Research in Law and Economics, Volume 19 (Research in Law and Economics)

RESEARCH IN LAW AND ECONOMICS Volume 19

This Page Intentionally Left Blank

RESEARCH IN LAW AND ECONOMICS VOLUME 19

EDITED BY

RICHARD O. ZERBE Jr University of Washington Seattle, WA 98195, USA

WILLIAM KOVACIC George Mason School of Law Arlington, VA 22201, USA

2000

JAI Press An Imprint of Elsevier Science Amsterdam – New York – Oxford – Shannon – Singapore – Tokyo

ELSEVIER SCIENCE Inc. 655 Avenue of the Americas New York, NY 10010, USA © 2000 Elsevier Science Inc. All rights reserved. This work is protected under copyright by Elsevier Science, and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier Science Global Rights Department, PO Box 800, Oxford OX5 1DX, UK; phone: ( + 44) 1865 843830, fax: ( + 44) 1865 853333, e-mail: [email protected]. You may also contact Global Rights directly through Elsevier’s home page (http://www.elsevier.nl), by selecting 'Obtaining Permissions’. In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (978) 7508400, fax: (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: ( + 44) 171 631 5555; fax: ( + 44) 171 631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Tables of contents may be reproduced for internal circulation, but permission of Elsevier Science is required for external resale or distribution of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier Science Global Rights Department, at the mail, fax and e-mail addresses noted above. Notice No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made. First edition 2000 Library of Congress Cataloging in Publication Data A catalog record from the Library of Congress has been applied for. ISBN: 0-7623-0308-5 ∞ 䊊

The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed in The Netherlands.

LIST OF CONTRIBUTORS David E. Ault

Economics Department, Box 1102 – School of Business, Southern Illinois University at Edwardsville, Edwardsville, Illinois, 62026-1102, USA. Tel.: (618)650-2542; Fax: (618)650-3047

Matthew Baker

Department of Economics U-63, University of Connecticut, Storrs, CT, 06269, USA. Tel.: (860)486-5810; Fax: (860)486-4463

Sheri Eisert

The Effects of Hospital Competition, Graduate School of Public Affairs, University of Colorado-Denver, 1380 Lawrence Street, STE. 500, Denver CO, 80204, USA. Tel.: (303)556-5984; Fax: (303)556-597

Adam Karp

Department of Statistics, University of Washington, USA. E-mail: [email protected]

Steven G. Medema

Department of Economics, University of Colorado at Denver, P.O. Box 173364, Denver, CO, 80217, USA. Tel.: (303)5568511

Thomas J. Miceli

Department of Economics U-63, University of Connecticut, torrs, CT, 06269, USA. Tel.: (860)486-5810; Fax: (860)486-4463

Reed Neil Olsen

Department of Economics, Southwest Missouri State University, 901 S. National Avenue, Springfield, MO, 65804, USA. Tel.: (417)836-5516; Fax: (417)836-4236 v

vi

Gilbert L. Rutman

Economics Department, Box 1102 – School of Business, Southern Illinois University at Edwardsville, Edwardsville, Illinois, 62026-1102, USA. Tel.: (618)650-2542; Fax: (618)650-3047

Stephen J. Spurr

Department of Economics, 2115 F.A.B., Wayne State University, Detroit, Michigan, 48202, USA. Tel.: (313)577-3232; Fax: (313)577-0149; E-mail: [email protected]

Steven Vinyard

The Daniel J. Evans School of Public Affairs and School of Law, University of Washington, Box 353055, Seattle, WA, 98195, USA. Tel.: (206)616-1676

Brian N. Wasankari

The Daniel J. Evans School of Public Affairs and School of Law, University of Washington, Box 353055, Seattle, WA, 98195, USA. Tel.: (206)616-1676

Richard O. Zerbe, Jr.

The Daniel J. Evans School of Public Affairs and School of Law, University of Washington, Box 353055, Seattle, WA, 98195, USA. Tel.: (206)616-1676; E-mail: [email protected]

CONTENTS LIST OF CONTRIBUTORS

v

STATISTICAL APPROACHES TO ASSESSING CHARGES OF ENVIRONMENTAL RACISM AND CLASSISM AGAINST THE U.S. ENVIRONMENTAL AGENCY Adam Karp

1

STATUTES OF LIMITATIONS FOR ACCIDENT CASES: THEORY AND EVIDENCE Matthew Baker and Thomas J. Miceli

47

EDUCATING ALICE: LESSONS FROM THE COASE THEOREM Steven G. Medema and Richard O. Zerbe, Jr.

69

BARS TO THE PAYMENT OF PREJUDGMENT INTEREST: THE SUPREME COURT AND THE FEDERAL EMPLOYERS’ LIABILITY ACT David E. Ault and Gilbert L. Rutman AN INTEGRATED THEORY OF THE COMMON LAW Brian N. Wasankari, Richard O. Zerbe, Jr. and Steven Vinyard

113 139

THE DURATION OF PERSONAL INJURY LITIGATION Stephen J. Spurr

223

THE EFFICIENCY OF MEDICAL MALPRACTICE LAW: A NEW APPRAISAL Reed Neil Olsen

247

THE EFFECTS OF HOSPITAL COMPETITION ON NONPROFIT AND FOR-PROFIT HOSPITAL MEDICAID SHARE Sheri Eisert

275

vii

This Page Intentionally Left Blank

STATISTICAL APPROACHES TO ASSESSING CHARGES OF ENVIRONMENTAL RACISM AND CLASSISM AGAINST THE U.S. ENVIRONMENTAL PROTECTION AGENCY Adam Karp1 ABSTRACT Discrimination law has evolved from litigating or prosecuting overt, individual cases of egregious behavior solely by means of anecdotal evidence and eyewitness testimony. Statistical evidence came to bear the imprimatur of the United States Supreme Court in the Seventies as a probative means of discerning guilt or liability, and has been used to shore up patterns of prejudice at a systemic level since. Courtrooms of the Twenty-First Century have struggled to define discrimination through a quantitative lens, nonetheless relying on qualitative evidence to assist the factfinder in rendering a verdict. Some definitions carry more precision and accuracy than others. Consider the inflammatory National Law Journal’s indictment of the United States Environmental Protection Agency (‘EPA’) as an example of the latter. In 1992, the National Law Journal ran a Special Investigation of the EPA, claiming that the federal Research in Law and Economics, Volume 19, pages 1–45. Copyright © 2000 by JAI/Elsevier Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0308-5

1

2

ADAM KARP

government had fostered a racist imbalance in hazardous site cleanup and its pursuit of polluters. Kudos to the columnists for bringing environmental equity into the spotlight of public debate and for forewarning and encouraging the EPA to conduct its enforcements reflectively, in order to avoid being on the receiving end of a Title VI lawsuit. Nonetheless, the methodology used by the National Law Journal belies a total understanding of the bureaucratic structure that pursued these actions and of the notion of statistical significance. This Article confines itself to Region X’s actions between 1995 and 1999, applying linear regression and other statistical tests to determine whether biases, found using the National Law Journal’s naïve methodology, stand after due consideration of chance. The NLJ approach finds evidence of bias, but the author also conducts more complicated and appropriate analyses, such as those contemplated by the National Guidance. After issuing some provisos, the author dismisses charges of racism or classism. While the National Guidance represents a positive first step in identifying environmental justice communities, those with an above-average proportion of lower-class or non-Caucasian inhabitants, it lacks statistical sophistication and econometric depth. This Article concludes by recommending the use of normalized racial distributions, Gini coefficients, and Social Welfare Functions to the EPA and to other organizations conducting environmental justice analysis.

I. INTRODUCTION The National Law Journal (‘NLJ’) published a 1992 report citing vast penalty and waiting time differentials as clear evidence of disparate racial impact (and almost went so far as to impugn racist motives to EPA attorneys and enforcement officers).2 Unfortunately, this study provided very little other than means and sample sizes. Furthermore, there was no consideration of confidence intervals or p-values, statistical tools that capture standard error as well as mean, and thereby help to provide a more complete picture of how seemingly appalling results may fall within the bounds of chance. The stories of victimized individuals and classes needed to be told, but the fashion in which the editors published this study gave short shrift to statistical analysis and sparse, if any, presentation of computation and methods. The numbers and percentages they did manipulate, however, showed an inadequate understanding of geographical distribution of race and income throughout the United States, as well as a working knowledge of how penalty ranges vary depending on which statute is used to clobber the polluter. Nonetheless, their study raised key points. I am optimistic that this analysis, though limited, will provide some

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 3

insight into the way Region X conducts its own enforcements. Part II proceeds in the spirit of the NLJ analysis, applying their techniques to Region X’s penalty-bearing enforcement actions between 1 October 1995 and 30 June 1999. It then appeals to the National Guidance as a basis for identifying environmental justice (‘EJ’) communities and uses a variety of more advanced statistical methods to test for significant disparities (e.g. chi-square, normal approximation to binomial, linear regression, and nonparametric tests). Part III recommends more sophisticated measures to the EPA for future EJ analyses in the hope that the extra computational burden will lead to credible and powerful conclusions.

