Integrated Neighborhoods

205

Rickles, Jordan, Paul M. Ong, with others. 2001. “Relationship between School and Residential Segregation at the Turn of the Century.” Discussion Paper. University of California–Los Angeles, Ralph and Goldy Lewis Center for Regional Policy Studies (June 11). Rivkin, Steven G. 1994. “Residential Segregation and School Integration.” Sociology of Education 67, no. 4: 279–92. Rumberger, Russell W. and Gregory J. Palardy. 2005. “Does Segregation Still Matter? The Impact of Student Composition on Academic Achievement in High School.” Teachers College Record 107, no. 9: 1999–2045. Sacerdote, Bruce. 2001. “Peer Effects with Random Assignment: Results for Dartmouth Roommates.” Quarterly Journal of Economics 116, no. 2: 681–704. Sanbonmatsu, Lisa, and others. 2006. “Neighborhoods and Academic Achievement: Results from the Moving to Opportunity Experiment.” Journal of Human Resources 41, no. 4: 649–91. Smith, Richard A. 1998. “Discovering Stable Racial Integration,” Journal of Urban Affairs 20: 1–26. Turner, Margery Austin, and Julie Fenderson. 2006. “Understanding Diverse Neighborhoods in an Era of Demographic Change.” Washington: Urban Institute. U.S. Census Bureau. 2007. “American FactFinder” (http://factfinder.census.gov [March 9, 2007]). U.S. Department of Education. 2008. “The National Assessment of Educational Progress Glossary of Terms” (http://nationsreportcard.gov/glossary.asp [March 10, 2008]). Watson, Tara. 2006. “Metropolitan Growth, Inequality, and Neighborhood Segregation by Income.” Brookings-Wharton Papers on Urban Affairs 2006, edited by Gary Burtless and Janet Rothenberg Pack, pp. 1–52. Wells, Amy Stuart, and Robert L. Crain. 1994. “Perpetuation Theory and the Long-Term Effects of School Desegregation.” Review of Educational Research 64, no. 4: 531–55. Wilson, William Julius. 1987. The Truly Disadvantaged: The Inner-City, the Underclass, and Public Policy. University of Chicago Press.

7 Spatial Development and Energy Consumption elena safirova, sébastien houde, and winston harrington

I

n recent years, American consumers spent over half a trillion dollars a year on energy. In 2000 the consumption of energy in the United States had tripled since 1949 and totaled about 98 quadrillion British thermal units (Btu). Although during this time period the amount of energy used per real dollar of U.S. GDP fell from 20,600 Btu to 10,600 Btu, population growth (from 149 million in 1949 to 281 million in 2000) and per capita GDP growth caused energy consumption per household to grow 63 percent, from 215 million Btu in 1949 to 350 million Btu in 2000.1 In the last two decades the concerns about energy prices and energy security seemed to be of less importance, and other consequences of energy consumption, such as environmental protection, were leading the quest for conservation. More recently, the issues of energy security are again on the front pages of the newspapers, and in addition to them, concerns about climate change make the search for recipes to restrain energy consumption more urgent. Recent evidence suggests that in the long run for each extra dollar earned, spending on energy amounts to $0.55 to $0.60, so it is unsurprising to find the demand for energy to be rising with people’s incomes.2 Still, the demand for The authors thank Royce Hanson, Rob Puentes, Hal Wolman, and other participants in the Conference on Urban and Regional Policy and Its Effects for helpful comments on an earlier draft of this chapter. We are grateful to Conrad Coleman and Janet Nackoney for assistance with the maps. Remaining errors are exclusively ours. Sébastian Houde was with Resources for the Future at the time of the conference and is now a doctoral student at Stanford University. 1. See EIA (2001). 2. See Gately and Huntington (2001) regarding long-term energy spending for each dollar earned.

206

Spatial Development and Energy Consumption

207

Table 7-1. Annual U.S. Energy Use and Rate of Change, by Sector In quadrillion Btu Year 1973 2006 Percent 1973–2006

Transportation

Industrial

Commercial

Residential

Total

18.6 28.4

32.7 32.1

9.5 18.0

14.9 21.1

75.7 99.7

1.3

–0.1

2.0

1.1

0.8

Source: Oak Ridge National Laboratory (2007).

energy in most instances is a derived demand, that is, people do not demand energy per se, but they like to consume goods and services that in turn require energy. This provides some justification for the hope of reducing energy consumption while minimizing impact on the welfare of consumers. In this chapter, we consider one particular approach to reducing energy consumption: modifying urban spatial structure to reduce the demand for energy in transportation and space heating. At first glance this would appear to be a promising approach. Table 7-1 reports on energy consumption in the four broad sectors of the U.S. economy. As shown, growth in energy use in the largest sector, industrial, has been stagnant for the past thirty years, while growth in the other three sectors has been steady and significant. It is these three sectors—transportation, residential, and commercial—that are most likely to be affected by changes in urban form. In general terms, urban form refers to the physical layout and design of a metropolitan area. Urban form includes density, street layout, transportation, employment and other urban activity areas, and urban design characteristics. Although existing urban forms are a product of numerous factors such as regional economic factors, regional development trajectory, a combination of policy and regulation factors, and so on, urban forms vary significantly across the world’s regions. Therefore, there is a strong hope that urban form can be affected by government policies and other factors.

Theory and Empirical Research The relationship between urban form and energy consumption was actively studied during the early 1970s, when concerns about security of energy supply gave rise to a wave of research aimed at evaluating the efficiency of the current state of energy consumption. Interest waned in the 1980s after the price of crude oil collapsed, but several events in the last decade or so have put the relationship between urban form and energy use back on the research agenda, including a

208

Elena Safirova, Sébastian Houde, and Winston Harrington

renewed concern about energy security and new concerns about global climate change. In addition, the policy prescriptions of the energy–urban form link seemed to point toward more compact cities, something that dovetailed nicely with local environmental concerns, as evidenced by the Smart Growth movement and the battle in some areas over rapid growth and suburban sprawl. The main link between energy use and urban form, of course, is urban transportation: abundant evidence has linked the spatial density of economic activities to the demand for vehicle use. Although making adjustments to urban form does not constitute direct energy conservation, it serves as a facilitating strategy that makes a variety of conservation activities possible.3 Starting with this idea, several research studies in the mid-1970s attempted to estimate the potential impacts of land use planning on energy consumption. The studies assumed either hypothetical cities or a hypothetical growth pattern of existing cities. The forecast reductions in total energy consumption ranged between 0.35 percent according to Dale Keyes and 46.3 percent according to the Council on Environmental Quality.4 Although those studies provided some valuable insights, their major drawback is that they did not explicitly model behavioral responses of individual households to changes in price signals. A secondary link is found in space heating requirements. Large multifamily dwellings, usually found in dense urban areas, have a lower ratio of exterior surface area to interior square footage, thus reducing the rate of heat or cooling loss. This link has received much less attention, especially from policy analysts. The majority of research on the relationship between urban form and energy consumption that followed those early studies focused on transportation-related energy consumption, with very little attention paid to building-specific energy consumption. Below we review those strands of literature separately.

Transportation-Related Studies It is usually perceived that the most significant impact of the built environment on energy consumption is made through travel. However, the research into the relationship between auto travel and built environment might address some issues that are different from energy consumption, such as emissions and other environmental concerns, urban sprawl as a general phenomenon, and so forth. Therefore, in this section we review the literature that addressed the relationship between the amount of auto travel and characteristics of urban form in general. 3. Keyes (1978). 4. The studies include Keyes (1978); Council on Environmental Quality (1975); Roberts (1978); Carrol (1978); Edwards (1978). A detailed review of these studies can be found in Anderson, Kanaroglu, and Miller (1996).

Spatial Development and Energy Consumption

209

Researchers studying the effect of the built environment on travel demand have analyzed the impact of density, city size, and mixed land use in human settlements of different scales, as well as the supply of public transit and the structure of the urban system.5 The most extensive and frequently cited studies on the impact of density on travel demand and on energy use by cars in thirty-two large cities in Europe, the United States, Australia, Asia, and Canada are by Peter Newman and Jeffrey Kenworthy.6 Using some simple regression analyses, Kenworthy and Newman concluded that differences in gasoline prices, income, and vehicle fleet efficiency explained only about 40 percent of the variation in the gasoline demand. However, they found that a large portion of the remaining variation can be explained by a simple measure of population density. They concluded that the urban population density is the single most important factor and called for policies of reurbanization to reduce transport energy demand and the associated environmental problems. Although higher densities may be expected to reduce the need to travel longer distances, the literature on this hypothesis is ambiguous. Ruth Steiner in her literature review concluded that, in aggregate, studies suggest that residents in high-density areas travel longer distances than residents in low-density areas do.7 David Levinson and Ajay Kumar, however, studied cities in the United States with more than 1 million inhabitants and came to another conclusion.8 After controlling for available opportunities, transport infrastructure, and the socioeconomic and sociodemographic characteristics of the residents, they found a positive relationship between metropolitan residential density and average commuting distance for auto and transit commuters. These ambiguous results could be related to metropolitan area size. In their study, Levinson and Kumar suggested that metropolitan residential density is a proxy for metropolitan area size. On the one hand, because large metropolitan areas offer more services and facilities than smaller ones offer, they may be associated with shorter travel distances and use of public transit. On the other hand, a dispersion of urban land use over a large area may lead to longer travel distances and a higher share of car trips. The complex interactions between metropolitan area size and travel distances have been extensively studied by Peter Gordon, Ajay Kumar, and Harry Richardson.9 They analyzed the relationship between metropolitan area size and distances traveled for both commuting and 5. For more comprehensive reviews, see Anderson, Kanaroglou, and Miller (1996); Badoe and Miller (2000); Crane (2000); Ewing and Cervero (2001); Handy (1996); Steiner (1994); Stead, Williams, and Titheridge (2000). 6. Newman and Kenworthy (1989, 1999); Kenworthy and Newman (1990). 7. Steiner (1994). 8. Levinson and Kumar (1997). 9. Gordon, Kumar, and Richardson (1989).

210

Elena Safirova, Sébastian Houde, and Winston Harrington

discretionary purposes in the United States. They found that the travel behavior is significantly different for central city and suburban residents. In particular, for central city residents, commuting trips increase with city size. In contrast, travel distances for nonwork trips decrease in cities with up to 1 million residents and increase for larger cities. At the same time, suburban residents have longer travel distances for both work and nonwork purposes, and their travel behavior is less affected by the size of the metropolitan area where they live. One of the important leitmotifs in the literature on the relationship between urban form and travel is the provision of public transit and its role in reducing auto travel. Several studies used the distance of household residence from public transit or from the central business district (CBD) as a measure of availability of public transportation.10 At the same time, in the empirical analysis it has proven difficult to control for other variables that affect automobile ownership and travel mode choice and to disentangle effects when the variables are determined jointly. Antonio Bento and colleagues attempted to rectify this problem by measuring transit supply by route miles in public transit networks normalized by the city area.11 They found that the probability of driving to work is lower the higher are population centrality and rail miles and the lower is road density. They also found that moving sample households from a city with characteristics of Atlanta to a city with characteristics of Boston reduces annual vehicle miles traveled (VMT) by 25 percent. Quantifying the benefits of compact and dense development might be a more difficult task than initially thought. For one thing, people with different demographic and socioeconomic characteristics might have strong preferences for a particular lifestyle. Frans Dieleman, Martin Dijst, and Guillaume Burghouwt have found that personal attributes of households and characteristics of their residential environment are very important determinants of mode choice and travel distance.12 Early studies did not take this into account at all and were later criticized because of that.13 Aggregate studies often failed to control for socioeconomic and demographic characteristics among households in different areas as well as for differences in transportation infrastructure and the cultural, political, historical, and economic differences among the areas.14 However, many disaggregate studies have not effectively accounted for the possibility that residential location is both a cause and an effect of residential density and vehicle usage. Thomas Golob and David Brownstone pointed out that the use 10. Train (1980); Boarnet and Sarmiento (1998); Crane and Crepeau (1998); Boarnet and Crane (2001). 11. Bento and others (2005). 12. Dieleman, Dijst, and Burghouwt (2002). 13. Handy (1996); Steiner (1994). 14. Gomez-Ibañez (1991).

Spatial Development and Energy Consumption

211

of city-wide or metropolitan area–wide data on urban form together with disaggregate travel data often does not solve the problem since that approach ignores potentially important influences on travel of differences in urban form between neighborhoods or larger geographic zones within a metropolitan area.15 Golob and Brownstone have used a model in which residential density, vehicle usage, and fuel consumption are jointly determined within the model. Analyzing data from Southern California, they found that a lower density of 1,000 housing units per square mile implies a positive difference of almost 1,200 miles per year and about 65 more gallons of fuel per household. Moreover, the contribution of vehicle choice to the latter number is about 20 gallons per year. Finally, some researchers have suggested that dispersed development might lead to less automobile use than what would occur in a monocentric city. They have argued that efficient travel patterns emerge as firms and households follow one another during the course of employment decentralization.16 The literature on this topic primarily addresses auto-based commuting, although decentralization would affect nonwork travel as well. At the same time, several authors have shown that polycentric urban form tends to be associated with higher levels of auto dependence and solo driving.17 Some evidence suggests that mode effects could partially be explained by insufficient transit connectivity between residences and workplaces.18 Overall, the literature on the effect of polycentric form is inconclusive with respect to the aggregate effect of polycentricity on the amount of travel in urban areas.