II. NLJ AND NATIONAL GUIDANCE APPROACHES A. NLJ Redux While the NLJ looked at 1091 of the 1214 national cases that the EPA concluded between 1985 and March 1991, I studied 436 of the 447 actions conducted by EPA Region X between 1 October 1995 and 30 June 1999.3 The NLJ further omitted 162 cases because they had no penalties.4 I omitted an additional 220 no-penalty cases for proposed and 230 no-penalty cases for assessed penalty analysis. Eleven cases were removed because their ZIP codes did not exist back in 1990, an important year given that all the demographic information used to perform these analyses was based on the 1990 Census. It may come as a surprise to some that the United States Postal Service routinely swaps and redefines the boundaries of its ZIP codes irrespective of the geographical grid (county, tract, block group) put in place by the Census Bureau. Though it is unlikely that the boundaries shifted during the fifteen months since the 1990 Census was taken, one might find error in the NLJ analysis for Docket listings in the five-year period preceding 1990. An identical critique could be leveled at my analyses and other ZIP code-based studies conducted in the latter half of this decade. Notwithstanding this potential for miscalculation, please consider the following EPA Region X results. They are based on the same naïve methodology used in the NLJ report – contrasting the average penalty for the least ethnically diverse quartile (1Q) with the average penalty for the most ethnically diverse quartile (4Q), with each penalty-bearing action ranked in ascending order by proportion of people of color. • For all the federal environmental laws aimed at protecting citizens from air, water and waste pollution, proposed penalties in white communities

4

ADAM KARP

($64,788) were 33% lower than in minority communities ($96,447), Regionwide. Assessed penalties were 40% lower in white communities ($40,416) than in minority communities ($67,336). • Penalties under hazardous waste laws (viz., RCRA 3000’s, 6000’s, 7000’s) at sites having the greatest white population were about 1600 percent higher than proposed penalties at sites with the greatest minority population, averaging $3,373,378 for the white areas, compared to $198,345 for minority areas. Assessed penalties had a far less striking disparity, since penalties in white communities ($176,333) were only 238% higher than in minority communities ($52,136). Penalties under RCRA 9000-series showed absolutely no bias in assessed penalties, with both white and minority communities averaging $175. No assessment of proposed penalties was conducted as only one of twenty-six cases had a proposed penalty.5 Other disparities of note are shown in Table 1. From Table 1, it appears that polluters pay less in communities of color under the Clean Air Act’s assessed penalties (32.8% higher in white communities), RCRA’s 3000-, 6000-, and 7000-series proposed (1600.8%) and assessed (238.2%) penalties, and SDWA’s proposed penalties (50.0%). Bear in mind, however, the small number of cases.6 If anything, the data show somewhat strong evidence that racially diverse communities are receiving, on average, higher penalties than white communities under virtually every statute. This conclusion, like the opposite ones reached by the NLJ, suffer from a variety of analytical defects, such as the following: (1) By aggregating cases from every State in the Region (or the nation), one does not control for the way race is distributed heterogeneously and nonuniformly between and within States. For example, a ZIP code with 10.0% people of color may fall in the top quartile of racial diversity for Iowa, yet take a position in the bottom quartile for ethnically-varied New York. (2) By collapsing over cases from every federal environmental law, one does not control for the standard range of penalties intrinsic to each law. For example, RCRA penalties are notoriously higher than for FIFRA or SDWA. (3) Merely citing a large percentage difference does not tell someone whether the disparity is statistically significant. That is, it fails to describe how the observed disparity may be accounted for purely by chance.

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 5

(4) Using means as the only descriptive statistic to contrast penalties does nothing to standardize the measures. One must take into account the variation of the penalties and sample sizes as well as the average. (5) ZIP codes provide a rough geographical scale by which to measure racial, or other, disparities. Most communities of color and impoverished areas do not occupy an entire ZIP code (but rather a census tract or block group). Numeric dilution can easily occur when a ZIP code envelops both white, middle-class and non-white, lower-class neighborhoods. (6) Finally, the NLJ report shows a lack of understanding how the enforcement chain of command operates. Targeting and enforcement decisions, while dictated in part through a series of negotiations between the Region and Headquarters, are primarily handled on an office-by-office, State-by-State, annually revisited basis, and not as some national ten-year programme. To accuse the EPA of a ‘racist imbalance’, as the NLJ does, after conflating all enforcement decisions over a six-year period from all ten regions seems to be a classic case of guilt by aggregation.7 B. Using the National Guidance to Study Region X Enforcements Given the limitations of the NLJ approach, I conducted a set of three analyses on the same data studied above, but this time using the EPA’s National Guidance, linear regression, and nonparametric tests to identify potential, statistically significant biases in how enforcement actions were performed. Three analyses follow: (1) Where did enforcement occur?, (2) Were there significant biases in penalties proposed and won?, (3) Was the EPA softer on violators in EJ communities than in non-EJ communities? Of the original 805 actions identified to the Docket database for 1 October 1995 through 30 June 1999, 350 actions were removed (e.g. COT and ONA actions, ‘early-stage’ actions such as administrative penalty complaints, and civil referrals) at the direction of Betty Wiese, EPA Region X Office of Enforcement & Compliance Administrator.8 Whether the NLJ filtered multistage actions in this fashion is unclear. By carefully scanning the CERTS, EnviroFacts, PCS, and Risk databases, as well as using Microsoft Streets Plus and the United States Postal Service ZIP code locator service online, I was able to not only correct but also to relocate virtually every Region X action to a proper, precise ZIP code (except for those that did not list an address, which amounted to eight actions, listed below).9 This brings the grand total of actions studied to 447.10 President Clinton’s Executive Order of 1994 charges all federal agencies with implementing environmental justice, defining EJ communities as those

6

ADAM KARP

that are either racially diverse or impoverished.11 One need not be both poor and heterogeneous to receive EJ protection. The Order prompted the EPA to craft a methodology that would identify, and thereby monitor and help to reduce, disproportionate health impacts endured by racially diverse and/or poor communities. In order to determine the thresholds for each factor, the National Guidance speaks of two thresholds for determining a minority population (hereinafter ‘N20’ and ‘N50’) and three thresholds for determining a lowincome population (one of which is used here, hereinafter ‘H15’ and ‘H25’).12 N20 grants a target area EJ status if it has a proportion of people of color greater than or equal to 120% of the State’s proportion; N50 raises the cut-off to 150% of the State’s proportion. H15 grants a target area EJ status if it has a proportion of households making less than $15,000 greater than or equal to the State’s proportion; H25 compares the proportions of households making less than $25,000. All four measures reference STF1A and STF3A of the 1990 Census. A complete EJ determination requires dual consideration of race and income, such as the four screens created below. The term ‘conservative’ refers to screens that will result in labeling fewer ZIP codes or block groups as EJ than ‘anticonservative’ ones. National Guidance Screens (1) (2) (3) (4)

N50/H15 (National Guidance Version I – most conservative) N20/H15 (National Guidance Version II – income conservative) N50/H25 (National Guidance Version III – race conservative) N20/H25 (National Guidance Version IV – most anticonservative)

After reducing the original 805 actions to 447 unique ones, I applied two EJ screens (N20H25, N50H15) to determine the proportion of actions that fell in EJ communities. One should also be aware that any ZIP codes that contained a majority of Alaska Native Area/American Indian (‘AIANA’) block groups were automatically deemed to be EJ communities. By using the Census Bureau’s Extract and Geocorr programs, I was able to identify AIANA block groups and, using the allocation percentages (i.e. the percentage of block group area allocated to a ZIP code), determine whether a majority of the inhabitants in the ZIP code resided in AIANA territories. Once identified, those AIANA ZIP codes were split from the non-AIANA ones. State race and income proportions were then computed based solely on the non-AIANA ZIP codes. This decision to split the territory has been endorsed by Region X’s Office for

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 7

Civil Rights and Environmental Justice and defended from a government-togovernment perspective (where the tribal lands are conceived of as sovereign). Another basis for splitting the population by AIANA classification was the concern that this sizable number of ZIP codes with people of color proportions in excess of 80% (as in Alaska, for instance) would artificially increase the State average level of diversity. Citizens inhabiting metropolitan and not reservation areas would thus be held to a higher, more conservative standard than those in other States with less sizable reservations. Consider how Alaska’s 4.06% difference (26.06% unsplit, 22.00% split) could impact the National Guidance’s N20 and N50 race thresholds. ZIP codes in predominantly metropolitan areas, would have to meet a threshold 4.9 (for N20) and 6.1 (for N50) percentage points higher than that calculated using the ‘split’ percentage. If one believes that the identification of target areas as environmental justice communities should be made with an eye toward comparing them to similarlysituated reference areas, then it seems unfair to hold, say, downtown Anchorage to a Nome or Bethel Census Area standard, with diversity proportions of 91% and 83%, respectively. See the results in Table 2, broken down by fiscal year and State, giving the lowest and highest EJ percentages determined by both screens. The N20H25 anticonservative screen tends to yield percentages higher than N50H15. Of special interest may be Table 3, which shows the proportions of ZIP codes identified as EJ using both the N20H25 and N50H15 screens. Astonishingly, between two-thirds and three-fourths of all ZIP codes are EJ. One might argue that a large portion of these EJ-identified ZIP codes was so identified because of the income, not the race, measure.13 To determine whether or not an office is underenforcing in EJ areas, one may want to use the percentages in Table 3 as floors beneath which an office should not fall. This approach assumes random and independent enforcement actions.14 In order to assess the level of EJ enforcement by particular offices, Betty Wiese determined which statutory actions belonged to which enforcing office.15 After separating the actions by law and statute, I proceeded to analyze the proportion of EJ actions by the two most active (‘primary and secondary’) offices for each State, further broken down by fiscal year. See Table 4 for details. The Office of Water’s NPDES program wins the position of most active office in the Region, with 101 actions since 1 October 1995, but its EJ percentages are modest (67% under N20H25, 66% under N50H15). The Office of Environmental Cleanup’s CERCLA program has the second highest number of actions, with 86; its EJ percentages are not terribly high (63% under N20H25, 52% under N50H15).

8

Table 1.

Table 2.

Quartile Analysis, by Statute, Regionwide

State/FY Numbers and Proportions of Actions from Reduced 447 Unique Actions

ADAM KARP

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 9

Table 3. Proportion of EJ ZIP Codes in Each State by Method

1. Analysis #1: Where do enforcements occur? One can determine whether certain offices under- or over-enforced in EJ ZIP codes by using the normal approximation to the binomial (continuity correction not made), and then taking the resulting Z score as a measure of deviation from the expected proportions (73.9% for N20H25 and 67.3% for N50H15). One computes this Z-score approximation using the following formula: z=

x ⫺ np 兹np(1 ⫺ p)

X represents the number of EJ ZIP codes actually subject to enforcement under each statute, n the total number of ZIP codes in the State or Region, and p the proportion of EJ ZIP codes within each State or Region. The product of n and p is the number of EJ ZIP codes one would expect to find subject to enforcement. The numerator of the Z-score is, therefore, the difference between the expectation and the observation. The denominator rescales the difference in terms of standard deviations from the mean. Hence, if a single standard deviation were five percent, then a 10% difference in the numerator would represent a two standard deviation departure from what one might anticipate with the greatest likelihood. The Z-score can be translated into a p-value, the probability that one would observe a proportion as or more extreme than the one observed in the sample, by chance alone. Z-scores and p-values share an inverse relationship: the larger the Z-score, the smaller the p-value, marshalling greater evidence in favor of drawing an inference of discrimination. If 73.9% of all N20H25-screen ZIP codes are EJ, then it stands to reason that a random sample drawn from this population might end up having between 65% and 82.5% EJ ZIP codes (i.e. plus or minus two standard deviations for a sample size of 100). Thus, if a particular office has an EJ proportion less than negative two standard deviations, one might wonder if such an underselection of EJ ZIP codes occurred purely by chance.

10

Table 4. EJ Activity in Primary and Secondary Offices (by FY)

ADAM KARP

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 11

Table 5. EJ Percentages by Office (Statute) for All Cases since 1 Oct. 1995

Whether to choose the one- or two-tailed p-value depends on how one wishes to test for bias in enforcement.16 If primarily concerned with underenforcement, I would suggest using one-tailed p-values.17 One paints an anti-conservative stroke when choosing the one-tailed approach. Reasonable pvalues, to make a prima facie case of discrimination, tend to be less than 0.05. In Table 5, I have highlighted p-values that fall below 0.05. It appears that the Office for Waste and Chemical Management needs to be investigated further for TSCA underenforcement in EJ communities, as well as the Office of Environmental Cleanup’s CERCLA program. There may exist potentially confounding factors that could explain why TSCA actions significantly depart from the expected proportions. More detailed (stratified) TSCA and CERCLA analysis in Table 6 attempts to isolate these confounders. Using the same p-value cut-off of 0.05, I highlighted those Z-scores that fell below –1.645. No continuity corrections were made in the normal approximations above.18 These results provide fairly convincing evidence that a high degree of bias exists in Oregon State CERCLA and Washington State TSCA actions; less evidence exists to suggest bias in Alaska State CERCLA actions. Viewed by fiscal year, Table 7 allows one to locate the origin of the disparity that affected the Region-wide CERCLA and TSCA percentages. In order to partially protect for the multiple comparison problem, I have arbitrarily dropped the p-value cut-off to 0.01 (two-tailed), or 0.005 (onetailed).19 I have also changed the statistical mechanism by which p-values are generated. The previous two stages of the analysis used the normal

12

ADAM KARP

Table 6. Additional CERCLA & TSCA Analysis by State

approximation to the binomial, but such a technique cannot be used when the product of the sample size and the expected proportion is less than 5, as is the case here.20 Therefore, the exact binomial cumulative distribution function is used to calculate one-tailed p-values. They are shown in Table 7. We now can see where the difficulty has arisen. Of the five CERCLA actions performed in Oregon for fiscal year 1997, only one action, according to N20H25, and none, according to N50H15, fell in an EJ ZIP code. The only statistically significant p-value belongs to the N50H15 screen. Of the eight actions performed under TSCA for fiscal year 1999, just one was located in an EJ ZIP code. As determined by both the conservative and anticonservative screens, this action appears to have extremely significant p-values. Five of these eight actions all occurred in 99223, a relatively affluent, not terribly diverse area of South Hill in Spokane, Washington.21 Table 7. Further TSCA Analysis by Fiscal Year for Washington State

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 13

Given this additional information, I offer three provisos22: (1) My analysis includes only nine of the twelve months in fiscal year 1999. While this fact alone should not discount the magnitude of the differential treatment found, it should also give pause, for the Office of Waste and Chemical Management may selectively enforce according to an annual agenda. OWCM enforcements in EJ neighborhoods during the final three months could restore the balance to expected levels. (2) Targeting differs from enforcing in an important way – regardless of how many inspections are done in EJ neighborhoods, if the only facilities that violate EPA regulations reside in non-EJ ZIP codes, then an apparent underenforcement in EJ areas may not indicate bias of any sort. It may lead one to bring heat on an office that should actually receive credit for sensitivity in targeting. (3) In addition to noting that violators simply may not be evenly split between EJ and non-EJ communities, one must also recognize that the number of facilities able to be monitored by each office may likewise be unequally divided. Each office may be performing quite a fair number of inspections in EJ areas, yet the potential universe of monitorable facilities has an unnaturally high percentage located in non-EJ areas.

2. Analysis #2: Revisiting penalty disparities In addition to determining where the Region performed its actions, one may also learn something from how the Region executed them. Consider, for instance, the proposed and assessed penalty levels, and the SEP amounts brokered on behalf of the nearby citizens. Take into account all Idaho actions. Imagine that there exists some ‘hypothetical Idaho’ (the population), an ever-changing alternative universe much like our own (with certain parameters), but one that admits variation to enter its processes. Picture this ‘hypothetical Idaho’ where a disagreement between the Idaho Operations Office Director of the OEC and the lead investigator in the OAQ, or a change in the winds of the political climate, could have altered the disposition and trajectory of each of the 100 Idaho enforcement actions. Choose a subset of 20 of these ‘hypothetical Idahos’, with different facilities chosen and dissimilar choices made at key points in the enforcement process. Of these 20, we choose one. This is a sample, the reality we lived in for the past four years and the sample we will use to make sense of the differences through statistics.23

14

ADAM KARP

To determine whether the monetary differences were statistically significant (i.e. they probably did not occur just by chance, but rather by some other form of human intervention), I used linear regression and, when appropriate, a nonparametric set of tests. The response variable was either proposed penalty (P), assessed penalty (A), or SEP amount (S). Race-income screens became the explanatory variables (N20 for N20H25, N50 for N50H15). The p-value, the statistical measurement with perhaps the greatest currency in all quantitative practice, tells the probability, computed under the null hypothesis of nondiscrimination, of observing a disparity between non-EJ and EJ areas of enforcement at least as extreme as the one observed in this sample. Clearly, a small p-value casts doubt on the plausibility of the null hypothesis, suggesting one reject the null in favor of an alternative hypothesis of discrimination. See Appendix C for the complete analysis. No significant disparities exist in the area of proposed or assessed penalties or SEP amounts, for any State in Region X. Those that did generate some concern, upon further analysis, turned out unremarkable when broken down by law and fiscal year – thereby averting the pitfalls of aggregation. Please be aware of certain limitations and warnings for future studies.24

3. Analysis #3: Hardball or softball? Could there exist some bias in how aggressively an office handles violators in EJ versus non-EJ ZIP codes? Might the office be ‘softer’, choosing the compliance approach instead of the penalizing ‘way of the warrior’? By counting the number of nonzero proposed penalties, nonzero assessed penalties, and SEP’s that fell within EJ or non-EJ ZIP codes, one can then run tests for independence. Such tests can determine whether the fact that a facility was located in a highly diverse or low-income ZIP code had any influence on (i.e. was ‘independent’ of) a State or office’s desire to seek a penalty or offer a SEP. Note that in the case of SEP’s, the fractions are based on the total number of actions where a penalty was proposed and not on the total number of actions. We can see from Table 8 that, proportionally speaking, penalties are proposed with greater frequency for EJ communities in Alaska and Idaho. Violators in EJ communities are assessed more often in Alaska. SEP’s tend to get introduced primarily in Idaho’s EJ ZIP codes. Using the chi-square test for independence, the only statistically significant differences involved Alaska’s proposed penalties under the N20H25 screen (p = 0.027), Idaho’s assessed penalties under the same screen (p = 0.012), Alaska’s assessed penalties under

Comparing Proportions of EJ and Non-EJ (‘ ~ ’) Actions Where Penalty proposed or Assessed or Sep created

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 15

Table 8.

16

ADAM KARP

the N50H15 screen (p = 0.037), and Idaho’s SEP’s under the N20H25 screen (p = 0.036). Oregon and Washington showed no statistically significant differences. By studying Alaska and Idaho at a fiscal year level in Table 9, one finds more evidence of an EJ-penalty proposal dependency. The more inclusive race and income screen (‘N20H25’) for 1996 presented slight to moderate evidence of proposed penalty bias (p = 0.056) but very strong assessed penalty bias (p = 0.003) in Alaska. Note that these are biases in favor of EJ communities. Most statisticians would agree that only p-values less than 0.01 are remarkable, particularly given the multiple comparison problem. The N50H15 screen, having more demanding race and income thresholds than N20H25, found Alaska’s 1996 proposed and assessed penalties revealing a moderate (p = 0.031) and extremely strong (p = 0.002) pro-EJ bias, respectively. These significant results suggest that the proportion of assessed penalties for Alaska’s fiscal year 1996 is rather dependent on EJ or non-EJ ZIP code status.25 It would seem logical initially only to worry about proposed penalty percentage disparities, since without the proposal actual assessment is unlikely. The stricter screen showed moderate evidence in favor of a proposed penalty percentage bias (p = 0.031), but even this significance test bears little fruit, since it has considered penalties proposed from every office that enforced federal laws in fiscal year 1996. Even with statistically significant assessed penalty disparities, one must be mindful that offices typically operate independently. To avoid spurious correlations, one can conduct a law-based analysis, testing for independence by statute for all four years in each State. Approached in this fashion, no test bore significant dependencies (e.g. the most ‘significant’ came from Washington’s Office of Air Quality26). One can therefore conclude that no statistically significant office biases exist for rates of penalty proposal, rates of penalty assessment, or SEP creation.