Studies of Energy Consumption in Buildings Energy consumption in buildings is another component that can contribute to overall metropolitan energy consumption. In the United States, buildings account for 36 percent of all energy consumed compared with 41 percent in the European Union (EU) and more than 50 percent in the United Kingdom.19 City comparisons with respect to this particular type of energy consumption are even harder to make in a meaningful way since such factors as climate, preexisting housing stock, and idiosyncratic population habits can make the overall picture even more complex. Nevertheless, energy use in buildings is rarely mentioned in discussions of the relationship between urban form and energy use. 15. Golob and Brownstone (2005). For use of city-wide and metropolitan-wide data on urban form together with disaggregate travel data, refer to Levinson and Kumar (1997); Bento and others (2005). 16. Gordon and Richardson (1997); Levinson (1998); Schwanen, Dieleman, and Dijst (2004). 17. Cervero and Wu (1997, 1998). 18. Modarres (2003). 19. Steemers (2003).

212

Elena Safirova, Sébastian Houde, and Winston Harrington

For example, a detailed review of interactions between land use, transportation, and environmental issues conducted by the Environmental Protection Agency discussed energy only in terms of energy use for transportation.20 At the same time, there is an existing literature on the relationship between building energy use and urban form that provides its own perspective on urban planning trade-offs. In the earlier literature Philip Steadman concluded that high-density linear development along transport routes would be more energy efficient than would compact central development, since buildings can be more energy efficient.21 (A linear pattern better permits natural lighting, ventilation, and passive solar gain; and infrastructure can be shared.) More recent studies are inconclusive. Some authors suggest that higher building densities reduce energy demand.22 Others believe that increasing density can increase energy demand because of a restriction of light and higher opportunity for solar gain.23 In a review of the issues related to housing, Koen Steemers concluded that the energy argument for and against densification is finely balanced and will depend on infrastructure issues (that is, whether there are opportunities for buildings to share water and energy networks).24 Although a more detailed analysis goes beyond the capabilities of our modeling framework, the issues related to building design as a factor affecting energy consumption should be kept in mind for further analysis.

Energy Consumption and Public Policy Although much of the literature reviewed above focused on the concept of a compact city that is characterized by a smaller footprint concerning energy and emissions, several important questions remain. One of them is whether the energy burden even in principle can be significantly reduced if urban form becomes more compact. As we have seen above, theory and empirical evidence are both inconclusive, and direct numerical comparisons of the effects involved might be required to find the answer. The most important question that we are trying to address here is whether land use changes alone can make a significant difference in energy consumption. To test this hypothesis, we conduct a series of experiments, which we call urban scenarios, that it is hoped would make urban form more compact. However, there are two important caveats. First, scenarios 20. EPA (2001). 21. Steadman (1979). 22. For examples, see Holden and Norland (2005); Mindali, Raveh, and Salomon (2004). 23. For examples, see Hui (2001); Larivière and Lafrance (1999). 24. Steemers (2003).

Spatial Development and Energy Consumption

213

are not policies, and we do not know how to achieve what is described in a particular scenario. Second, if there were a way to achieve the state of affairs described in the scenario, it might require either very high and hard-to-estimate costs or significant changes in individual preferences of the city residents that are also hard, if at all possible, to change. We also would like to determine which policies can achieve outcomes preferable to the status quo. It is one thing to say that cities of different urban form have different patterns of energy consumption and a quite different thing to assume that realistic policies can turn an inefficient city into an efficient one. We will call the second type of experiments urban policies. An important difference between the policies and the scenarios is that policies are direct instruments that come with implementation recipes. People’s preferences under a policy stay the same as in the status quo. Therefore, we can conduct cost-benefit analyses of different policy instruments and compare policies on the basis of the net benefits they are likely to bring.25 William Anderson, Pavlos Kanarouglou, and Eric Miller suggested that the most efficient way to study further the relationship between urban form and urban energy efficiency would be through a comprehensive study of possible outcomes of alternative policies.26 In particular, they said that the current (circa 1995) generation of land use and transportation models should be extended in two ways—by incorporating a range of policy instruments as exogenous variables, that is, as model parameters that can be easily changed, and by building modeling blocks that would translate travel demand and land use changes into energy and emissions. In this chapter, we took their suggestion and developed a methodology to analyze how urban policies might affect metropolitan energy consumption. The main benefit of using a land use and transportation framework to evaluate urban policies and scenarios is that transportation and land use in urban areas are very much intertwined. Likewise, scenarios and policies that directly affect either transportation or land use decisions will indirectly affect the other component. Therefore, it is hard to predict which policies—predominantly land use policies or predominantly transportation policies—would be more effective at reducing energy consumption, unless such a complex methodology is employed. The motivation of the discussion to follow is to evaluate to what extent an ideal compact urban form leads to energy savings and investigate how much savings can be achieved through policy intervention. The rest of the chapter is struc25. In this framework, costs of the policies can include tangible items such as implementation costs as well as other costs, for example political costs. Although they are harder to estimate, political costs sometimes serve as a major impediment to policy adoption and implementation. 26. Anderson, Kanarouglou, and Miller (1996).

214

Elena Safirova, Sébastian Houde, and Winston Harrington

tured as follows. The following section outlines our methodology and model. We then briefly describe the scenarios and policies that we simulate using the model. Next we discuss an array of other urban policies that cannot be accurately simulated using our model but whose effects can be compared with those of the policies we are able to simulate. The chapter concludes by discussing the limitations of our approach and sketches a road map for future research.

Methodology In this section, we briefly describe the structure and main features of the modeling framework used to conduct policy simulations. Also, we state energy-related assumptions made in this study and describe the energy-related status quo for the Washington, D.C., metropolitan area that we use as a baseline for policy simulations.

Model Overview The model we use in this chapter is an integrated model of land use, economic activity, and transportation (see appendix A-1 for model description). The model provides an attractive tool for evaluation of urban policies that affect residential and transportation decisions of inhabitants of urban areas. Because both local travel and locational decisions of firms and residents in the metropolitan area are modeled explicitly, scenarios and policies that affect such decisions can be represented, evaluated, and compared. An important element of the model is that it automatically computes the changes of the residents’ economic welfare that are the result of a particular policy. The computation of welfare serves as a basis for cost-benefit analysis, an evaluation of the economic efficiency of a particular policy. Finally, the model reveals the actual policy mechanism. In other words, the simulations demonstrate how exactly decisions of the economic actors are changed as a result of policy and therefore eliminate the impression that policies operate within black boxes. At the same time, our model in its current form has only forty land use zones and about 350 transportation links. Although such structure leads to fast run times and enables quick analysis of multiple policies, it also poses some limitations. First of all, not all policies can be meaningfully represented in the model. This especially concerns policies implemented at a microscale in the areas that are much smaller than existing modeling zones. Another limitation is that the model does not explicitly consider what determines vehicle ownership choices and therefore is not able to reflect a shift to smaller, more efficient cars as a result of higher fuel prices. Finally, the model only shows the effects of the policies in the long run; it is not capable of demonstrating what intermediate

Spatial Development and Energy Consumption

215

changes will occur in the urban area before all policy effects take place. Consequently, although our model is a quite detailed model of the entire regional economy, it is not equally good at predicting the consequences of all policies.

Details of Energy Modeling This section presents the methodology used to produce annual estimates of residential and vehicular energy use with our model.27 vehicular energy use. The transportation simulation component of our model computes the costs of travel, which consist of monetary costs and time costs. The monetary costs of driving include fuel costs, fuel tax, vehicle depreciation, wear and tear, maintenance, and insurance. The model predicts fuel use, which varies by speed and vehicle miles traveled. Fuel use is measured in gallons and corresponds to a combination of gasoline and diesel products. Our methodology does not include fuel used by buses in the public transit system. Our model simulates transport for an average week day of the year. The annual estimate of fuel use thus corresponds to a working year (250 days).28 For the purposes of easy comparison with residential energy use, we convert gallons of gasoline into Btu.29 residential energy use. To estimate residential energy usage, we combine population distribution numbers from our model with energy consumption data from the Energy Information Administration’s residential energy consumption surveys.30 In particular, we use annual energy consumption (end use only) data disaggregated by household members and by the following four building types: single family housing units, detached (SFD); single family housing units, attached (SFA); apartments in multifamily housing buildings of two to four units (MF24); and apartments in multifamily housing buildings of five units or more (MF5). These consumption data correspond to national averages, but we adjust them for the Washington, D.C., metropolitan area’s climate, considering the difference between the average national consumption per household and the average consumption per household for a climate zone of fewer than 2,000 cooling degree days and between 4,000 and 5,999 heating degree days. Our method of producing energy use estimates will capture energy savings in residential use obtained by two sources. First, energy savings and losses from substitution between single and multifamily housing will be included. Second, different population distributions by income class across the region will also affect energy outcomes. 27. More details on modeling assumptions and a brief description of the baseline (status quo) energy profile can be found in appendixes A-2 and A-3, respectively. 28. Because we model traffic for an average week day, we consider only working days to compute annual vehicular energy use. Therefore, we underestimate the total effect. 29. 1 gallon = 126,000 Btu. 30. EIA (2001).

216

Elena Safirova, Sébastian Houde, and Winston Harrington

Modeled Urban Scenarios and Policies Our simulations are divided into two categories. We first model three urban scenarios that consist of changes to the urban form of our modeling region, Washington, D.C. For the second category of simulations, we consider policies commonly discussed in the context of urban sprawl and energy consumption. Each of them aims to induce behavioral responses of the same nature as the ones observed under the three urban scenarios.

Urban Scenarios We assume here that drastic changes in urban form simply occur and that they consequently change the energy footprint of the metropolitan area, but we do not attempt to determine what policies might lead to such outcomes or even try to determine how difficult or costly such changes might be. These scenarios show, respectively, how changes in individual preferences, building density, and road capacity, consistent with denser urban forms, affect energy use. individual preferences. Individuals in our model choose where to work and where and in what type of housing to live. In addition to the dollar benefits, each choice of work location and housing type has a relative inherent attractiveness associated with it. In our model the inherent attractiveness of each option is characterized by a fixed parameter and is calibrated to match the observed data. With such characterization, it is relatively simple to consider a scenario in which the relative attractiveness of the choice bundles changes in a way consistent with preferences to live in a denser area. More specifically, we consider a scenario in which living inside the Beltway, the highway that rings the District of Columbia (Interstate 495 and part of Interstate 95, shown in figure 7-1), is considered 25 percent more attractive than it was previously. Under this assumption our model predicts that 124,000 residents move into the area. building density. In our model each residential building type (single family and multifamily) is associated with a structural density (square feet of floor space per acre). The structural density varies by zone to capture the fact that urban development in downtown Washington, D.C., is denser than development in the far suburbs of Frederick County, Maryland, for example. For this urban scenario, we consider that all residential development inside the Beltway is 20 percent denser and that the increase in density comes only from additional SFA and MF5 housing units. road capacity. Our model describes in a stylized but disaggregated manner the transportation network of metropolitan Washington, D.C. Capacities of arterial roads and freeways are determined by a nonlinear relationship between

Spatial Development and Energy Consumption

217

Figure 7-1. Map of the Washington, D.C., Metropolitan Area

the speed and vehicle miles traveled on that road.31 By scaling these relationships, it is possible to simulate changes in road capacity. Lower road capacity is consistent with denser urban area. However, a reduction in road capacity will increase congestion, ceteris paribus. And the converse is true for higher road capacity. In the model, an increase in congestion might have an ambiguous effect on residential pattern. In response to higher congestion costs, people can switch routes, time and mode of travel, but also residence and work locations. If this latter type of response is significant, lower road capacity could lead to a more dispersed city. To better understand this complicated series of effects, we model two scenarios. In the first one we increase the road capacity of the transportation network situated inside the Beltway by 25 percent. In the second simulation we consider a 25 percent decrease in road capacity.

Urban Policies For the policies, the urban forms respond to the design of the policies and are not changed according to ad hoc assumptions. We model three policies. The 31. In START this relationship is described by speed-flow ⫻ distance curves. See Houde, Safirova, and Harrington (2007).

218

Elena Safirova, Sébastian Houde, and Winston Harrington

first one is the so-called Live Near Your Work (LNYW) program, which is a policy that provides a monetary incentive to live closer to the work location. Like the scenario of a change in preferences in favor of higher density, this policy would be expected to induce residents to move toward the center of the region. Second, we model an inclusionary zoning (IZ) program, which is a policy that requires developers to provide affordable housing and denser development. This policy induces an effect similar to that of the scenario in which building density increases. Finally, we consider a VMT tax; this policy, like a change in road capacity, directly affects the cost of driving. The rest of this section further describes each of the simulations. live near your work program. In the Washington, D.C., metropolitan area, LNYW programs have already been established.32 In this region and elsewhere, the design of LNYW programs usually intends to provide a closing cost assistance grant to first-time home buyers who choose a property within a certain distance from their work location. For our simulations, we modeled a LNYW program that provides a closing cost assistance grant of $8,000, which is in the higher range of existing programs.33 To provide an illustrative example that fits the requirements of the model, we relax some of the eligibility criteria. For example, because the exact geographic locations of the buildings within the model’s zones are not defined, it is not possible to consider eligibility criteria based on a specific distance from residence to workplace. For our simulation, we consider a LNYW program that provides a grant to residents living and working inside the Beltway. Unlike existing programs, we assume that residents meeting this location criterion before the start of the policy and ones moving to meet the criteria both receive the grant. Our version of LNYW is therefore more generous but also more consistent with an economy in long-run equilibrium.34 inclusionary zoning program. Montgomery County in Maryland implemented the first IZ program in the United States in 1973. Since then, 135 32. See appendix table 7A-2. 33. In our model, individuals make decisions on the basis of their annual income, not permanent income. To be consistent with the model assumptions, the LNYW grant must then be converted into annual payments. For the conversion, we assume that the grant corresponds to the net present value of an ordinary annuity (that is, an annuity payable at the end of each period) payable over thirty years at a 5 percent interest rate. Under these assumptions, the grant corresponds to an annual payment of approximately $520. 34. Our version of the LNYW program is notably more generous because it is not restricted to first-time home buyers. This is due to the fact that homeownership is not modeled explicitly. Considering that residents meeting the location criterion before the start of the policy also benefit from the grant is consistent with our assumption that the prediction of the model is a long-run equilibrium. It implicitly assumes that the grant is capitalized in the value of the land and property and thus also accrues to these existing residents, not only to the new ones. This is a realistic assumption since we have perfect competition, low elasticity for the supply of land, and a permanent policy.