III. SUGGESTIONS FOR FUTURE ANALYSIS Both the NLJ and National Guidance approaches to exposing enforcement biases against minority or low-income communities suffer from oversimplification. While conducting the ZIP code analyses described above, I worked on a project to identify EJ communities at a much smaller, Census

Table 9. Comparison of Proportions Continued by Fiscal Year

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 17

18

ADAM KARP

Bureau-based block group level.27 The EPA should enlist these advanced techniques to better capture the racial and economic dimensions of EJ, and thereby make fairer permitting, targeting, and enforcement decisions.

A. Racial Recommendations Using 1990 Census Data downloaded from the Bureau’s useful EXTRACT program, I managed to retrieve the following fields:

P0080001: Whites P0080003: American Indian P0080005: Other P0100002: White Hispanics P0100004: Indian Hispanics P0100006: Other Hispanics

P0080002: Blacks P0080004: Asian/Pac. Islander P0100001: Total Hispanics P0100003: Black Hispanics P0100005: Asian Hispanics

I proceeded to compute proportions of people of color within each block group/ZIP code (Total Population – Whites + White Hispanics). Treating each block group/ZIP code as a binomial random variable, one can then take advantage of the less tedious normal approximation to the binomial. From this approximation one derives a Z-score and one-tailed p-value (‘greater than the mean of all block groups in the State’). Graphically analyzing the block groups and ZIP codes by plotting a histogram of the people of color proportions, one immediately sees that no histogram portrays a normal distribution. Idaho, Oregon, and Washington looked primarily lognormal. Alaska, while lognormal for the people of color proportions ranging from 0%–75%, had a bimodal distribution given the sizable swell in the 80%–100% range. At this stage, one can use a logarithmic transformation to force the race distribution from lognormal (in the case of Idaho, Oregon, and Washington) to normal. In the case of Alaska, I split the population (and then the other States) into AIANA and non-AIANA block groups. After this split, I succeeded in transforming all four States to normality. The purpose of generating Z-scores (viz., standard deviations on the standard normal curve) was to create a racial screen cut-off not confounded by an

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 19

asymmetric, highly skewed distribution. With Z-scores computed,28 one could then identify those block groups or ZIP codes that exceeded a fixed p-value (or number of standard deviations). The standard cut-offs were the following: upper 5% (1.645 standard deviations), upper 10% (1.282 standard deviations), and upper 25% (0.674 standard deviations).29 I created five race-based EJ screens (N20, N50, Z5, Z10, Z25). N20 and N50 refer to the National Guidance’s 1.2 and 1.5 times the State proportion of people of color, respectively.30 The Z’s refer to the Z-score computation, where any block group (or ZIP code) in the top 5% (far right tail of distribution) gets the EJ classification; Z10 refers to the upper 10% and Z25 the upper 25%. By splitting the populations into AIANA and non-AIANA lands, Table 10 shows a general decrease in the level of diversity in non-AIANA block groups – except for Oregon. One can explain the rise here by the fact that an overwhelming proportion of AIANA Oregon lands are populated by Caucasians, a curious fact in itself.31

B. Econometric Recommendations: P0800001: Less than $5000 P0800002: $5000-$9,999 P0800003: $10,000-$12,499 P0800004: $12,500-$14,999 P0800005: $15,000-$17,499 P0800006: $17,500-$19,999 P0800007: $20,000-$22,499 P0800008: $22,500-$24,999 P0800009: $25,000-$27,499 P0800010: $27,500-$29,999 P0800011: $30,000-$32,499 P0800012: $32,500-$34,999 P0800013: $35,000-$37,499 P0800014: $37,500-$39,999 P0800015: $40,000-$42,499 P0800016: $42,500-$44,999 P0800017: $45,000-$47,499 P0800018: $47,500-$49,999 P0800019: $50,000-$54,999 P0800020: $55,000-$59,999 P0800021: $60,000-$74,999 P0800022: $75,000-$99,999 P0800023: $100,000-$124,999 P0800024: $125,000-$149,999 P0800025: More than or equal to $150,000 P080A001: Median Household Income P081001: Aggregate Household Income Less than $150,000 P081002: Aggregate Household Income Greater than or Equal to $150,000 P114A001: Per Capita Income

20

ADAM KARP

Table 10. Block Group State Averages, Split and Unsplit

After lengthy computations, one can use these Census variables to generate a Gini coefficient of relative inequality and, from that, a Social Welfare Function of absolute equality (similar to the Generalized Lorenz Curve). A Gini near 0 implies perfect equality (where every person has an equal amount of income), whereas a Gini near 1 implies perfect inequality (where a huge gap exists between the have and have-nots). One should be cautious, however, when interpreting Ginis, as two populations with the same coefficient may nonetheless have radically different income distributions. Generally speaking, the higher the Gini coefficient, the greater the level of inequality. We risk making bad decisions, though, when we ignore the influence of per capita income and look to the Gini measure alone. I chose to adopt the Social Welfare Function (‘SWF’) for the following reason: Imagine a block group where the per capita income is $5000, but because everyone is more or less equally impoverished, the Gini coefficient of 0.05 implies widespread equality. Such a result, without further information, would lead one to overlook this block group as deserving of EJ status. On the other hand, consider a block group where the per capita income is $500,000, except that 1% of the block group pockets $5 million, while the remaining 99% make only a meager $454,545. Here the Gini coefficient would be 0.89, a clear sign of radical inequality. Does this mean that one should rush to label this affluent community EJ, but not the former? The SWF for the uniformly poor community would be 0.95*$5000, or $4750, while the SWF for the nonuniformly rich community would be 0.11*$500,000 = $55,000. Now one can see the general welfare enjoyed within each community. Clearly, the poorer community should get the EJ mark.32,33

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 21

In addition to computing block group Ginis and SWF’s, I computed the overall State Gini and State SWF by summing over all income and household categories in every block group (and ZIP code) before using the Gini formula. Deviations from the State SWF were then calculated, thereby giving each block group’s (ZIP code’s) percentage above or below the State SWF. As far as EJ identification is concerned, I would recommend that the qualifying block group have a deviation no greater than 0% (i.e. less than or equal to the State SWF). Unfortunately, my analysis did not consider regional differences in living costs and inflation. For this reason, I recommend that some regional, intra-State analysis be performed with multiple baselines tailored to a block group or ZIP code’s location within the State. One might do this by computing County SWF’s and percent deviations therefrom. The National Guidance does not speak of sophisticated econometric measures. Rather, it takes a headcount of those households within the relevant geographical area who earn less than or equal to $15,000 or $25,000. These headcounts, expressed as a percentage, do a good job of telling the number of households making a pitifully small amount of money, but they do not tell anything about the extent of immiseration of the poor. The poverty gap, though not addressed by the National Guidance, better measures the level of immiseration than the mere headcount, for it considers the amount of deviation of each household from the $15,000 or $25,000 cut-off.34 One expresses the poverty gap in dollars, since it gives the amount of money it would take to raise the average poor person up to the $15,000 (or $25,000) level. Table 11 lists the Ginis, SWF’s, Headcounts, and Poverty Gaps for each State, split and unsplit. Notice the across-the-board increase in social welfare, decrease in inequality and poverty that occurs upon cleaving off the Census-identified AIANA (or American Indian/Alaska Native Area) lands. This will no doubt result in a more conservative set of income measures. I created four income-based EJ screens (H15, H25, S15, and S). H15 and H25 stand for the National Guidance headcount measure for the $15,000 and $25,000 cut-offs, respectively. S15 refers to the SWF and identifies any block group (or ZIP code) with a deviation less than or equal to –15% as EJ; S brings the deviation level up to 0%. C. Alternative Screens Extensive calculations using 1990 block groups for all four States in Region X provide convincing evidence of agreement between the National Guidance and the author’s alternative measures for EJ identification by race and income. One

22

Table 11. Block Group State Ginis, SWF’s, Headcounts, and Poverty Gaps

ADAM KARP

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 23

may attempt to index the (anti-)conservatism of both approaches based on the following list: (1) Z5/S15 (Very Conservative Benchmark) (2) Z10/S15 (Conservative Benchmark) (3) Z25/S5 (Compare to National Guidance’s N20H15 – income conservative, race anticonservative) (4) Z15/S (Compare to National Guidance’s N50H25 – race conservative, income anticonservative) (5) Z25/S (Very Anticonservative Benchmark) For my analyses, I used N50/H15, N20/H25, Z10/S15, and Z25/S. I chose to classify any National Guidance measure using H15 as more conservative than one using H25 based on the computations at the ZIP code level. At the block group level, below one sees the EJ percentages identified using each of the four screens. Tables 12 through 14 corroborate the ‘conservative-anticonservative’ ranking above, looking at race (Table 12) and income (Table 13) measures individually, as well as jointly in a screen (Table 14). The column headings ‘**Trib’ refer to the number of AIANA block groups in each State, while ‘**Non’ consider only non-AIANA block groups. The most conservative race measure in Table 12 is consistently S15 (capturing roughly 30% of all block groups), except for Alaska. The most liberal race measure, oddly, is H25 (averaging about 57%), though it holds the position as close runner-up to Washington’s AIANA-split S (50.28% versus 50.61%). By automatically identifying the 142 Alaska and 187 Washington AIANA block groups as EJ, and then recomputing the State headcounts and SWF’s, we have redefined the degree of conservatism traditionally associated with each race measure. This explains the disparity observed when ethnic diversity percentages include or exclude AIANA lands.35 Table 13 defines Z5 as the most conservative income measure (capturing roughly 6% of all block groups, except for Alaska, where over 20 percent are identified as EJ). Still more conservative than any National Guidance measure is Z10. N50 probably presents the moderate approach. The more anticonservative measure, virtually matching N20 at every step, is Z25. One cannot overlook the fact that Alaska and AIANA Washington have much higher EJ proportions than any other State, for every measure. The decision to declare all tribal lands as EJ explains this significant increase in EJ identification.36 The most conservative race-income screen, Z10S15, rakes in about 33 percent of all block groups, and almost half in AIANA Alaska. N50H15, the more conservative of the two National Guidance screens, identifies over half of all block groups as EJ. Unexpectedly, Z25S is more conservative than N20H25

24

Table 12.

Income Screens (block groups)

ADAM KARP

Table 13. Race Screens (block groups)

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 25

26

Table 14. Either/Or Screens (block groups)

ADAM KARP

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 27

Table 15.

for every non-AIANA State. Therefore, the most anticonservative race-income appears to be N20H25.37 D. To Split or Not to Split By splitting the Washington population into (non-)reservation lands, one finds an increase in the identifications made by the race measures, a decrease made by the income measures, but a modest, collective increase when race and income are merged into a screen. The proportion of people of color dropped from 13.27% to 12.59%, a decrease of just over five percent, while the SWF increased from $8640.54 to $8810.11, an increase of roughly two percent. Table 15 indicates a four to nine percent increase in collective identifications when one uses the ‘split’ measures – a racial proportion of 12.59% and SWF of $8810.11. Generating EJ screens based only on demographics from nonAIANA lands relaxes identification. AIANA block groups, as mentioned earlier, achieve EJ status by fiat. The Alaska split caused significant changes in identification levels. The proportion of people of color fell from 26.19% to 18.18%, a drop of over 30 percent, while the SWF increased by just over three percent, or $365.15, from $10,656.68 to $11,021.83.38 E. Why We Should Be Aware of a Race-Only Screen One cannot dispute the positive correlation between the proportion of people of color and the high rate of low-income poor, but do not regard the income and race measures as exchangeable. By studying the measures independently, one discovers a large disparity between the number of Washington State actions39 performed in EJ areas according to income versus race alone. Table 16 shows the three screens with the race measure separated from its accompanying income measure (N20 from H25, Z10 from S15, Z25 from S). Notice how the paired race and income indicators vary so tremendously. The three race measures (N20, Z10, and Z25) identify only 12–22 percent of all WA cases as EJ, while the three income measures (H25, S15, and S) are between

28

Table 16.

EJ Race/Income Segregated Analysis of WA

ADAM KARP

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 29

30 and 59 percent. In all three cases, the income measure identifies nearly 150 percent more cases as EJ than its respective race measure. We find less pronounced disparities at the FY96 level. Fiscal year 1997, however, shows a strong disparity for all indices: 8–18 percent for race versus 35–65 percent for income. Consider the greater identification percentages within each measure, posting perhaps the highest deviations, overall, of any fiscal year: N20H25–400%, Z10S15–338%, Z25S–250% . Washington State demonstrates satisfactory EJ compliance based on race and income combined, as well as income alone. EJ compliance by race alone, however, appears to suffer. If based on the current, unsplit State average for people of color (13.25% per capita), then there is below-average enforcement by race for every fiscal year except 1996, using the Z10 race measure. The Z25 race measure shows above-average enforcement, while the N20 race measure shows borderline enforcement in 1997 and above-average enforcement elsewhere. Whether this suggests that the definition of EJ should be recalibrated to emphasize racial justice over economic justice (or at least provide equal consideration) remains a quite important matter for political discussion.40

IV. CONCLUSION Statistics provide insight, perhaps better than any other form of evidence, into acts of alleged discrimination against classes of people and entire neighborhoods by faceless corporate entities or governmental apparatuses. The NLJ study, while admirable in bringing environmental equity to national attention, wielded statistics clumsily in its attempt to fell the mighty EPA. A similar analysis of 1995–1999 Region X EPA data exposed striking disparities for RCRA actions, but upon closer inspection held those differences to be statistically insignificant. Only the first analysis provided evidence of bias in the geographic selection of enforcement actions, but then only in two offices and for a single fiscal year. The last two analyses did not generate any extraordinary results, except that pro-EJ bias may exist in Alaska and Idaho. For future analysis, each office should make diligent efforts to provide accurate latitude-longitude data for every facility inspected and subjected to enforcement and use Year 2000 Census Data at the earliest possible convenience. That way, a more precise block group analysis could have either accentuated the ‘significant’ or discounted the ‘insignificant’ results found from the tests conducted in these three analyses. With the help of officers within and outside Region X,41 I used National Guidance approaches to reassess Region X’s enforcement performance since

30

ADAM KARP

fiscal year 1995. Though promising, the National Guidance currently suffers from oversimplifying assumptions. By acknowledging how racial demography varies within a State, one can model this important spatial component of racial clustering when conducting EJ analysis.42 Econometric measures, such as poverty gaps and Social Welfare Functions, offer a more rigorous perspective to EJ designation and better illuminate the level of financial well-being and immiseration of the poor than the headcounts used by the National Guidance. Finally, one discovers that environmental justice protects not just people of color. In Washington State, especially, it captures a large number of lowincome households, a significant number of which are Caucasian. This questions the very definition of EJ (allowing for designation by race or income) as failing to adequately protect people of color, even by the proxy of economic status. After all, how should the EPA reconcile a conflict between two EJ communities, one qualifying primarily by race, the other by income?

NOTES 1. B.A. Honors, 1995, Gonzaga University; J.D., 1998, University of Washington School of Law; Candidate for M.S. Statistics, University of Washington Department of Statistics (2000). I would like to give special recognition to Joyce Kelly, Director of the United States Environmental Protection Agency, Region X’s Office for Civil Rights and Environmental Justice, for whom I served as a contractor during the Summer of 1999. The author is not an employee of the U.S.E.P.A. and the views expressed herein are the author’s own and do not necessarily reflect those of the U.S.E.P.A. I also thank Peter Guttorp and Galen Shorack, professors of statistics at the University of Washington, for their support. 2. Marianne Lavelle, & Marcia Coyle, UNEQUAL PROTECTION: The Racial Divide in Environmental Law, at S2, Nat’l L. J. (21 Sept. 1992). 3. Id., at S4. 4. Id. 5. This case fell in the bottom quartile for people of color (12.40%) and, despite the large proposed penalty of $146,007, no penalty was actually assessed. 6. In the RCRA 3/6/7 analysis, only the means of the four least diverse and the four most diverse community-based actions were compared. In the SDWA analysis, only two cases are being compared (the least diverse action versus the most diverse action). Though the percentages indicate some element of bias, the small sample sizes should give one pause. 7. The perils of aggregation await those who make accusations of discrimination without controlling for lurking variables, or confounders. Known to many as Simpson’s paradox, one may discover bias when the data are collapsed (or aggregated) but find quite the opposite when the data are analyzed at each level of the confounding factor

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 31 (i.e. disaggregated). Consider the example of sex bias in graduate admissions for two competitive departments, Biochemistry and English: Sex of Applicant

Department

# Admitted

# Applying

% Admitted

Male

Biochemistry

202

410

49

Male

English

45

153

29

Female

Biochemistry

75

148

51

Female

English

102

304

34

When broken down by department, females fare better than males, with admission rates of 51% versus 49% for Biochemistry and 34% versus 29% for English. Aggregated data may lead one to reach the misleading conclusion that women face discrimination campus-wide, with admission rates of 44% for men and 39% for women. Sex of Applicant

Department

# Admitted

# Applying

% Admitted

Male

Bio & Eng

247

563

44

Female

Bio & Eng

177

452

39

The confounding factor here was Department. Applicants to the English department face stiffer competition than those who apply to the Department of Biochemistry. Because more women than men applied to the department with a universally lower rate of acceptance (66.5% of the applicants to English and 26.5% of the applicants to Biochemistry were female), women had a larger proportion of low-success applications than men. One method of adjusting for this discrepancy involves disaggregating the data, as done above. 8. See Appendix A for details. 9. See Appendix B for details. 10. Quality assurance for these data was performed without error for 5% of all ZIP codes. 11. Exec. Order No. 12,898: Federal Actions to Address Environmental Justice in Minority and Low-income Populations (1994). 12. The three relative thresholds are (1) ‘below poverty threshold or two times the poverty threshold’, (2) ‘below $15,000 or $25,000’, based on Census STF3A data, or (3) qualified census tracts ‘below 60% of AMGI values (imbedded)’. Office of Environmental Justice, U. S. Environmental Protection Agency, National Guidance for Conducting Environmental Justice Analyses (Peer Review Version) 53–56 (1998). 13. See infra, Section III.E. 14. Such assumptions do not adequately model repeat violations or the nonuniform distribution of facilities throughout a State (unless one can demonstrate a high correlation between facility density and race or income distribution). 15. Legal Breakdown: CAA: Office of Air Quality, CERCLA: Office of Environmental Cleanup, CWA 402, 404: Wetlands, Office of Ecosystems, CWA 311’s: Oil Pollution Control Act, Office of Environmental Cleanup, CWA 300’s: NPDES, Office of Water, EPCRA 313: Office of Waste and Chemical Management, EPCRA (All else):