Spatial Development and Energy Consumption

219

cities and counties have adopted a similar program.35 Different IZ programs are quite similar in their design. In a nutshell, they have requirements for any new development, exceeding a certain threshold in the number of units, to set aside affordable housing units. To make the building of affordable housing more attractive, a density bonus is assigned to the project, which allows developers to increase the density in excess of the existing zoning and land use laws.36 Different criteria of eligibility for affordable housing exist. In addition to income, such factors as age, disabilities, place of work, and household size are also considered. Given that the supply of affordable housing units is most likely below the demand, the units are usually assigned by a way of lottery. Once occupied, the units have a control period during which the units cannot be sold or rented at the market rate. To better understand the effects of the IZ program on energy use, we simulate such a program in a counterfactual way. In the present case, we create a counterfactual scenario of a world where an IZ program has been implemented over the entire area inside the Beltway for all past developments and has been fully successful. Therefore, under this scenario there is a stock of affordable housing units corresponding to the one specified by the program, and simultaneously the urban density across the county is increased by the full percentage corresponding to the bonus provided.37 Appendix table 7A-3 provides more details about the assumptions of our simulated IZ program. vmt tax. Since our principal policy target is transportation, we simulate a tax of $0.10 per vehicle mile traveled (VMT) in the metropolitan area.38 The revenues collected are redistributed equally and as a single payment to all residents. The current gasoline tax, which is about $0.40 per gallon, is unaffected by this VMT tax. In our model, this VMT tax is implemented as a change in the price of fuel. Since fuel economy depends on the speed of vehicles on various roads, our tax, strictly speaking, is not a pure distance tax, but it is very close to one in its effects.

Discussion of Results On the basis of the simulation results, we can draw important conclusions. First of all, except for the urban scenarios where road capacity is changed, all other 35. Smart Growth Network (1995). 36. In practice, both mandatory and voluntary IZ programs are used. 37. In a sense, since we assume that the program is fully successful, the results represent an upper bound on the effectiveness of an IZ program. 38. This VMT tax is set at the “optimal” level, that is, it maximizes residents’ welfare in terms of congestion reduction and redistribution of the tax revenues to the residents. It should be noted that the welfare function does not take into account the benefits of other outcomes, including the effect on energy conservation.

220

Elena Safirova, Sébastian Houde, and Winston Harrington

scenarios induce energy savings, and all policies improve the economic welfare of residents. Second, for all scenarios and policies, the overall changes in vehicular energy use are more significant than the changes in residential energy use. This is true in absolute and in percentage terms. The total energy savings observed are primarily a consequence of a reduction of vehicle fuel consumption. Third, except for the VMT tax, the potential for energy savings of all scenarios and policies is low: less than 1 percent. Nevertheless, the population does move to denser building types, as shown in table 7-4. We refer readers to tables 7-2 to 7-4 for more details. Table 7-2 presents changes in residential and vehicular energy use for the different urban scenarios and policies. For the three policies, welfare changes are also reported. Tables 7-3 and 7-4 separate vehicular energy use from residential energy use to show which sector is more responsive. In the rest of this section, we further investigate the source of these results for each scenario and policy.

High Preferences to Live Inside the Beltway As noted earlier, the exogenous increase in the inherent attractiveness of living inside the Beltway induces 124,000 individuals to move into the area. This corresponds to a 10.69 percent increase in the population. However, the population living in the periphery outside the Beltway decreases by 4.19 percent (table 7-5). This scenario significantly increases the density of the population distribution in the Washington Metropolitan Area. On the one hand, under this scenario, there are few individuals who substitute from single family to multifamily housing. On the other hand, residents moving inside the Beltway are now more likely to live in single family units attached (SFA) rather than in detached units (SFD). Similarly, individuals choosing to live in multifamily housing but moving inside the Beltway are now more likely to be in an apartment building with five units or more (MF5).39 The reduction in residential energy use comes from the latter substitution. Even though residential energy use increases in single family attached and in multifamily units, a decrease in energy consumption in detached single family houses dominates (table 7-4). On the transportation side, the movement of the population affects VMT (table 7-6), mode choice, and congestion. People living inside the Beltway have shorter travel distances for commuting and shopping trips, which decreases the average trip distance (table 7-7). They have also better access to public transit 39. As described in appendix A-2, the disaggregation of the population distribution from single family housing into SFA and SFD categories and from multifamily housing into MF24 and MF5 are based on exogenous coefficients, which are indexed by zones. Table 7A-1 shows that in the central zones, the proportion of SFA relative to SFD is higher. The same is true for the proportion of MF5 relative to MF24.

Source: Authors’ calculations. … = Not applicable. a. 1 gallon = 126,000 Btu.

Policy Live Near Your Work program enacted inside the Beltway Inclusionary zoning program enacted inside the Beltway VMT tax of $0.10 per mile

Urban scenario High preferences to live inside the Beltway area Increase in residential housing density inside the Beltway area Increase in road capacity of 25 percent inside the Beltway Decrease in road capacity of 25 percent inside the Beltway

Category

–0.01

–208.2

…

–133.7

305

–35,139.7

–737.4

–0.01

–9.5

1051

–0.09

–226.4

–0.004

–6.9

94

–16.10

–0.34

–0.10

–0.29

0.20

428.3

0.01

118.7

… –634.6

–0.28

–618.1

–0.12

–194.1

–0.78

Percent change

…

–1704.4

Billions of Btu

–0.07

Percent change

Annual change in vehicular energy use (gasoline, converted into Btu)a

–115.7

Billions of Btu

Annual change in residential energy use (end use)

…

Millions of dollars

Overall welfare gains

Table 7-2. Welfare Gains and Energy Savings

–35.273.4

–746.9

–233.3

–655.4

440.1

–812.2

–1820.1

Billions of Btu

–10.39

–0.20

–0.06

–0.18

0.12

–0.22

–0.49

Percent change

Total change in energy use

6.93

Model baseline

Source: Authors’ calculations. … = Not applicable. a. 1 gallon = 126,000 Btu.

1,453

5.81 1,732

1,726

1,727

6.91

6.91

1,736

6.94

1,730

1,727

6.91

6.92

1,719

Millions of gallons

6.87

Millions of gallons

Policy Live Near Your Work program enacted inside the Beltway Inclusionary zoning program enacted inside the Beltway VMT tax of $0.10 per mile

Urban scenario High preferences to live inside the Beltway area Increase in residential housing density inside the Beltway area Increase in road capacity of 25 percent inside the Beltway Decrease in road capacity of 25 percent inside the Beltway

Category

Daily fuel consumption (gasoline)

Table 7-3. Annual Change in Vehicular Energy Use

218,253

183,114

217,516

218,027

217,619

…

–16.10

–0.34

–0.10

–0.29

0.20

–0.28

217,635 218,682

–0.78

Percent Change

216,549

Billions of Btua

Annual (250 days) fuel consumption (gasoline, converted into Btu)

…

–279

–6

–2

–5

3

–5

–14

Annual changes in millions of gallons

…

–35,139

–737

–226

–634

428

–618

–1,704

Annual changes in billions of Btu

Annual change in fuel consumption

Source: Authors’ calculations.

Model baseline (billions of Btu)

Policy Live Near Your Work program enacted inside the Beltway Inclusionary zoning program enacted inside the Beltway VMT tax of $0.10 per mile

Urban scenario High preferences to live inside the Beltway area Increase in residential housing density inside the Beltway area Increase in road capacity of 25 percent inside the Beltway Decrease in road capacity of 25 percent inside the Beltway

Category

Percent

3,007

10,638

156,672

–0.09

–0.01 –0.27

0.30

–0.11 39,524

–0.01

–0.08 –5.02

5.00

–1.83

103,488

–0.004 0.04

–0.01

–0.01

0.01

–0.03

0.01

–0.01

0.01

0.004

–0.004

0.02

–0.01

0.14

–0.12

0.62

–4.27

4.55

–1.81

–0.06

–0.17

All types

2.63

Apartment in multifamily housing of 5 units or more

1.57

Apartment in multifamily housing of 2–4 units

0.44

Single family attached housing unit

–0.75

Single family detached housing unit

Table 7-4. Annual Change in Residential Energy Use, by Housing Type

Source: Authors’ calculations.

1,164,632

0.79 5.62 0.66

Policy Live Near Your Work program enacted inside the Beltway Inclusionary zoning program enacted inside the Beltway VMT tax of $0.10 per mile

Model baseline (number of residents)

10.69 4.22 0.10 –0.18

Inside the beltway

Urban scenario High preferences to live inside the Beltway area Increase in residential housing density inside the Beltway area Increase in road capacity of 25 percent inside the Beltway Decrease in road capacity of 25 percent inside the Beltway

Category

Percent

Table 7-5. Change in Population

2,974,502

–0.31 –2.20 –0.26

–4.19 –1.65 –0.04 0.07

Outside the beltway

172,594 172,297 147,430 172,461

Policy Live Near Your Work program enacted inside the Beltway Inclusionary zoning program enacted inside the Beltway VMT tax of $0.10 per mile

Model baseline

Source: Authors’ calculations. … = Not applicable.

171,620 172,390 173,175 172,002

Daily VMT (thousands of vehicle miles)

Urban scenario High preferences to live inside the Beltway area Increase in residential housing density inside the Beltway area Increase in road capacity of 25 percent inside the Beltway Decrease in road capacity of 25 percent inside the Beltway

Table 7-6. Daily Change in Vehicle Miles Traveled

…

132 –163 –25,031

–841 –71 713 –459

…

0.08 –0.09 –14.51

–0.49 –0.04 0.41 –0.27

Daily change in VMT Thousands of vehicle miles Percent change

Source: Authors’ calculations.

Model baseline

Policy Live Near Your Work program enacted inside the Beltway Inclusionary zoning program enacted inside the Beltway VMT tax of $0.10 per mile

Urban scenario High preferences to live inside the Beltway area Increase in residential housing density inside the Beltway area Increase in road capacity of 25 percent inside the Beltway Decrease in road capacity of 25 percent inside the Beltway

Category

Percent

12.27 miles

–2.42

–2.11 7.92 miles

–0.18

0.06

–0.08

–0.14

–0.03

0.05

–0.25

–0.29

–0.18

–0.03

–0.01

Road network outside the Beltway

–0.16

Road network inside the Beltway

11.02 miles

–2.30

–0.40

–0.13

45.84 mph

0.80

0.56 42.49 mph

0.32

0.14

0.17

–2.04

0.003

0.09

–0.07

1.37

–0.04

0.37

0.26

–0.42 0.001

0.28

Road network outside the Beltway

–0.67

Road network inside the Beltway

Average speed (car only)

–0.65

All road network

Average trip distance (car only)

Table 7-7. Changes in Average Trip Distance and Average Speed of Travel over the Road Network

44.98 mph

0.71

0.26

0.05

–0.34

0.27

0.33

–0.01

All road network

Source: Authors’calculations. HOV = high occupancy vehicle; SOV = single occupancy vehicle.