32

ADAM KARP

Office of Environmental Cleanup, FIFRA: Office of Ecosystems, RCRA 9000’s: Underground Storage Tanks, Office of Water, RCRA 3000’s, 6000’s, 7000’s: Office of Waste and Chemical Management, SDWA: Office of Water, TSCA: Office of Waste and Chemical Management. 16. A two-tailed approach considers both over- and underenforcement in EJ communities, since the Type I error (the fixed or observed alpha, or level of significance) is split between both the upper and lower tail. A one-tailed approach looks only to overenforcement or underenforcement in EJ communities (not both), since the Type I error is placed wholly in the tail of interest. 17. The United States Supreme Court struggled with the notion of statistically significant deviations of observed from expected hires of Black schoolteachers in Missouri. See Hazelwood S. D.v. United States, 433 U. S. 299, 317–18, 318 n.5 (1977) (Stevens, J., dissenting). Justice Stevens’ law clerk found a p-value of nearly 0.05 (from a Z-score of –1.71, p-value of 0.0436), relatively convincing evidence of discrimination against Blacks. Stevens did not state explicitly that his clerk found the one-tailed pvalue (0.10 for two-tailed). Conventionally, only p-values less than 0.05 raise an eyebrow, whether derived from a one- or two-tailed test. Two-tailed p-values seem more appropriate when testing the alternative hypothesis of discrimination through overrepresentation (‘reverse discrimination against Whites’) or underrepresentation (‘traditional, sociologically plausible definition of discrimination’) of Blacks. The null hypothesis that race is completely independent of hiring, or that personnel decisions are made randomly as to race, may best be tested using the one-tailed, alternative hypothesis of discrimination against Blacks. In the case of EPA enforcements, where a finding of discrimination would only be made if EJ communities received less environmental protection (through underenforcement), not vice versa, the one-tailed test seems most appropriate. For a detailed discussion of the impact of one- vs. two-tailed tests in employment discrimination, see Fienberg, The Evolving Role of Statistical Assignments as Evidence in Courts 39–40 (1989) and Finkelstein & Levin, Statistics for Lawyers 125–26 (1990). 18. Corrected, the Z-scores would be less significant, but still beyond the –1.645 threshold. 19. For more information on multiple-test problems, see Sanford Weisberg, Applied Linear Regression (2nd ed.), 116–17 (1985) and R. Miller, Simultaneous Inference (2nd ed.), (1981). 20. Douglas G. Kelly, Introduction to Probability 366 (1994). 21. None of these actions involved the same defendant. One pair of actions was filed on the same day against USA Waste of Washington and Western Refuse Company. 22. Methodological Note: Some facilities were subject to enforcement more than once, either due to recidivism, perhaps, or huge facility size (with numerous, similar violations). Repeated enforcements could skew the true percentages. 23. Galen Shorack, Professor of Statistics at the University of Washington, remarked that though we have the whole population of actions over this nearly four-year period, we can look at these t-scores and p-values as “backdrops against which to view the data.” 24. (1) We are using a rough geographic scale for making EJ identifications (ZIP code, as opposed to a more refined census tract/BNA or block group). (2) Linear regression performs best with symmetric variables. Virtually no data set for the response variable (penalty or SEP amount) was mound-shaped, or normally distributed. Even the

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 33 nonzero, penalty-only analyses (i.e. those analyses that omitted actions where no penalty was proposed or assessed) were non-normal. Taking the log of the nonzero proposed penalties largely cured this non-normality problem. (3) There may be a small percentage of ZIP codes that provide the mailing address, rather than the address of the facility subject to an enforcement or compliance action. However, actions with unknown or post office box ZIP codes were identified and replaced by alternative, valid ZIP codes. 25. Note: What we are concerned with here is not the disparity in assessed amount between actions in EJ versus non-EJ areas, but rather the disparity in assessed penalty percentage, regardless of amount. 26. Clean Air Act: N20H25 and N50H15 proposed penalty percentage bias (p = 0.047), assessed penalty percentage bias (p = 0.147). 27. The statistically advanced racial and econometric analyses that follow were born largely, but not entirely, in consultation with Marty Halper, HQ Science Advisor on Environmental Justice, and Arthur Lubin, Region V Statistician. For example, the econometric recommendations were based on independent research and communication with scholars overseas. 28. By subtracting the log-transformed mean from the log-transformed proportion (of individual ZIP code or block group) and dividing by the log-transformed standard deviation. 29. Methodological Note: Block group averages were computed by omitting block groups with no people. Individual State averages were computed by summing over all relevant race categories and then using the people of color formula described above. The ZIP code analysis used the block group averages instead of the individual State averages when producing the N20 and N50 racial screens. I chose the former over the latter because a ZIP code average captures the existence of racial pockets better than a Statewide, uniform average. 30. See Office of Environmental Justice, supra note 12 at id. 31. One may find it interesting that Oregon had the highest proportion (27.4% under N20H25 and 31.5% under N50H15) of AIANA block groups that failed to meet either National Guidance standard for being EJ. Proportion of AIANA Block Groups Failing to Meet EJ Cut-Offs, by Screen State

N20H25

N50H15

AK

10.6%

9.9%

ID

21.4%

26.8%

OR

27.4%

31.5%

WA

25.7%

32.1%

32. Computational Note: The SWF had the following formula, as recommended by Goetz Kluge: SWF = Per Capita Income *(1 – Gini). For a defense to using econometric measures other than headcount for inequality and poverty, see J. J. Thompson, A Tool for Measuring Income Inequality, 51 Nieman Reports 42 (1997) (where I followed her spreadsheet recommendations for computing the Gini coefficient); Goetz Kluge’s

34

ADAM KARP

Inequality Measures, http://ourworld.compuserve.com/homepages/SMIPP/frmentro.htm (15 July 1999); Stephen Jenkins, The Measurement of Income Inequality, in Lars Osberg, Economic Inequality in the United States 3–38 (1984); A. K. Sen, On Economic Inequality (1973); D. L. Blackwood & R. G. Lynch, The Measurement of Inequality and Poverty: A Policy Maker’s Guide to the Literature, 22 World Dev. 567 (1994). It closely resembles Shorrocks’ relative income measure, the Generalized Lorenz Curve (GL(y,p) = ␮L(y,p), where L = Lorenz Curve and ␮ = mean income of distribution y). See Blackwood at 575. 33. Methodological Note: Ideally, one should compute a Gini and SWF from data in the same units. One may notice that my SWF was computed by using a per household Gini but a per capita income. I would not expect huge differences in the resulting SWF between household and capita, but I thought it important to clarify my assumptions and methods. 34. See Blackwood, supra note 28, at 569–70. 35. Quality assurance for these data was performed without error for 10% of all block groups. 36. Quality assurance for these data was performed without error for 10% of all block groups. 37. The union of the race and income screens was not computed for Idaho and Oregon, due to time constraints. Quality assurance for these data was performed without error for 10% of all block groups. 38. No pre-/post-split information is available for Alaska’s merged screens because the pre-split racial distribution was bimodal and incapable of being transformed to normality; this prevented me from computing Z-scores and rendering Z10S15 and Z25S results. 39. Not based on split into AIANA and non-AIANA lands. 40. Note that requiring a block group to clear both race and income hurdles will significantly reduce the proportion of neighborhoods that may claim EJ protection. 41. Region X support came from Joyce Kelly, Director of the Office for Civil Rights and Environmental Justice; Betty Wiese, Office of Enforcement and Compliance; and Ray Peterson and Jim Hileman, Office of Environmental Assessment. Outside support came from Marty Halper, HQ Science Advisor and Project Lead on National Guidance, and Arthur Lubin, Region V Statistician. 42. Credit for the Z-score race analysis must go to Arthur Lubin and Marty Halper, who have suggested a more complete consideration of racial distribution.

Appendix A REMOVAL RULES (1) OMIT the following types of actions: COT (Case Other) and ONA (2) COUNT ONCE if BOTH OCCUR (drop first, ‘early-stage’ action): APC (Admin. Penalty Complaint) & APO (Admin. Penalty Order) CVR (Civil Jud. Ref.) & CDE (Consent Decree Entered) (3) COUNT ONCE if BOTH OCCUR (same Docket ID): ACO/FFA (drop FFA) APO/FC (drop FC) ACO/SCR (drop SCR) ACO/FC (drop FC) APO/ACO/FC (drop ACO and FC) ACO/APO (drop ACO) ACO/APC (drop ACO) (4) INCLUDE: SCR (Superfund Cost Recovery) FC (Field Citation) FFA (Federal Facility Compliance Agreement) (5) Unspecified statutes clarified by Betty Wiese (i.e. actions where the statutory section was unclear from the Docket report): AK96 CWA (BNK) ← CWA 404 AK97 CWA (CIV) ← CWA 300’s AK97 CWA (PRN) ← CWA 300’s AK97 RCRA (BNK) ← RCRA 9000 AK98 CWA (CIV) ← CWA 300’s AK98 RCRA (CIV) ← RCRA 3/6/7 AK99 CWA (CIV) ← CWA 311 AK99 CWA (COL) ← CWA 300’s ID96 RCRA (CIV) ← RCRA 3/6/7 ID98 CWA (CIV) ← CWA 300’s ID98 RCRA (CIV) ← RCRA 9000

36

ADAM KARP

ID99 CWA (CIV) ← CWA 300’s OR98 CWA (CIV) ← CWA 404 WA96 CWA (BNK) ← CWA 404 WA97 EPCRA (COL) ← EPCRA 313 WA97 RCRA (BNK) ← RCRA 3/6/7 WA99 EPCRA (COL) ← EPCRA 313 (6) If Same Docket ID-Same Statutory Section, then favor SCR’s and CDE’s. If Same Docket ID-Different Laws, then choose one with penalty (e.g. Bunker Hill -give credit to TSCA because of penalty; dropped CERCLA action).

Appendix B LOG OF ZIP CODE RELOCATIONS AND REMOVALS Actions Moved from One State to Another: Arrow Transp. Company (10–1998–0019/ WA0002189157) moved from OR to WA: 98421. Bainbridge Island Park MOVED from WA to AK96 (10–1996–0026) SDWA 1414G2 ACO. Hatley Construction Co. (10–1997–0061/ ID0001765080) moved from ID to OR: 97868. Neil Allen MOVED from WA to ID97 (10–1997–0134) TSCA 207 ACO. Nerco Delamar Co. (10–1997–0081/ ID0000227462) moved from ID to OR: 97910. Northland Fisheries MOVED from WA to AK98 (10–1997–0117) CWA 309G2 APO. Pacific Hide and Fur (10–1993–0234)/(IDIDEA223349) moved from ID to WA because ZIP 98270. Pacific Hide and Fur (10–1998–0157/ IDIDEA223474) moved from ID to WA: 98270. Southern Pacific Transportation (10–1993–0300/CAD006913206) moved from OR to OUT (CA): 94105. Actions Removed: Alaska Pulp Corp. REMOVED because invalid zip code (0) – (10–1994–0176/AKIDEA223429). AK97. Anthony Alosi (10–1999–0153/ IDIDEA223620) REMOVED from ID: 88510 (?). NOTE: We have another ID Alosi on same date. Arthur Leeson REMOVED because Nevada Zip code (89415) (10–1993–0302) CAA COL CDE. Blue Mountain Growers, Inc. REMOVED because zip was 0 (10–1996–0001). OR96. Pacific Hide and Fur (three actions) REMOVED because two zips were 99999 and one was 0 (10–1992–0308/10–1996–0031/10–1995–0069). All ID96. Pacific Hide and Fur Co. Depot REMOVED (10–1998–0132/ IDD098812878) because of invalid zip code (99999) AK99.

38

ADAM KARP

Actions with Post Office ZIP or Unknown ZIP, moved and relocated: AK: 99500 Marathon-Spurr Platform – moving to (99611) because of Kenai address. AK: 99510 Arco Kuparuk CPF–3 – moving to (99734) because of Prudhoe Bay address. AK: 99510 Bainbridge Industrial Park – moving to (99611) because of latlong. AK: 99510 Shell Western – East Foreland, Platform A, and Platform C – all three actions moved to (99611) because of Kenai lat-longs. AK: 99510 Unocal Trading Bay – moving to (99611) because it had same post office ZIP as the Kenai actions. AK: 99512 Alyeska Pipeline Service Co. (two actions) moving to (BarrowPt. Hope Census) because of Umiat reference ( ~ EJ presumption). AK: 99519 Dignon Co. Inc. – moving to (99829) because of Hoonah, AK reference. AK: 99619 USCG-Kodiak – moving to (99615) because of Kodiak address. AK: 99652 Charles Godin – moving to (99726) because of Big Lake reference. AK: 99697 USCG-Kodiak – moving to (99615) because of Kodiak address. AK: 99699 River Terrace PCE Site – moving to (99669) because of street address in Soldotna. AK: 99710 Steven Olsen – moving to (99730) because of lat-long on Steese Hwy. AK: 99738 Alaska Dept. of Natural Resources – moving to (99780) because of address (Spruce St. is in Tok, AK). AK: 99811 Trident Seafoods Corp. – moving to (99553) because of Akutan reference. AK: 99825 Elfin Cove Corp. – moving to (99829) because of Otter Way address. AK: 99830 SOS Value Mart Inc. – moving to (99833) because of Keku Rd. address. ID: 83205 FMC – moving to (83201) because of FMC action also off HWY 30. ID: 83653 (DNE) Armour and Co. – moving to (83651) because of Armour’s (in Enviro database) at this zip code. OR: 97200 Asbestos Services International (10–1997–0141) – moving to (97225) as found in Microsoft Streets Plus – 97200 ZIP code does not exist in Census90. WA: 98124 Boeing Company (P. O. Box) – moving to (98108) because of nearby address (9725 E. Marginal Way S versus 9797)

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 39

WA: 98227 Oeser Co. – moving to (98225) because of Oeser action also in 98225. WA: 98282 Annema – moving to (98284) because of Sedro Woolley address. Checked with USPS. WA: 98282 Dettling Dairy Farm – moving to (98292) using Microsoft Streets Plus. WA: 98301 Marks – moving to (99301) because of probable typo. Checked with USPS. WA: 98353 Manchester Lab (Manchester) – moving to Port Orchard (98366) because of similar EPA Manchester Lab action at this new zip code. WA: 98357 Makah Tank Farm (Neah Bay) – moved to (98381), less than 15 miles from the Sekiu Airport and Sekiu, WA. Used Microsoft Streets Plus. WA: 98939 Simplot Soilbuilders Moxee Cit (Moxee City) – 7528 Postma Rd. – moving to (98936) using Microsoft Streets Plus. WA: 99220 WWP – moving to (99202) because of probable typo. Compare: E. 3100 Mission Ave. is in 99202.

This Page Intentionally Left Blank

Appendix C Nonzero Penalty Analysis for Region, by State, Fiscal Year, and Law

42

ADAM KARP

ALASKA STATE

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 43

IDAHO STATE

44

ADAM KARP

OREGON STATE

Statistical Approaches to Assessing Charges of Environmental Racism and Classism 45

WASHINGTON STATE

This Page Intentionally Left Blank

STATUTES OF LIMITATIONS FOR ACCIDENT CASES: THEORY AND EVIDENCE Matthew Baker and Thomas J. Miceli ABSTRACT This paper provides a theoretical and empirical analysis of statutes of limitations for accident cases. The theoretical model formalizes the tradeoff underlying a finite statute of limitations: while a shorter statute limits injurers’ exposure to liability, thereby curbing incentives for care, it also limits costly litigation associated with legal error. The model yields several comparative static results that are tested using cross-state variation in statutes of limitations for personal injury cases and accidental damages to property. In order to minimize the impact of historical inertia in the statutes, we used census data from 1910 and statute lengths from 1916. This represents the earliest period when there were both adequate census data to construct the explanatory variables, and a convenient survey of the statutes for the forty-seven states. Despite difficulties in constructing the data, the empirical model performs reasonably well in explaining variation in the statute lengths.

I. INTRODUCTION This paper provides a theoretical and empirical analysis of statutes of limitations for accidental damage cases. From an economic perspective, the Research in Law and Economics, Volume 19, pages 47–67. Copyright © 2000 by JAI/Elsevier Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0308-5

47

48

MATTHEW BAKER & THOMAS J. MICELI

trade-off underlying a statute of limitations in general is well-known.1 On the one hand, a statute limits injurers’ exposure to liability, thereby reducing incentives for care, but on the other hand, it reduces legal error resulting from stale evidence and saves on litigation costs. The optimal period of limitation balances these two effects. The theoretical contribution of this paper is to formalize this trade-off in the context of a simple accident model with legal error and litigation costs.2 The benefit from developing a formal model is that it yields specific comparative static results that can form the basis for empirical analysis. To our knowledge, the only existing empirical analysis of statutes of limitations is by Netter, Hersch & Manson (1986), in the context of claims for title under adverse possession law. As in their analysis, we examine the determinants of cross-sectional variation in statute lengths across states, but we apply the analysis to statutes of limitations for property damage cases and personal injury cases. The empirical model performs reasonably well given that data and practical limitations prevented us from keying the analysis to the years when individual statutes were passed. In general, the model performed better for the property damage statutes than for the personal injury statutes. This may reflect greater ease in measuring property losses in economic terms. This paper is organized as follows. Section II develops the basic accident model and shows that, in the absence of legal error, there is no basis for setting a finite statute length. Section III, therefore, introduces legal error and derives the optimal statute length based on the trade-off described above. It goes on to derive several comparative static results that form the basis for the empirical analysis. Section IV describes the data and presents the results of the empirical tests of the model. Finally, Section V concludes.

II. THE BASIC MODEL To examine the impact of a statute of limitations, it is necessary to assume that not all victims file suit immediately following an injury. One reason for delay by victims in filing suit may be that the injury is not immediately evident. Another is that the injurer might avoid immediate detection. (Although we will not explicitly model the latter factor, we will take it into account in specifying the empirical model in Section IV.) For purposes of the model, we abstract from the specific reason for delay and simply treat the length of time after an accident that a plaintiff files suit, denoted t, as a random variable with distribution function G(t), where G⬘ ⬅ g > 0, G(0) = 0, and G( ∞ ) = 1. Thus, under a statute of limitations of length L, a fraction G(L) < 1 of all possible suits will have been filed by time L.

Statutes of Limitations for Accident Cases: Theory and Evidence

49

Consider first the impact of a statute of limitations for accident cases in a model where litigation is costly, but courts do not make errors in assessing negligence. The model is one of unilateral care in which the injurer makes a dichotomous choice: care or no care. The injurer’s cost of care ( = the social cost) is c. The probability of an accident is pc if the injurer takes care and pn if he does not, where pn > pc. The victim’s damage in the event of an accident is D, which we take to be the same for all victims. Finally, the cost of litigation is ci and cv for injurers and victims, respectively. We assume that the magnitude of damages is large enough to make suits against negligent injurers profitable; that is, D > cv. This assumption both simplifies the model and focuses on the statute of limitations as the sole deterrent of lawsuits. A more general model, with a continuum of damage levels (some of which are less than cv) and with variation in the cost of care across injurers, would be more realistic but would not yield qualitatively different conclusions.3 The injurer’s actual care choice under a negligence rule with no errors and a statute of limitations of length L is determined as follows. If we assume that the victim observes the injurer’s care choice, then the injurer can avoid all lawsuits by taking care at cost c. In contrast, by not taking care, the injurer expects to face liability for all accident cases prior to L. Thus, his expected damages plus litigation costs are pnG(L)(D + ci).