10,092,421

515,198

646,105

1,689,457

21.71 20.21 16.41

16.41

1,1705,440

2.13 4.69 3.46

0.66

Model baseline (trips per day)

0.72 1.19

0.85

–17.80

1.11 3.70

0.48

–0.087

–0.39

–0.024

–0.62

–2.17

–0.19

0.036

0.23

0.14

1.99

4.96

3.52

0.79

0.043

0.024

3.65

Walking and biking

3.55

Train

2.70

Bus

0.25

HOV

–0.21

SOV

Policy Live Near Your Work program enacted inside the Beltway Inclusionary zoning program enacted inside the Beltway VMT tax of $0.10 per mile

Urban scenario High preferences to live inside the Beltway area Increase in residential housing density inside the Beltway area Increase in road capacity of 25 percent inside the Beltway Decrease in road capacity of 25 percent inside the Beltway

Category

Percent

Table 7-8. Changes in Travel Mode

228

Elena Safirova, Sébastian Houde, and Winston Harrington

and thus make greater use of it (table 7-8). Yet the significant increase in population contributes to congestion of the road and transit networks. The increase in congestion can be seen from the lower average speed of travel on the roads. For the present scenario, average speed decreases inside the Beltway by 0.67 percent (table 7-7). Changes in speed also affect vehicle fuel consumption rates, but unlike for VMT, the relationship is not straightforward.40

Increase of Residential Housing Density Inside the Beltway Under this scenario, we have simulated a 20 percent increase in the density of urban development inside the Beltway and further stipulated that the increase was from additional SFA and MF5 housing units only. This simulation is equivalent to an increase of 20 percent in the supply of such units. We first observe that this increase in supply lowers market rents in the area. This attracts a number of individuals to live there. Approximately 49,000 individuals move (which corresponds to a 4.22 percent population increase inside the Beltway and a 1.65 percent population decrease outside the Beltway; see table 7-5). Compared with the previous scenario, fewer individuals move, but there is a greater decrease in residential energy use (table 7-4). This is due to the fact that the increase in housing density comes only from new SFA and MF5 units, which are associated with lower energy use. Even if the movement in population has a similar pattern to that of the preceding scenario, the nature of the transportation effects is different. The reduction in VMT is basically nil (table 7-6), but the reduction in average trip distance is noticeable (table 7-7). Unlike in the preceding scenario, the movement of population inside the Beltway does not induce congestion; in the present scenario, average speeds on the roads inside and outside the Beltway increase (table 7-7). Transit usage increases, but so do car trips (table 7-8). In sum, on average trips are shorter, faster, and more likely to be done by transit. However, the overall number of trips increases, but VMT remains almost unchanged.

Changes in Road Capacity. As mentioned in the section “Details of Energy Modeling” above, lower road capacity is associated with denser development but also with congestion, which could have the perverse effect of leading to a more dispersed metropolitan landscape. For example, in our scenarios, a 25 percent decrease in road capacity inside the Beltway induces individuals and firms to relocate to the peripheral zones. The effect is, however, small: only 2,000 individuals move (table 7-5), and the increases in production in the different zones and economic sectors never exceed 40. In our model, the relationship between speed and fuel consumption is characterized by a fourth degree polynomial.

Spatial Development and Energy Consumption

229

0.15 percent. For an increase in road capacity, the nature of the effects is the same, but of the opposite sign. Furthermore, the movement is even smaller: 1,200 individuals move inside the Beltway (table 7-5). For these two scenarios, these small changes in population distribution have few effects on residential energy use. The transportation effects are more important, however. It is interesting that even though higher road capacity decreases congestion (table 7-7), overall, it leads to higher VMT (table 7-6). Part of the reason is that travelers abandon public transit and opt for cars as a means of transportation (table 7-8). Therefore, car fuel consumption increases. Given the small changes in residential energy use, this scenario leads to higher energy consumption. The opposite is true for a reduction in road capacity.

Live Near Your Work Program Inside the Beltway In a within-Beltway LNYW program, 9,250 individuals decide to move from outside to inside the Beltway (table 7-5). Furthermore, economic activity is relocated, to a certain extent, to this area. This displacement is caused by two factors. First, there is the movement of population. Second, the additional income from the LNYW grant is spent primarily at shopping destinations inside the Beltway. As a result, the within-Beltway LNYW program induces not only the workforce to move but also encourages the relocation of the firms that benefit from the increase in the labor supply and the demand for goods and services in the area. Although people move inside the Beltway, the magnitude of the change is too small to affect residential energy use significantly. The policy leads to an overall increase in VMT of 0.08 percent (table 7-6). It is particularly interesting to note that under the within-Beltway LNYW program the influx of population inside the Beltway does not reduce VMT. Overall the average speed slightly increases (table 7-7), which is the only cause for the small reduction in vehicle fuel consumption (table 7-3). This policy is still successful in achieving its primary objective: reducing commuting distance. For commuting trips only, we observe a reduction of 0.21 percent in the average trip distance, which results in a reduction of 0.44 percent in VMT. It is the increase in consumption inside the Beltway, fueled by the LNYW grant, that has the unintended consequence of leading to more shopping trips (and more trips overall; table 7-8) and thus mitigates the decrease in the overall VMT achieved by commuters. Therefore, this policy has a small effect on vehicle energy consumption.

Inclusionary Zoning Program Enacted Inside the Beltway The IZ program has an important effect on the population distribution. More than 65,000 individuals move inside the Beltway; this is about 15,000 more than that in the urban scenario in which only the housing density increases by

230

Elena Safirova, Sébastian Houde, and Winston Harrington

20 percent. The requirement for affordable housing is therefore an effective way to attract new residents. Of interest, the overall change in residential energy use in the IZ program is smaller than it is in the urban scenario with higher density. The reason for such a result is that the IZ program creates a greater incentive to move to an attached single family house (SFA) rather than to an apartment in a building of 5 units or more (MF5). Our assumptions regarding the design of the policy are an important factor. A higher average rent, or lower set aside requirement for the affordable single family housing units, could lead to different results. The present policy has been modeled in accordance with existing IZ programs, but the results are sensitive to the details of program rules. If the movement in population has little effect on the overall residential energy use, this is not true on the transportation side. VMT decreases by 0.09 percent partly as a result of shorter average trip distance (table 7-6). Coupled with substitution in mode choices toward public transit (table 7-8), this is enough to relieve congestion, as shown by the higher average speeds of travel (table 7-7). vmt tax. The VMT tax is the policy that achieves the largest reduction in energy use. As expected, the bulk of the reduction comes from transportation, but it is interesting that the VMT tax is effective at reducing residential energy use. For example, it does better than the infill policies. The VMT tax also induces individuals to move to the center of the region. Almost 11,000 people move inside the Beltway. The increase is particularly concentrated in the District of Columbia and Arlington County, Virginia. These are the two places where the share of SFD relative to SFA and the ratio of MF5 relative to MF24 are the highest across the metropolitan area. It explains why the decrease in residential energy use in those areas is significant. The VMT tax, because of the substantial decrease of travel that it induces, is an effective way to increase urban density. On the transportation side, the decrease in fuel use is drastic. This decrease is due to the important decline in VMT, which is caused by three things. First, people move to the center of economic activity; therefore, the average trip distance to work and shopping locations is consequently reduced (table 7-7). Second, there is an important switch to public transit and nonmotorized modes of travel (table 7-8). Both of these effects contribute to congestion relief (table 77). The third and a more subtle cause of VMT reduction comes from the fact that the VMT tax causes some people to stop working and, therefore, to stop commuting. The model assumes that the vast amount of revenue collected from the VMT tax—nearly $1.18 billion per year—is distributed equally among all residents of the metropolitan area. Some workers facing high commuting costs

Spatial Development and Energy Consumption

231

and simultaneously receiving a generous tax rebate would simply prefer not to work because their commuting costs are so high and because the tax rebate is large enough to enable them to afford not to work. This is particularly true for low-income individuals for whom commuting costs represent a larger share of their budget and to whom the tax rebate is more valuable.41

Other Potential Urban Policies The list of policies that could affect energy use in urban areas is very long. Here we do not intend to cover a comprehensive list of policies but will try to discuss three large groups of potentially relevant ones.

Congestion Pricing A set of policies that are likely to affect an urban energy footprint are various transportation policies that are often called congestion pricing policies. Although particular policy schemes may significantly vary, the best-known, real-world example of such policies is the London Congestion Pricing Scheme that was implemented in 2003; considered a success, it has doubled the charge area since then.42 Although congestion pricing is still facing significant opposition, pricing experiments are planned or underway in many U.S. metropolitan areas. Our previous research has shown that congestion pricing schemes hold the promise to reduce road congestion and improve the well-being of urban travelers. At the same time, those policies are likely to have much less impact on energy consumption. In another study, we analyzed five distinct congestion pricing schemes alongside the same level of VMT tax as the one analyzed above. In particular, we modeled a comprehensive toll (congestion pricing on all roads), a freeway toll (the same as comprehensive but with pricing only on highways), and three different cordon-type pricing schemes (where entering a particular area of the city incurs a toll) covering central parts of metropolitan areas (table 7-9). It should be noted that these policies are more extensive than any such policies currently in place anywhere in the United States, and thus they provide a fair indicator of the potential for such policies to achieve both congestion reductions 41. The assumption of lump sum distribution, that is, that the tax rebate is distributed equally to all residents, is highly unrealistic. It is more likely that the funds will enable reductions of existing taxes, new public works spending, or some combination of the two. For example, some of those who now support taxing automobile use, especially those in the environmental community, condition their support on the use of a substantial portion of the revenues for investment in transit. Others try to earmark the funds for new road construction. 42. New York City proposed a similar scheme in 2007 and received federal funding for the project.

232

Elena Safirova, Sébastian Houde, and Winston Harrington

Table 7-9. Six Second-Best Transportation Policies: Optimum Fees and Effects on VMT

Policy

Toll rates, Average cost Percentage of where charged per VMT VMT affected (dollars) (cents per mi.)

VMT tax Comprehensive toll Freeway toll Double cordon Beltway cordon Downtown cordon

100 100 26 7a 7a 1.1a

0.10 per mile Variable Variable D: 2.18 B: 3.43 B: 2.77 D: 4.70

Base case

7.9 3.3 0.7 0.4 0.3 0.2

Total estimated VMT (millions percent (per day change 147.4 160.5 169.0 170.5 171.1 171.3

–14.6 –7.1 –2.1 –1.3 –0.9 –0.8

172.7

Source: Harrington, Houde, and Safirova. (2007). B = Beltway; D = downtown. a. Percentage of trips, not VMT (vehicle miles traveled).

and welfare gains. We found that although yielding the highest welfare gains (660 million dollars annually) the comprehensive toll turned out to be less effective at reducing total VMT in the metropolitan area than the VMT tax (7.1 percent as opposed to 14.6 percent). At the same time, from tables 7-2 and 7-3, we know that the VMT tax policy appears to be effective primarily because it succeeds at significantly reducing vehicle miles traveled. To the extent that other transportation policies are less effective at reducing vehicle miles, they will yield only a fraction of a reduction in energy use. The same logic will apply to policies, including the VMT tax, imposed at levels much lower than the optimal level. For example, it should not be surprising that a $0.01 per mile VMT tax would reduce total VMT and therefore energy consumption by only a fraction of the amount by which a $0.10 VMT tax would.

Nonurban Policies Many of the policies that could affect urban structure, perhaps most, are not designed to do so, but they could have unintended consequences. This very likely would be true of most federal policies, which are concerned with energy use more at the national level than at the local level. Consider, for example two prominent federal policies, the Corporate Average Fuel Economy (CAFE) standards and the home mortgage interest deduction. CAFE standards may reduce fuel use in vehicles by improving fuel economy, but they also reduce the fuel costs of driving, thereby potentially increasing demand.

Spatial Development and Energy Consumption

233

Experiments with CAFE in our model showed that this so-called rebound effect will increase vehicle use by about 8 percent and presumably accidents and traffic congestion as well.43 We have not modeled the mortgage interest deduction, but it is possible that its effects would be important. By reducing the cost of housing, this tax break encourages consumers to build larger houses on larger lots, increasing the demand for local travel and residential heating and cooling.

Local Policies There are many local policies that also have the potential to affect energy use, of which we mention two. One is increased transit use. We have used our model to estimate the benefits of the local transit system in the Washington, D.C, metropolitan area. We find that its benefits are very large.44 However, we have not looked in any detail at the question of whether enhanced transit can reduce energy use in the metropolitan area. We think it is unlikely, mainly because even with very vigorous policies to encourage transit use the overwhelming majority of trips in the metropolitan area will continue to be made by car. Transit is very effective for certain kinds of trips, in particular rush hour trips into downtown Washington, D.C., and to Northern Virginia. For trips to other locations and at other times, its mode share is very low. Nonetheless, transit provides valuable services at all times to the poor and disabled. We should mention also transit-oriented development (TOD), a collection of policies to encourage economic development around rail transit stations. Locating housing and employment development within walking distance of transit stations will, according to theory, reduce the demand for work trips and perhaps other types of trips as well, to the extent that such centers become magnets for other types of development. Unfortunately, our model’s ability to analyze TOD is limited by the necessarily small number and large size of the spatial areas represented in the model. TOD is a policy that is focused on a small geographic area; in fact, the limit that people can be reliably induced to walk is between a quarter and a half mile. Because our model divides the Washington, D.C., metropolitan area into fairly large geographic zones, we cannot model this policy effectively. However, we are skeptical about TOD’s potential to significantly reduce energy consumption. The small size of the zones could put a limit on the overall scale of TOD, regardless of how successful it is in individual applications. After all, the number of transit stops, an essential ingredient of a TOD policy, is limited.

43. Parry, Fischer, and Harrington (forthcoming). 44. Nelson and others (2007).

234

Elena Safirova, Sébastian Houde, and Winston Harrington

Concluding Remarks In this chapter, we have simulated several urban scenarios and policies with the purpose of analyzing their comparative impact on energy consumption. Important novel features of the analysis include the interaction between land use and transportation decisions built into our framework and the inclusion of both transportation-related and building heating and cooling–related effects. We find that a VMT tax of $0.10 per mile has the potential to substantially reduce energy consumption while all other policies, as well as the hypothetical scenarios, are likely to be much less effective. But even if the effects of the VMT tax are larger than those of the other policies examined, at a reduction of 10 percent, they are still modest, especially when compared with the scale of the intervention. After all, a tax of $0.10 for each mile traveled is a very large tax. The low sensitivity of travel to the costs of trips suggests that policies to reduce fuel use directly might be more promising approaches. Trying to devise urban policies to reduce energy consumption is a dubious enterprise for another reason. As noted in the beginning of the chapter, the two main reasons to be concerned about fuel consumption are global climate change and energy security. If there are any benefits from urban policy to reduce energy use, they will be enjoyed nationally, if not globally. The metropolitan area implementing the policy cannot capture its benefits. If urban policy that was adopted for other reasons incidentally reduces energy consumption, that is all well and good. But policies with the main goal of reducing energy consumption should be national, not local. Localities, on the other hand, would do much better for themselves if they designed urban policies to correct local externalities— congestion, local air pollution, provision of public open space—and left the energy policy to federal and international entities.

Appendix A-1. Model Description Our model, called LUSTRE, combines two preexisting models: the Regional Economy and Land Use (RELU) model and the Strategic and Regional Transport (START) modeling suite. The RELU model was developed by Alex Anas and Elena Safirova with the purpose of creating a theoretically sound modeling tool for the analysis of interactions between transportation, land use, and economic activity. The model is meant to be integrated with a detailed transportation model. The START modeling suite was developed by MVA Consultancy.45 More recently, this model was calibrated for the Washington, D.C., metropolitan area (referred to as Washington START) and used to conduct a wide range of 45. May, Roberts, and Mason (1992).

Spatial Development and Energy Consumption

235

policy simulations.46 LUSTRE is calibrated for the Washington, D.C., metropolitan region for the year 2000; the transportation network and characteristics of the economy both are specific to this region.

RELU Description RELU is a spatially disaggregated static computable general equilibrium model that represents the long-term economic equilibrium in a regional economy. For this chapter, the spatial representation corresponds to the Washington, D.C., metropolitan area divided into thirty-six economic zones, plus four outer zones that act as sinks, that is, they attract economic activity but are not equilibrated together with the model zones. RELU follows the structure of Alex Anas and Rong Xu in its modeling philosophy, although several new features have been added.47 For an exposition of the model in greater detail, we refer the reader to Elena Safirova and others.48 The following presents the salient features of the model. RELU has seven types of economic agents that are explicitly represented. There are four types of individuals that correspond to a given skill level. These individuals are the consumers and the workforce in the regional economy; their total number is held fixed across simulations. The three other agent types are producers, landlords, and developers. Although the government is not explicitly modeled, income and property taxes are present. Individuals maximize their utility based on a series of discrete and continuous choices. After deciding whether to work or to be unemployed, individuals choose a triple corresponding to their residential and workplace locations and type of housing. Conditional on these discrete choices, individuals decide how much housing to rent and how much retail goods and services to purchase at each available retail location. The costs of traveling to a given work or shopping destination are taken into account. Although leisure is not represented in the model, aggregate labor supply is elastic because of voluntary unemployment and the variation in time spent traveling to shop. The production sector consists of four basic industries: agriculture, manufacturing, business services, and retail; construction and demolition industries are represented as well. The producers are perfectly competitive profit-maximizing agents, with a Cobb-Douglas production function between four groups of inputs: labor, capital, buildings, and intermediate inputs. At the same time, within input groups, substitution is characterized by a constant elasticity of substitution (CES) function. All primary industries, except retail, provide interme46. Safirova and others (2004); Safirova and others (2005); Safirova, Gillingham, and Houde (2007). 47. Anas and Xu (1999). 48. Safirova and others (2006).

236

Elena Safirova, Sébastian Houde, and Winston Harrington

diate inputs to other sectors. Retail output is consumed by individuals only or exported out of the economic region. Prices of intermediate inputs include freight costs; hence, firm reallocation is affected by the costs of shipping goods. Landlords manage floor space in a profit-maximizing way in a perfectly competitive market. Rents and operating costs are taken as given. Landlords decide whether or not to offer a unit amount of floor space on the rental market. Floor space in buildings is disaggregated into four types: single family housing, multifamily housing, commercial use, and industrial use. Developers, like landlords, are profit-maximizing agents. They determine how much vacant land should be converted into buildings or vice versa. Construction and demolition prices and other costs are taken as given. Potential rents for the building also affect developers’ decisions. Each individual owns a certain share of the real estate.

Washington START Description START contains two submodels referred as the supply-side and the demandside models. The supply-side model consists of the transportation network disaggregated into forty travel zones (START’s travel zones correspond to RELU’s economic zones, thirty-six plus four). Each zone has three stylized transportation links (inbound, outbound, and circumferential) and a number of other “special” links that represent the principal highway segments and bridges of the region. The traffic quality for each link is characterized by a speed-flow curve. The rail network of the region combines the Washington Metrorail system and suburban heavy rail systems: Maryland Rail Commuter (MARC) and Virginia Railway Express (VRE). The three rail systems are modeled. A highly stylized route network represents bus travel, with bus accessibility in any zone determined by the density of stops, frequency of service, and reported bus travel times. Transit crowding costs and parking search costs are explicitly included in the model. The model also accounts for existing high-occupancy vehicle (HOV) lanes. The supply-side model computes the generalized cost of travel, taking into account the time and monetary elements of traveling. Time components include the time spent traveling, transit waiting time, parking search time, and transit crowding penalties. Monetary components include car operating costs, car depreciation costs, parking fares, tolls, and transit fares. The value of time is a function of the travelers’ wage rate and varies by trip purpose. The demand-side model is a strategic model centered on nested logit models. In START, trip purposes and origins are taken as given. Agents choose whether or not to generate a trip, the destination, mode, time of day, and route (in LUSTRE, trip generation and destination are delegated to RELU). Nest order may be interchanged for different purposes. The model distinguishes four travel

Spatial Development and Energy Consumption

237

modes: single-occupancy vehicle (SOV), HOV, transit (which has two submodes: bus and rail), and nonmotorized travel (walking and biking). It also represents three time periods: morning peak, afternoon peak, and off peak. Travelers maximize their utility of travel according to a generalized cost of travel that combines time and money costs explicitly modeled in the supply module, as well as idiosyncratic preferences. The overall structure of START is iterative. The trips computed in the demand-side model are loaded into the supply-side network, which uses the loads to compute costs of travel that are passed back to the demand-side module. This process iterates until the costs of travel converge to equilibrium values.

Model Integration Figure 7A-1 summarizes RELU and START and the integration procedure. First, RELU takes the time and monetary costs of travel as given. The RELU simulation yields trip demands (in addition to other land use and economic effects), disaggregated by purpose and origin-destination pair, and wage rates. Trips are loaded into START, and RELU-determined wage rates translate into value of time for START. Thereafter, START computes the generalized costs of travel. Any transportation policies are taken into account in this step. Computed generalized costs are sent back to RELU. This iterative process between the two models continues until trip demands and costs converge. Although the LUSTRE model has an intrinsically dynamic structure, in its present stage of development, we exercise the model in its long-term equilibrium version. Therefore, when policy simulations are performed, we obtain results that correspond to changes in the long-term equilibrium of the urban area.

Welfare Measure The strength of LUSTRE resides in its ability to compute welfare measures, which account for the changes in transportation, as economic variables. LUSTRE’s welfare measures are provided by RELU and are based on a utility function for individuals. We posit that the utility function of consumers is CobbDouglas between housing and aggregate consumption, while the subutility of all retail goods is a constant elasticity substitution (CES) function. Wages and prices of retail goods are net of travel costs. The utility of an individual is conditional on one’s discrete choices regarding place of work, residence, and housing type. Each discrete choice bundle has an inherent attractiveness associated to it. Finally, individuals have idiosyncratic utility, which differs by consumers. Assuming that they are iid (independently and identically distributed) Gumbel with dispersion probability, this gives rise to multinomial logit choice probabil-

238

Elena Safirova, Sébastian Houde, and Winston Harrington

Figure 7A-1. Flow Diagram of LUSTRE

ity. The welfare measure for workers of skill class f in the model can then be expressed as a logsum: Wf =

1 λ U ln ∑ e f ijk| f , λ f ijk

where ␭f is the dispersion parameter for the distribution of unobserved charac~ teristics of workers of skill class f and U i j k | f is the indirect utility function for

Spatial Development and Energy Consumption

239

such workers conditional on residential location i, employment location j, and housing type k. See the mathematical appendix in Safirova and others for more details.49

Appendix A-2. Energy Modeling Details Because LUSTRE only considers two building types (single family, SF, and multifamily, MF, housing), we use data from Census 2000 to further disaggregate the LUSTRE population into the four categories of building types considered by the Energy Information Administration.50 Census 2000 provides the distribution of households within these four building types for different regions of the Washington, D.C., metropolitan area. Therefore it is possible to split, exogenously, the number of household members living in SF housing into detached single family (SFD) and attached single family (SFA) categories. Similarly, for household members living in MF housing, they can be broken down into the categories of apartments in multifamily housing buildings of two to four units (MF24) and apartments in multifamily housing buildings of five units or more (MF5). Because the LUSTRE population corresponds to individuals, this disaggregation assumes that, for each zone, the ratio of individuals (household members) living in detached single family units to the number of individuals living in attached single family units is the same as for all households. This is a realistic assumption, given that the average household size is the same for the two building types.51 We further assume that this ratio is constant across skill level.52 For household members living in multifamily housing units, we have adjusted the ratio to account for the fact that households living in MF5 are on average smaller than households living in MF24 (2.0 persons per household compared to 2.3).53 Note that this disaggregation increases the accuracy of our scenarios. Yet, the decision to live in SFA or SFD, and analogously in MF24 or MF5, is exogenous. In this context, one could think that 49. Safirova and others (2006). 50. Energy Information Administration (2001). 51. Energy Information Administration (2001). 52. Intuitively, one might think that there is a greater proportion of higher-income individuals living in SFD as opposed to SFA units, particularly in the suburbs of Washington, D.C. It might also be true for MF24 relative to MF5, but that is less clear. Luxury apartments can be found in big structures in the region. The fact that we do not consider that the proportions of people living in SFD rather than SFA and MF24 rather than MF5 differ by income groups may cause some bias for policy simulations. Indeed, if the movement in population is not uniform across skill level, suppose that only less affluent are moving, a higher share of SFA units will be occupied, ceteris paribus. However, if the reverse is true, that is, highincome individuals move more, then more SFD units will be occupied. 53. Energy Information Administration (2001).

240

Elena Safirova, Sébastian Houde, and Winston Harrington

this choice is imbedded in LUSTRE’s individual decisionmaking and characterized by a Leontief relationship. Overall, the residential energy usage (REU) computed by LUSTRE is given by: ⎛ e SFA iri SFA i Popi , f ,SF + e SFD (1 − ri SFA )i Popi , f ,SF + ⎞ REU = ∑ ⎜ MF 5 MF 5 ⎟, iri i Popi , f , MF + e MF 24 (1 − ri MF 5 )iP Popi , f , MF ⎟⎠ if ⎜ ⎝e

where e is the energy consumption coefficient (millions of Btu per year) for each residential building type per household member;54 riSFA is the ratio of the number of household members living in SFA housing units to the number of households living in single family housing units (SF), disaggregated by zone i, from Census 2000 (see table 7A-1); riMF5 is the ratio of the number of households members living in MF5 units to the number of households living in multifamily housing units (MF), disaggregated by zone i, from Census 2000 (see table 7A-1); Popi,f,SF is the population of skill level f, living in zone i, and in a single family housing unit, from LUSTRE; Popi,f,MF population of skill level f, living in zone i and in a multifamily housing unit, from LUSTRE.

Appendix A-3. Baseline Energy Profile In the baseline case, out of the total of 156.6 billion Btu of annual residential energy use, 103.5 billion Btu are consumed by the residents of detached single family homes. The rest of residential energy consumption falls on the other three residential categories (SFA: 39.5 billion Btu, MF24: 3.0 billion Btu, and MF5: 10.6 billion Btu). One can observe that less than 10 percent of residential energy in the Washington, D.C., metropolitan area is consumed by residents of multifamily housing. At the same time, baseline vehicular energy use is 218 billion Btu. In other words, in absolute terms residential energy consumption and consumption of energy in transportation are of the same order of magnitude in the baseline picture, and therefore a potential reduction in any of them can make a dent in the total amount of energy used. Figures 7-A2 and 7-A3 show respectively the distributions of the population and per capita residential energy consumption by zone. Although there are 54. Energy consumption per household member amounted to 40.89 for SFD; 38.72 for SFA; 35.42 for MF24; 21.17 for MF5 (in millions of Btu); see EIA (2001).

Spatial Development and Energy Consumption

241

Table 7A-1. Share of Household Members Living in Different Housing Types for Different Regions of the Washington Metropolitan Area Percent Share of household Members living in SFA relative to single family housing

Share of household members living in MF5 relative to multifamily housing

District of Columbia

66.67

79.53

Inner core

Total Arlington County, VA Alexandria City, VA

37.98 24.98 58.74

90.17 88.38 92.52

Inner suburbs

Total Montgomery County, MD Prince George’s County, MD Fairfax County, VA Fairfax City, VA Falls Church City, VA

27.13 25.93 23.00 31.35 24.85 25.53

91.92 92.92 90.80 92.02 94.16 92.08

24.95 4.53 20.11 21.47 32.49 33.23 12.87 43.67 33.52

81.68 70.96 71.13 72.26 90.31 87.45 73.64 86.99 47.86

6.62 4.55

64.87 58.26

Region

Outer suburbs Total Calvert County, MD Charles County, MD Frederick County, MD Loudoun County, VA Prince William County, VA Stafford County, VA Manassas City, VA Manassas Park City, VA Far suburbs

Total Clarke County, VA

Source: U.S. Census Bureau, U.S Department of Commerce. U.S. Census 2000. MF5 = apartments in multifamily housing buildings of five units or more; SFA = attached single family housing units.

slight variations, one can see that per capita residential energy consumption does not vary significantly across the study area.

Appendix A-4. Urban Policy Details The descriptions of the existing Live Near Your Work (LNYW) programs in the Washington, D.C., metropolitan area are presented in table 7A-2. The details of the simulated inclusionary zone (IZ) program are given in table 7A-3.

242

Elena Safirova, Sébastian Houde, and Winston Harrington

Figure 7A-2. Population by Zone thousands

Population by Zone 302

237

172

108

43

Figure 7A-3. Annual Residential Energy Consumption

Spatial Development and Energy Consumption

243

Table 7A-2. LNYW programs in the Washington Metropolitan Area Program

Description

Live Near Your Work (LNYW) program, Maryland Department of Housing and Community Development

Closing cost assistance grant of 3 percent to borrowers who are purchasing a home within 10 miles of their place of employment or within the boundaries of the local jurisdiction where they are employed; available for first-time homebuyers.

DC Employer Assisted Housing Program (EAHP)

Grants and deferred loans of up to $11,500 to employees of the District of Columbia government who are first-time home buyers in Washington, D.C.

Alexandria, Virginia Employee Homeownership Incentive Program (EHIP)

Deferred payment, 0 percent interest loans up to $5,000 for public employees who purchase homes in the city of Alexandria.

Forgivable loan, forgiven at 1/36 per month. If the eligibility Arlington, Virginia Live Near Your Work program criteria are met for three years, loan becomes a grant. At least one family member must be a permanent full-time employee of Arlington County or the Arlington School Board. Government employees receive LNYW assistance of $5,400. School employees receive assistance of 1 percent of the purchase price, up to a maximum of $5,400.

Table 7A-3. IZ Program Simulated with LUSTRE Area covered

IZ program inside the Beltway

Set aside requirements for affordable housing

12 percent of units for multifamily housing; 15 percent for single family housing

Bonus density

20 percent of all units provided are set as SFA and MF5 housing units

Threshold number of units

No threshold

Eligibility criteria

Below median income (skill level 1 and 2 in LUSTRE); lottery determines the renters

Average rent

$10 per square foot, for multifamily housing; $5 per square foot for single family housing

Source: Authors’ simulation assumptions. IZ = inclusionary zoning; MF5 = apartments in multifamily housing buildings of five units or more; SFA = attached single family housing units.

244

Elena Safirova, Sébastian Houde, and Winston Harrington

References Anas, Alex, and Rong Xu. 1999. “Congestion, Land Use and Job Dispersion: A General Equilibrium Model.” Journal of Urban Economics 45, no. 3: 451–73. Anderson, William P., Pavlow S. Kanaroglou, and Eric J. Miller. 1996. “Urban Form, Energy and the Environment: A Review of Issues, Evidence, and Policy.” Urban Studies 33, no. 1: 7–35. Badoe, Daniel A., and Eric J. Miller. 2000. “Transportation-Land Use Interaction: Empirical Findings in North America, and Their Implications for Modeling.” Transportation Research D 5, no. 4: 235–63. Bento, Antonio M., and others. 2005. “The Effects of Urban Spatial Structure on Travel Demand in the United States”. Review of Economics and Statistics 87, no. 3: 466–78. Boarnet, Marlon, and Randall Crane. 2001. “The Influence of Land Use on Travel Behavior: Specification and Estimation Strategies.” Transportation Research A 35, no. 9: 823–45. Boarnet, Marlon, and Sharon Sarmiento. 1998. “Can Land Use Policy Really Affect Travel Behavior? A Study of the Link between Non-work Travel and Land Use Characteristics.” Urban Studies 35, no. 7: 1155–169. Carrol, T. Owen. 1978. “Calculating Community Energy Demand.” In Energy and the Community, edited by Raymond J. Burby and A. Flemming Bell, pp. 24–45. Cambridge, Mass.: Ballinger. Cervero, Robert, and Kang-Li Wu. 1997. “Polycentrism, Commuting, and Residential Location in the San Francisco Bay Area.” Environment and Planning A 29, no. 5: 865–86. ———.1998. “Sub-centring and Commuting: Evidence from the San Francisco Bay Area, 1980-90.” Urban Studies 35, no. 7: 1059–076. Council on Environmental Quality. 1975. “The Costs of Sprawl in the U.S.A.” Ekistics 239: 266–72. Crane, Randall. 2000. “The Influence of Urban Form on Travel: An Interpretive Review.” Journal of Planning Literature 15, no. 1: 3–23. Crane, Randall, and Richard Crepeau. 1998. “Does Neighborhood Design Influence Travel? A Behavioral Analysis of Travel Diary and GIS Data.” Transportation Research D 3, no. 4: 225–38. Dieleman, Frans M., Martin Dijst, and Guillaume Burghouwt. 2002. “Urban Form and Travel Behavior: Micro-level Household Attributes and Residential Context.” Urban Studies 39, no. 3: 507–27. Edwards, Jerry L. 1978. “The Effect of Land Use of Transportation Energy Consumption.” In Energy and the Community, edited by R. J. Burby and A. Flemming Bell, pp. 71–93. Cambridge, Mass.: Ballinger. [EIA] Energy Information Administration. 2001. A Look at Residential Energy Consumption in 2001. Washington: U.S. Department of Energy. [EPA] U.S. Environmental Protection Agency. 2001. Our Built and Natural Environments: A Technical Review of the Interactions between Land Use, Transportation, and Environmental Quality. Washington. Ewing, Reid, and Robert Cervero. 2001. “Travel and the Built Environment: A Synthesis.” Transportation Research Record 1780: 87–114. Gately, Dermot, and Hillard G. Huntington. 2001. “The Asymmetric Effects of Changes in Price and Income on Energy and Oil Demand.” Economic Research Reports 2001-01. New York University: Chauncey V. Starr Center for Applied Economics (January).

Spatial Development and Energy Consumption

245

Golob, Thomas F., and David Brownstone. 2005. “The Impact of Residential Density on Vehicle Usage and Energy Consumption.” Working Paper 011. University of California–Irvine, Energy Institute (February). Gomez-Ibañez, Jose A. 1991. “A Global View of Automobile Dependence: Review of ‘Cities and Automobile Dependence: An International Sourcebook’ by Peter W. G. Newman and Jeffrey R. Kenworthy.” Journal of the American Planning Association 57, no. 3: 376–79. Gordon, Peter, Ajay Kumar, and Harry W. Richardson. 1989. “Congestion, Changing Metropolitan Structure and City Size in the United States.” International Regional Science Review 12, no. 1: 45–56. Gordon, Peter, and Harry W. Richardson. 1997. “Are Compact Cities a Desirable Planning Goal?” Journal of the American Planning Association 63, no. 1: 95–106. Handy, Susan L. 1996. “Methodologies for Exploring the Link between Urban Form and Travel Behavior.” Transportation Research D 1, no. 2 (December): 151–65. Harrington, Winston, Sébastien Houde, and Elena Safirova. 2007. “A Simulation of the Effects of Transportation Demand Management Policies on Motor Vehicle Emissions.” Paper presented at the Transportation, Land Use, Planning, and Air Quality conference. Orlando, Florida, July 9–11. Accepted for publication in the proceedings of the conference, to be published by the American Society of Civil Engineers. Holden, Erling, and Ingrid T. Norland. 2005. “Three Challenges for the Compact City as a Sustainable Urban Form: Household Consumption of Energy and Transport in Eight Residential Areas in the Greater Oslo Region.” Urban Studies 42, no. 12: 2145–166. Houde, Sébastien, Elena Safirova, and Winston Harrington. 2007. “Washington START Transportation Model: Description and Documentation.” Discussion Paper 07-43. Washington: Resources for the Future. Hui, Sam C. M. 2001. “Low Energy Building Design in High Density Urban Cities.” Renewable Energy 24, no. 3: 627–40. Kenworthy, Jeffrey R. and Peter W.G. Newman. 1990. “Cities and Transport Energy: Lessons from a Global Survey.” Ekistics 34, nos. 4–5: 258–68. Keyes, Dale L. 1978. “Land Use and Energy Conservation: Is There a Link to Exploit?” In Energy and the Community, edited by Raymond J. Burby and A. Flemming Bell, pp. 101–26. Cambridge, Mass.: Ballinger. Larivière, Isabelle, and Gaetan Lafrance. 1999. “Modeling the Electricity Consumption of Cities: Effect of Urban Density.” Energy Economics 21, no. 1: 53–66. Levinson, David M. 1998. “Accessibility and the Journey to Work.” Journal of Transport Geography 6, no. 1: 11–21. Levinson, David M., and Ajay Kumar. 1997 “Density and the Journey to Work.” Growth and Change 28, no. 2: 147–72. May, A. D., M. Roberts, and P. Mason. 1992. “The Development of Transport Strategies for Edinburgh.” Proceedings of the Institution of Civil Engineers: Transport 95, no. 1 (February): 51–59. Mindali, Orit-Rotem, Adi Raveh, and Iian Salomon. 2004. “Urban Density and Energy Consumption: A New Look at Old Statistics.” Transportation Research A 38, no. 2: 143–62. Modarres, Ali. 2003. “Polycentricity and Transit Service.” Transportation Research A 37, no. 10: 841–64. Nelson, Peter, and others. 2007. “Transit in Washington, D.C.: Current Benefits and Optimal Level of Provision.” Journal of Urban Economics 62, no. 2: 231–51.

246

Elena Safirova, Sébastian Houde, and Winston Harrington

Newman, Peter W.G. and Jeffrey R. Kenworthy. 1989. “Gasoline Consumption and Cities: A Comparison of U.S. Cities with a Global Survey.” Journal of the American Planning Association 55, no. 1: 24–37. ———. 1999. “The Costs of Automobile Dependence: A Global Survey of Cities.” Transportation Research Record 1670: 17–26. Oak Ridge National Laboratory. 2007. Transportation Energy Data Book 2007. Washington: U.S. Department of Energy (http://cta.ornl.gov/data/index.shtml [November 16, 2007]). Parry, Ian W., Carolyn Fischer, and Winston Harrington. Forthcoming. “Do Market Failures Justify Tightening Corporate Average Fuel Economy (CAFE) Standards?” Energy Journal. Roberts, James S. 1978. “Energy Conservation and Land Use: Prospects and Procedures.” In Energy and the Community, edited by Raymond J. Burby and A. Flemming Bell, pp. 47–70. Cambridge, Mass.: Ballinger. Safirova, Elena, Kenneth Gillingham and Sébastien Houde. 2007. “Measuring Marginal Congestion Costs of Urban Transportation: Do Networks Matter?” Transportation Research A 41, no. 8: 734–49. Safirova, Elena, and others. 2004. “Welfare and Distributional Effects of Road Pricing Schemes for Metropolitan Washington, D.C.” In Road Pricing: Theory and Evidence, edited by Georgina Santos, pp. 179–206. Amsterdam: Elsevier Science. Safirova, Elena, and others. 2005. “Choosing Congestion Pricing Policy: Cordon Tolls vs. Link-Based Tolls.” Transportation Research Record 1932: 169–77. Safirova, Elena, and others. 2006. “Congestion Pricing: Long-Term Economic and Land Use Effects.” Discussion Paper 06-37. Washington: Resources for the Future. Schwanen, Tim, Frans M. Dieleman, and Martin Dijst. 2004. “The Impact of Metropolitan Structure on Commute Behavior in the Netherlands: A Multilevel Approach.” Growth and Change 35, no. 3: 304–33. Smart Growth Network. 1995. Getting to Smart Growth: 100 Policies for Implementation. Washington: International City/County Management Association (www.smartgrowth.org/ pdf/gettosg.pdf ). Stead, Dominic, Joe Williams, and Helena Titheridge. 2000. Land Use, Transport and People: Identifying the Connections.” In Achieving Sustainable Urban Form, edited by Katie Williams, Elizabeth Burton, and Mike Jenks, pp. 174–86. London: E & FN Spon. Steadman, Philip J. 1979. “Energy and Patterns of Land Use”. In Energy Conservation through Building Design, edited by Donald Watson, pp. 246–60. New York: McGraw Hill. Steemers, Koen. 2003. “Energy and the City: Density, Buildings and Transport.” Energy and Buildings 35, no. 1: 3–14. Steiner, Ruth. 1994. “Residential Density and Travel Patterns: Review of the Literature.” Transportation Research Record 1466: 37–43. Train, Kenneth. 1980. “A Structured Logit Model of Auto Ownership and Mode Choice.” Review of Economic Studies 47, no. 2: 357–70.

Index

AAMC (Association of American Medical Colleges), 33 Aaronson, D., 85 Abecedarian Early Childhood Intervention, 164. See also Preschool education Accountability, 117, 140 ACIR (Advisory Commission on Intergovernmental Relations), 109 Acs, Z., 45 Advisory Commission on Intergovernmental Relations (ACIR), 109 Affordable Gold program (Freddie Mac), 80, 81 African Americans. See Minority homeownership; Minority students Agency loss model, 119 Akron, University of, 44 Alabama, property tax rate limitations in, 115 Alaska, revenue and expenditure limitations in, 115 Ambrose, B., 92

American Dream Downpayment Initiative of 2003, 93 Anas, A., 234 Anderson, N., 116, 121, 124 Anderson, R., 46 Anderson, W., 213 Andrews, L., 45 Annual Housing Survey, 78 Anselin, L., 45 Antipoverty effects of homeownership, 62 An, Xudong, 92 Apgar, W., 67 Arizona, revenue and expenditure limitations in, 113, 115 Arkansas, revenue and expenditure limitations in, 115 Aron, L., 87–88 Asia, density and travel demand in, 209 Association of American Medical Colleges (AAMC), 33 Atlanta, auto travel and urban form in, 210 Australia, density and travel demand in, 209

247

248

Index

Auto travel and urban form, 210 Bain, D., 94–95 Bania, N., 46 Barker, D., 85 Bartik, T. J., 2, 3–5, 15, 17–18, 21 Belfield, C., 2, 9–10, 15, 155 Bell, M., 2, 7–9, 15, 17, 109 Belsky, E., 67, 71, 72, 73, 75, 78, 79, 82 Benchmarking and TELs, 140 Bento, A., 210 Besley-Smart model and TELs, 120–21, 123 Besley, T., 120–21 Bigelow, J., 167 Biotechnology, 22–23, 29 Blackwell, M., 39 Bland, R., 127 Boehm, T., 68–69, 75, 78, 85 Bostic, R., 78, 92 Boston: auto travel and urban form in, 210; racial integration in, 188; Route 128 development, 23–24, 43 Boyle, M., 85 Brennan, G., 119 Brookings Institution, 1, 23 Brown, T., 127 Brown v. Board of Education (2007), 181 Brownstone, D., 210–11 Brunori, D., 2, 7–9, 15, 17, 109 Buchanan, J., 119 Budget-maximizing bureaucrat model, 119 Buildings and energy consumption, 211–12, 216 Burghouwt, G., 210 Bush (George W.) administration housing policy, 60 CAFE (Corporate Average Fuel Economy) standards, 13, 232–33 Calem, P., 78 California: density and travel demand in, 211; property tax limitations in, 114, 115, 134–35; Proposition 13 in, 7, 17, 111, 112–13, 114, 134–35, 136; TELs in, 109, 128–29

Canada: density and travel demand in, 209; low-income homeownership in, 85, 93 Canner, G., 79 Card, D., 193 Carnegie Mellon University, 44 Carr, J., 78 Center for Economic Development (Carnegie Mellon University), 44 Chicago Child-Parent Centers, 159, 164 Child Care and Development Fund (CCDF), 157, 176 Children: and low-income homeownership, 82–86, 96. See also Economically integrated neighborhoods and children’s educational outcomes; Preschool education Clinton administration housing policy, 60 Cobbs, S., 39 College graduates, impact of, 41, 52–54 Colleges and universities. See Higher education Collins, M., 68, 86 Colorado, revenue and expenditure limitations in, 7, 113, 114, 127 Community Advantage Secondary Market Demonstration Program (CAP), 69–70 Community colleges, 26. See also Higher education Community Development Block Grants, 90 Community Reinvestment Act, 90 Competitive economy through industrybased strategies. See Health care services; Higher education Comprehensive tolls, 231 Congestion pricing, 231–32 Connecticut, public opinion on tax limitations in, 125, 138 Consumer education and counseling for homeownership: DHA program for, 64, 66; and initiatives for low-income homeownership, 90, 91, 93–95, 97, 98; need for, 87; pilot program for, 61 Cordes, J., 2, 7–9, 15, 109, 127 Cordon-type pricing schemes, 231 Cornia, G., 127 Corporate Average Fuel Economy (CAFE) standards, 13, 232–33

Index Cost-benefit calculations, 16 Cost of preschool education, 166, 168, 172–73 Council on Environmental Quality, 208 Cowan, S., 95 Credit counseling and low-income homeownership, 90 Crime reduction and preschool education, 162–65, 166, 169–70, 174–75, 177 Cross-cutting themes, 13–20 Cutler, D., 123–24 Davis, W., 69–70, 74 Demand effects, 33 Deng, Yongheng, 92 Denver Housing Authority home ownership program, 5, 64, 65–66 Desrochers, P., 45 De Tray, D., 134, 138 Di, Zhu, 68, 71, 82 Dielman, F., 210 Differential effects, 16–18 Dijst, M., 210 Disabled individuals and homeownership, 72 Discrimination and low-income homeownership, 87, 90, 96 District of Columbia Public School (DCPS) preschool program, 173–74 Dobos, D., 94 Downes, T., 130, 135 Down payment subsidies, 90, 92–93, 96 Duda, M., 71, 78 Dye, R., 127, 130–31, 133 Dynan, K., 79 Early Childhood Longitudinal Study, 159 Earned Income Tax Credit, 97 Eberts, R., 46 Economically integrated neighborhoods and children’s educational outcomes, 2, 9–12, 181–205; and differential effects, 17; effects of economic integration on school children, 183–87; evidence on integration of neighborhood and schools, 187–90; literature review on, 187–93;

249

policy implications, 201–02; preliminary evidence from New York City neighborhoods and schools on, 193–201, 202; relationship between integrated neighborhoods and integrated schools, 183–84; and school integration, effect of, 184–86 Economic Development Administration, 44 Education. See Economically integrated neighborhoods and children’s educational outcomes; Higher education; Preschool education; School financing and property tax limitations Education and counseling for homeownership. See Consumer education and counseling for homeownership Eitle, Tamela, 189 Ellen, I.G., 2, 9–12, 17, 181, 188 Elmendorf, D., 123–24 Energy consumption. See Spatial development and energy consumption Energy Information Administration, 215 Engelhardt, G., 92–93 Enrollment zones in school districts, 183 Environmental Protection Agency (EPA), 212 Equal Credit Opportunity Act, 90 Erickcek, G., 2, 3–5, 15, 17–18, 21 European Union (EU): buildings and energy consumption in, 211; density and travel demand in, 209 Export-base industries, 2, 23, 24–26, 32–35 Family Self-Sufficiency (FSS, DHA), 60, 61, 90, 98 Fannie Mae, 60, 90, 91–92 Federal Housing Administration (FHA), 80, 81, 91 Feins, J., 94, 95 Feldman, M., 45 Fernandez, J., 134, 138 Figlio, D., 131–32, 135 Financial literacy programs, 90 First-time homebuyers, assistance for, 12, 97–98. See also Live Near Your Work (LNYW)

250

Index

Flippen, C., 69 Florida: preschool in, 157; property tax limitations in, 113, 114; TELs in, 138 Fogarty, M., 46 Footloose businesses, 23 Foreclosures, 80, 81, 95 Foundations for Home Ownership Program (FFHO, DHA), 65–66, 70, 72, 74, 81, 82. See also Low-income homeownership Freddie Mac, 60, 80, 81, 90, 91–92, 94 Freeway tolls, 231 Fugitive literature, 33 Full disclosure laws, 115–16, 127, 140 Gabriel, S., 92 Galles, G., 128–29 Galster, G. G., 2, 5–7, 60, 85, 87–88 Game theory and TELs, 120–21, 123 Geographic mobility, 3, 14, 35, 76, 83, 167 Geolytics, 194 George Washington University Institute of Public Policy, 1 Georgia, 114, 157 Glaeser, E., 168–69 Glickman, M., 129 Golob, T., 210–11 Gonzalez-Vega, C., 94 Goodman, J., 91 Gordon, P., 209–10 Gottlieb, J., 168 Governance, effective. See Tax and expenditure limitations (TELs) Government-sponsored enterprises (GSEs), 91–92. See also specific agencies and programs Gradeck, R., 45 Grand Rapids–Wyoming and Kalamazoo–Portage metropolitan areas, 37–41 Great Depression, property tax limits during, 111–12 Green, R., 85, 86 Harkness, J., 85, 86 Harrington, W., 2, 12–13, 14, 15, 206

Hatarska, V., 94 Haurin, D., 72–73, 75–76, 78–79, 85, 88 Haurin, J., 85 Head Start, 157–58, 160, 161–62, 173, 176. See also Preschool education Health care services, 2–5, 21–59; and demand stimulus effects of expanding export base, 24–26, 33–35; and differential effects, 17–18; and double-counting of expenditures, 36; as export-base industry that stimulates demand for local goods and services, 2, 32–33; and export-base percentage, 33–35; and future research, 19; and human capital, 26–27, 41–43; and job creation and earnings effects on local human capital, 26–27; and local standards of fair wages and fair labor practices, 27–28, 47–50; multiplier effects for, 35–36; and opportunity cost of expansion, 36–37; and outcome measures, 15; reference calculation of impact of, 37–41; and research and development (R & D) spillover, 27, 43–47, 50, 51–52; technical appendix for analysis of, 52–56; types of impacts of, 24–28, 31–41; variations across metropolitan areas in, 28–31 Heckman, J., 8–9 Herbert, C., 67, 71, 73, 75, 78, 79, 93 Higher education, 2–5, 15, 21–59; college graduates, effect and impact of, 52–54; and demand stimulus effects of expanding export base, 24–26, 33–35; and differential effects, 17–18; and doublecounting of expenditures, 36; as export-base industry that stimulates demand for local goods and services, 2, 32–33; and export-base percentage, 33–35; and future research, 19; and human capital, 26–27, 41–43; and job creation and earnings effects on local human capital, 26–27; and local standards of fair wages and fair labor practices, 27–28, 47–50; multiplier effects for, 35–36; and opportunity cost of

Index expansion, 36–37; and private earnings return to local residents from increased local educational attainment, 55–56; reference calculation of impact of, 37–41; and research and development (R &D) spillover, 27, 43–47, 50, 51–52; technical appendix for analysis of, 52–56; types of impacts of, 24–28, 31–41; variations across metropolitan areas in, 28–31 High/Scope Perry preschool program, 164 Hill, E., 46 Hirad, A., 80, 94–95 Holland, D., 135, 138 Home equity loans, 82 HOME Investment Partnerships program, 90 Home Mortgage Disclosure Act (HMDA), 90 Homeownership. See Low-income homeownership HOPE IV, 11–12, 201. See also Economically integrated neighborhoods and children’s educational outcomes Houde, S., 2, 12–13, 14, 15, 206 Housing and Urban Development (HUD), Department of, 60–61, 90, 92, 94 Housing supply and low-income homeownership, 90–91 Human capital development: and health care services, 26–27, 41–43; and higher education, 26–27, 41–43. See also Preschool education Ichniowski, C., 127 Idaho, property tax limitations in, 113 Illinois: property tax limitations in, 115, 133; school financing in, 127, 130–31 Implementation mechanisms, 18–20 Import substitution, 25 Inclusionary zoning (IZ) program, 218–19, 230, 241, 243 Incremental costs, 16 Indiana: preschool in, 164; revenue and expenditure limitations in, 115 Individual Development Accounts, 90, 97

251

Information and low-income homeownership. See Consumer education and counseling for homeownership Insurance hypothesis and TELs, 116, 121, 124 Interest rates and homeownership, 77–78, 87 Investments in industry by state and local governments. See Health care services; Higher education Joyce, P. G., 115, 127, 128 Kalamazoo (Mich.), 37–41 Kanarouglou, P., 213 Karoly, L., 167 Kentucky Rural Health Works Program, 33 Kentucky, University of, 33 Kenworthy, J., 209 Keyes, D., 208 Kirlin, J., 94–95 Kumar, A., 209–10 Labor market practices, 76. See also Health care services Lang ,Yi, 68, 71 Laosirirat, P., 127 Latinos. See Minority homeownership; Minority students Lendel, I., 46 Leslie, L., 33, 36 Lester, Richard, 44 Leviathan model and TELs, 119–20, 121, 123 Levinson, D. M., 209 Listokin, D., 87 Liu, Xiadong, 68, 71 Live Near Your Work (LNYW) program, 12, 218, 229–31 Living wage laws, 27–28 Lo, C., 112 Local government investment in industry. See Health care services Local government tax and expenditure limits. See Tax and expenditure limitations (TELs)

252

Index

Local Innovation Systems Project (MIT), 44 Localism, 117–18 Locke, G., 80 Logan, J., 188 London Congestion Pricing Scheme, 231 Lotteries, 129 Louie, J., 72 Louisiana, spending limits in, 113 Low-income homeownership, 2, 5–7, 60–108; and ability to weather financial setbacks, 71–76, 96; and barriers to sustainable homeownership, 86–89, 96; children, effect on, 82–86, 96; Denver Housing Authority’s Foundations for Home Ownership Program, 5, 64, 65–66; education and counseling for, 93–95; and financial distress probability, 76–82, 97; and initiatives to sustain homeownership, 90–95, 96–97; and liberalization of mortgage terms, 91–93; policy implications for, 19, 96–98; and preschool education, 168; rationale for, 17, 61–65; and TELs, 136, 137; as wealth-building strategy, 66–71, 96 Lubell, J., 86 LUSTRE model, 219, 233, 234. See also Spatial development and energy consumption Maine, revenue and expenditure limitations in, 115 Marginal costs, 16 Marriage termination and homeownership, 76 Maryland, property tax rate limitations in, 115 Masnick, G., 82 Massachusetts: property values in, 133; Proposition 21⁄2, 124–25, 133, 136; revenue and expenditure limitations in, 113, 115, 138 Massachusetts Institute of Technology (MIT), 44 Mass transit, 236. See also Spatial development and energy consumption

Mayer, S., 192 McArdle, N., 72 McGuire, T., 127, 130–31, 133 McHenry, P., 33, 36 McMillen, D., 127 Measure 5 (Ore.), 135, 137 Median voter and taxpayer support for TELs, 122 Meredith v. Jefferson County Board of Education (2007), 181, 191 Merriman, D., 133 Metropolitan economic development policies, 22 Michigan: property tax limitations in, 111, 113, 115; spending limits in, 113 Michigan State University, 24 Middle class, growing. See Low-income homeownership Milliken v. Bradley (1974), 182 Miller, E., 85, 213 Mills, E., 70–71 Minnesota, Mortgage Foreclosure Prevention Program in, 95 Minority homeownership: and ability to weather financial setbacks, 74–75; and barriers to homeownership, 87, 88; delinquency rates of, 80; and foreclosures, 81; interest rates for, 77–78; mortgages for, 86–87; and refinancing, 63; strategies for, 97; wealth of, 67, 68–70. See also Low-income homeownership Minority students: enrollment in preschool, 157, 158, 173; and racial segregation and integration of schools, 181–82, 187–90, 191–92 MIT (Massachusetts Institute of Technology), 44 Mobility, 3, 14, 35, 76, 83, 167 Monroe, A., 91 Moreno, A., 95 Mortgage Foreclosure Prevention Program, 95 Mortgages: and barriers to homeownership, 67, 71, 86–87, 96; and delinquency and defaults, 75, 80–81, 82, 89, 94; and edu-

Index cation and counseling, 94–95; and foreclosure, 80, 81, 95; and home equity, 82; and initiatives for low-income homeownership, 90; interest rates, 77–78, 87; liberalization of terms of, 91–93; and minority homeowners, 86–87; and predatory lending, 89; and probability of financial distress, 76–82; refinancing of, 63, 79, 82, 89; subprime, 79–80, 89; and tax policy, 13, 67, 98, 232–33; volume of, 61. See also Low-income homeownership Moving to Opportunity (MTO), 192–93 Mullins, D., 115, 127, 128 MVA Consultancy, 234 National Association of State Universities and Land-Grant Colleges (NASULGC), 33 National Conference of State Legislatures, 117 National Household Education Surveys Program (NHES), 157 National Longitudinal Survey of Youth (NLSY), 74–75, 85 Nebraska, property tax rate limitations in, 115 Neighborhood Change Database (NCDB), 194 Neighborhood Reinvestment Homeownership Pilot program, 73, 74, 82 Neighborhoods for families, creating. See Economically integrated neighborhoods and children’s educational outcomes Nevada, property tax limitations in, 111 Newburger, H., 77, 81 New Jersey, revenue and expenditure limitations in, 113, 133, 134, 136–37 Newman, P., 209 Newman, S., 85–86 New Mexico, property tax limitations in, 114 New York City, economic integration and educational outcomes of students in, 193–201, 202

253

New York, property tax limitations in, 114 Nichols, J., 91 Non-export-base activities, 23 Nonresident taxpayers and TELs, 122 Norris, D., 116–17 North Carolina: community colleges in, 26; Research Triangle in, 23–24, 43 Oakley, D., 188 Ohio, property tax limitations in, 111, 115 Oklahoma: preschool in, 157, 164; property tax limitations in, 111, 114 Onder, Z., 91 Ong, P., 189 Ontario (Canada), 85 Opportunity cost of expansion, 36–37 Oregon: Measure 5, 135, 137; revenue and expenditure limitations in, 113, 115, 135; school financing and property tax limitations in, 131–32 Orzechowski, S., 67 O’Sullivan, A., 131, 134–35 Outcome measures, 14–15 Painter, G., 129 Panel Study of Income Dynamics (PSID), 68, 85 Parcel, T., 85 Parents Involved v. Seattle School District No. 1 (2007), 181, 191 Passmore, W., 79 Patents obtained by universities, 44–45 Paytas, J., 45 Peer effects in education, 184–85, 192 Pharmaceutical industries, 22–23, 29 Philadelphia 500 program, 77 Philadelphia, foreclosures in, 81 Phillips, S., 94 Pilot programs: for consumer education and counseling for homeownership, 61; Neighborhood Reinvestment Homeownership Pilot, 73, 74, 82 Pindus, N. M., 50 Pitcoff, W., 82 Policy design and implementation, 18–20

254

Index

Polymer research, 44 Poterba, J., 116, 128 Predatory lending, 63, 89, 90. See also Lowincome homeownership Pre-K Incentive Program, 173 Preschool education, 2, 8–10, 155–80; academic benefits of, 158–60; behavioral benefits of, 158, 160–61; and building human capital and competitive environment, 165–69; and case study of expanding preschool in Washington, D.C., 172–75; and costs and benefits of expanding preschool coverage, 16, 171–76; and differential effects, 17; and economic benefits of preschool, 158–65; and enrollments in preschool, 156–58; and funding for preschool programs, 175–76; and outcome measures, 15; and policy concerns in expanding preschool, 171–72; spillover benefits of, 169–71; and urban development, 19, 165–71 Preston, A., 127 Principal-agent model, 119 Property tax limitations. See Tax and expenditure limitations (TELs) Property values and TELs, 132–33 Proposition 21⁄2 (Mass.), 133, 136 Proposition 13. See California PSID (Panel Study of Income Dynamics), 68 Public hospitals, subsidies for, 36–37 Public Use Microdata Sample (PUMS), 29, 30, 48. See also Health care services; Higher education Quality neighborhoods for families. See Economically integrated neighborhoods and children’s educational outcomes Quercia, R., 69–70, 73–74, 82, 87, 95 Quigley, J., 46 Racial segregation and integration of schools, 181–82, 187–90, 191–92 Rail networks, 236 Reagan administration: and housing policy, 60; and property tax limitations, 112

Reardon, Sean, 189–90 Redfearn, C., 87 Reeder, W., 87 Refinancing of mortgages and homeownership, 79, 82, 89 Regional Economic Models Inc. (REMI), 37–41. See also Health care services; Higher education Regional Economy and Land Use (RELU), 234, 235–36, 237. See also Spatial development and energy consumption Registered Home Ownership Savings Plan, 93 Reid, C., 68, 75–76 Rent Voucher Homeownership, 98 Renuart, E., 89 Research and development (R&D) in higher education and health care services, 24, 27, 43–47, 50 Research, implications for future, 19–20. See also specific topics Research Triangle (N.C.), 23–24, 43 Resident Opportunities for Self-Sufficiency (ROSS) programs (DHA), 65 Retsinas, N., 71, 78 Richardson, C., 89 Richardson, H., 209–10 Rickles, J., 189 Rivkin, S., 189 Rohe, W., 73–74, 82 Rosenthal, S., 72–73, 75–76, 78–79, 87–88 Rothstein, J., 193 Route 128 development (Boston), 23–24, 43 Rueben, K., 116, 128, 132 Safirova, E., 2, 12–13, 14, 15, 206, 234 Sanderson, A., 33, 36 San Francisco, preschool in, 17 Santiago, A. M., 2, 5–7, 60 Schlottmann, A., 68–69, 75, 78, 85 School financing and property tax limitations: and constitutionality of property tax financing, 112; and educational outcomes, 132, 136, 137; and educational

Index spending, 130–32; and elimination of property taxes, 113; and property tax revenue, 127; state and local share of, 125, 126, 129. See also Tax and expenditure limitations (TELs) Schuetz, J., 78 Schwartz, A. E., 2, 9–12, 17, 181 Section 8 Homeownership program, 77, 80 Section 235 homeownership program (HUD), 94 Sepielli, P., 67 Sexton, R., 128–29 Sexton, T., 134–35 Shadbegian, R., 116, 126, 128, 132 Shapiro, J., 53, 168–69 Sheffrin, S., 134–35 Shocks, effect on local labor, 54–56 Siegfried, J., 33, 36 Silicon Valley, California, 23–24, 43 Single parents and homeownership, 72 602 Nonprofit Property Disposition program (HUD), 60 Slaughter, S., 33, 36 Smart, M., 120–21 Social benefits, 16 Sokolow, A.D., 125 South Dakota, revenue and expenditure limitations in, 115 Spatial development and energy consumption, 2, 12–13, 206–46; and baseline energy profile, 240–41; and buildings and energy consumption, 211–12, 216; and congestion pricing, 231–32; and energy modeling, 215, 239–41; and inclusionary zoning (IZ) program, 218–19, 230, 241, 243; and increased residential housing density, 228; and individual residential preferences, 210, 213, 216, 220, 228; and LNYW program, 218, 229–31, 241, 243; and local policies, 233; and methodology of study, 214–16; and modeled urban scenarios and policies, 14, 212–13, 216–19, 231–33, 241–43; model overview, 214–15, 234–39; and nonurban policies,

255

232–33; and outcome measures, 15; and public policy and energy consumption, 15, 212–15; and residential energy use, 215; results, 219–31; and road capacity, 216–17, 228–29; theory and empirical research on, 207–12; and transportation, 208–11, 215, 228–29; and VMT tax, 219, 220, 228, 229, 230–31, 232 State and local government investment in industry. See Health care services; Higher education State lotteries, 129 State tax and expenditure limitations (TELs) legislation. See Tax and expenditure limitations (TELs) Stegman, M., 69–70, 74 Stiefel, L., 2, 9–12, 17, 181 Stowell, J., 188 Strategic and Regional Transport (START) modeling suite, 234–35, 236–37. See also Spatial development and energy consumption Strauss, L., 94 Subprime mortgages, 79–80, 89 Survey of Consumer Finances, 62 Sustin, S., 78 Sweden, university development and local economic development in, 46 TANF (Temporary Assistance for Needy Families), 176 Tax and expenditure limitations (TELs), 2, 7–9, 109–54; and accountability, 140; administrative implementations of, 114–15; and assessment limitations by states, 114–15, 141–43; assessment of policy options on, 139; and benchmarking, 140; benefit-cost evaluation of, 140; binding and nonbinding limits, 115–16, 126–27; decisions by jurisdictions to support or reject, 123–25; and differential effects, 17; distributional effects of, 133–35, 138–39; and effect on local finances and public services, 136–37; and effect on public finances, 125–35,

256

Index

136–37, 138; and effects on taxes and spending, 126–29; empirical evidence of, 123–25, 138; and fiscal illusion, 122–23; future research implications for, 137–40; and government spending, 128–29; history of, 111–13; and Leviathan model, 119–20, 121, 123; and localism, 116–18, 136–37; motives for supporting TELs, 135–36; and outcome measures, 15; override mechanism for, 116; overrides, 139–40, 141–50; political economy of, 116–23; popular support for, 118–25, 135–36; and property tax limitations today, 113–16; and property tax rate limits by state, 115, 144–47; and property tax revenue, 127; and property values, 132–33; public opinion surveys on, 123; and public services, 130–32, 136–37; and revenue and expenditure limits by state, 115–16, 148–50; state and local limits for, 114; and taxpayer self-interest, 121–22 Tax policy: Earned Income Tax Credit, 97; for homeownership, 13, 67, 98; truth-intaxation requirements, 115–16, 127, 140. See also Tax and expenditure limitations (TELs) Temple, J.A., 124 Temporary Assistance for Needy Families (TANF), 176 Texas, TELs and property tax revenue in, 127 Thibodeau, T., 92 Thistle, P., 78 Trachtenberg School of Public Policy and Public Administration, 1 Transit-oriented development (TOD), 233 Transportation in spatial development and energy consumption. See Spatial development and energy consumption Truth-in-taxation requirements, 115–16, 127, 140 Tsen, W., 93 Turner, M. A., 1

United Kingdom, buildings and energy consumption in, 211 United States: buildings and energy consumption in, 211; density and travel demand in, 209 Universities. See Higher education Urban and Regional Policy and Its Effects conference, 1 Urban development, 207, 208, 210; preschool’s role in, 19, 165–71. See also Spatial development and energy consumption Urban Institute, 1, 194 Utah, revenue and expenditure limitations in, 113, 127 Van Order, R., 81 Varga, A., 45–46 Vehicle mile traveled (VMT) tax, 219, 220, 228, 229, 230–31, 232 Vigdor, J., 122, 124–25, 132–33, 137, 138 Virginia, property tax limitations in, 115 Wages in higher education and health care services, 27–28, 47–50 Walters, L., 127 Washington: property tax limitations in, 111; school financing and property tax limitations in, 131–32; spending limits in, 113 Washington, D.C.: preschool in, 10, 171, 172–75; spatial development and energy consumption in metropolitan, 12–13, 214–43 Washington START, 234–35, 236–37 Watcher, S., 78 Waters, E., 135, 138 Watson, Philip E., 112 Wealth acquisition and homeownership, 62–64, 66–71, 96. See also Low-income homeownership Weber, B., 135, 138 Weinberg, D., 39 West Virginia, property tax limitations in, 111

Index W.E. Upjohn Institute for Employment Research, 37 White, M., 85, 86 Wial, Howard, 1 Wilhelmsson, M., 46 Wiranowski, M., 81 Wolman, Hal, 1

257

Wyoming, property tax limitations in, 115 Yuan, Bing, 2, 7–9, 15, 17, 109 Yun, John, 189–90 Zeckhauser, R., 123–24 Zorn, P., 80, 81, 94–95