(1)

The condition for the injurer to take care is therefore given by c < pnG(L)(D + ci).

(2)

Since the right-hand side is increasing in L, incentives for care are stronger the longer is the statute of limitations, all else equal. Condition (2) is shown graphically in Fig. 1(a), where L* > 0 is the minimum statute of limitations that induces the injurer to take care. Now consider the social condition for care. If the injurer takes care, no suits are filed, but victims still incur expected damages of pcD. Social costs in this case are therefore c + pcD.

(3)

If the injurer does not take care, expected damages plus litigation costs are pn[D + G(L)(ci + cv)].

(4)

The condition for care to be efficient is thus c < (pn ⫺ pc)D + pnG(L)(ci + cv).

(5)

50

MATTHEW BAKER & THOMAS J. MICELI

Fig. 1.

Statutes of Limitations for Accident Cases: Theory and Evidence

51

In discrete care models without litigation costs, the Hand formula for care to be efficient is4 c < (pn ⫺ pc)D.

(5⬘)

Clearly, this condition implies condition (5). The extra term on the right-hand side of (5) represents the litigation costs that are incurred when the injurer does not take care. Since there are no trials when the injurer takes care in this version of the model, these costs simply amplify the social desirability of care. Figure 1(b) is drawn under the assumption that condition (5⬘) (and therefore (5)) holds. Thus, care is socially desirable for all L. This is seen by the fact that social costs with care in (3) are everywhere below social costs without care in (4). The darkened segments of the curves show the relevant ranges given that injurers only take care when L ≥ L*, as shown in Fig. 1(a). (Note that the positive slope of the upper curve in Fig. 1(b) is due to higher litigation costs as the statute of limitations is lengthened.) It follows that social costs are minimized by setting L ≥ L*. That is, the statute of limitations should be long enough to give injurers an incentive to take care. Note, however, that the model suggests no reason to impose any upper bound on L. That is, once L* is exceeded and injurers take care, a further increase in L does not raise social costs. (This is shown by the darkened segment to the right of L* in Fig. 1(b).) The reason is that no suits are filed by victims when injurers take care since, under a perfectly functioning negligence rule, they will collect no damages. Thus, a shorter statute of limitations in the current model can only have the detrimental effect of reducing incentives for care by truncating damages for the injurer (given the assumption that care is efficient). But there is no offsetting cost for lengthening the statute. The result that all injurers take care and no suits are filed as long as L ≥ L* is, of course, unrealistic, and is an important weakness of economic models of negligence. One way to overcome this result is to suppose that courts make errors in evaluating injurer compliance with a negligence standard.5 And since justifications for a finite statute of limitations are typically based on arguments about legal error, it is appropriate to extend the model in this direction.

III. INTRODUCING LEGAL ERROR Suppose that, in determining whether an injurer violated the due standard of care under a negligence rule, courts make two kinds of errors. The first occurs when a court finds an injurer not liable when he in fact took less than due care. Let q1 be the probability of this ‘type 1’ error. The second type of error occurs

52

MATTHEW BAKER & THOMAS J. MICELI

when a court finds an injurer liable when he in fact took due care. Let q2 be the probability of this ‘type 2’ error. We assume that these probabilities are constants and, in particular, that they are independent of the time lapse t between the occurrence of the accident and the trial. This simplifies the analysis greatly and is sufficient to establish the key results.6 We further assume that 1 ⫺ q1 > q2, which implies that the court, though prone to error, is right more often than it is wrong in evaluating an injurer’s actions.7 This is a sensible assumption since if it does not hold, trials would be unsuitable procedures for resolving negligence claims. The possibility of legal error changes the incentives of victims to file suit once an accident occurs. First, if the injurer takes care, there is now a probability q2 that he will be found liable. Thus, the victim’s expected return from a suit is q2D, which we assume is large enough to make a suit profitable (q2D > cv). Again, this simplifies the model without affecting the basic results. (More generally, we need only assume that some suits are profitable as a result of type 2 errors.) Given that q2 is independent of t, the injurer’s expected cost of taking care is now c + pcG(L)(q2D + ci).

(6)

If the injurer does not take care, the victim’s expected return from a suit is (1 ⫺ q1)D. Note that the above assumptions, that q2D > cv and 1 ⫺ q1 > q2, imply that lawsuits are profitable in this case as well (i.e. (1 ⫺ q1)D > cv). The expected costs of a negligent injurer are therefore8 pnG(L)[(1 ⫺ q1)D + ci].

(7)

If (6) is less than (7), the injurer will choose to take care. After rearranging, this condition can be written c < G(L){[pn(1 ⫺ q1) ⫺ pcq2]D + (pn ⫺ pc)ci}.

(8) 9

where the right-hand side is positive given pn > pc and 1 ⫺ q1 > q2. This condition is graphed in Fig. 2(a). As in Fig. 1(a), L* represents the shortest statute of limitations that just induces the injurer to take care.10 Now consider social costs in the presence of legal error. If the injurer takes care, there will be some suits filed, in contrast to the case with no error. In a sense, these suits are "frivolous’ since the defendant is not negligent; however, plaintiffs do expect to recover at trial with a positive probability q2 (Hylton, 1990b).11 The existence of these suits will play an important role in the determination of the optimal statute of limitations. Social costs in this case are c + pcD + G(L)pc(ci + cv). When the injurer does not take care, social costs are given by

(9)

Statutes of Limitations for Accident Cases: Theory and Evidence

Fig. 2.

53

54

MATTHEW BAKER & THOMAS J. MICELI

pnD + G(L)pn(ci + cv).

(10)

The condition for care to be socially desirable is thus c < (pn ⫺ pc)D + G(L)(pn ⫺ pc)(ci + cv).

(11)

Given (5⬘), condition (11) holds for all L. This is shown in Fig. 2(b), where the darkened segments again show the relevant ranges according to the injurer’s actual care choices. There is, however, an important difference between social costs as shown in Figs 1(b) and 2(b). Note that in Fig. 1(b), the ‘no care’ cost curve is upward sloping in L, but the ‘care’ curve is flat. This reflects the fact that there are no lawsuits in the latter case. In contrast, in Fig. 2(b), both curves are upward sloping in L since there are some suits even when care is taken. Costs are increasing in L even when the injurer takes care because the longer statute of limitations allows more ‘frivolous’ suits to be filed. The impact of this difference is that it is now socially beneficial to put an upper bound on L in order to limit the costs of litigation, even when care is taken. This can be seen by the fact that the cost expression in (9) is increasing in L, whereas the expression in (3) is independent of L. By setting L at the lowest possible level that induces care–namely L*–the number of ‘frivolous’ suits, and hence litigation costs, are minimized. The preceding suggests that the optimal statute of limitations under a negligence rule is L*, but it is possible that social costs are still lower if injurers are subject to no liability, which is, in effect, a statute of limitations of zero. The reason that this case may be desirable is that setting L = 0 would eliminate frivolous lawsuits – and hence litigation costs – altogether. This benefit, however, must be weighed against the fact that when injurers are never liable, they have a reduced incentive to take care. Combining these two factors yields social costs of pnD, which is (10) with L = 0. A statute of limitations of L* is optimal if, and only if, pnD is greater than expression (9) evaluated at L*, or c + pcD + G(L*)pc(ci + cv) < pnD,

(12)

where L* solves (8) written as an equality. This comparison is shown by the leftmost points on the darkened segments in Fig. 2(b). If (12) holds, we can derive comparative statics by writing (8) as an equality and by differentiating with respect to L* and the parameters of the model. This yields ⭸L*/⭸c > 0.

(13)

⭸L*/⭸ci < 0

(14)

⭸L*/⭸D < 0

(15)

Statutes of Limitations for Accident Cases: Theory and Evidence

55

⭸L*/⭸q1 > 0

(16)

⭸L*/⭸q2 > 0.

(17)

According to (13), as the injurer’s cost of care increases, the statute of limitations should be longer. This is true because the injurer needs to face greater expected liability, all else equal, in order to be willing to take care. By (14), higher litigation costs for the injurer (ci) decrease the statute of limitations since higher trial costs increase deterrence, all else equal. According to (15), a higher average damage-level lowers the statute of limitations because the injurer’s expected liability is increased, thereby increasing deterrence. Finally, (16) and (17) imply that an increase in the probability of either type of error results in a higher statute of limitations because legal errors of either type reduce deterrence, all else equal.

IV. EMPIRICAL ANALYSIS In this section we use the model to explain differences across the United States in statutes of limitations for both personal injuries and property damages. Our sample includes the length of statutes of limitations in 47 states for the year 1916, and uses data from the 1910 General Census, the 1909 Census of Manufacturing, and the 1904 Census of Crime. Before discussing the results of the empirical analysis, we note data problems and problems we encountered in specifying the empirical model. A. Data and Model Specification The biggest data problem is the long history of statutes of limitations in the United States. To the present day, state statutes of limitations bear signs of their origins in English law. Statutes of limitations for private actions, beyond simple actio personalis moritor cum persona (‘the tort dies with the person’), first appeared in English law in the early fifteenth-century in 21 James I, c. 16, under the title ‘An Act for Limitations of Actions, and for Avoiding the Suits of Law’. 21 James I, c. 16 specified periods of six, four, and two years for property damage, personal damage, and slander, respectively.12 The form of these statutes was generally adopted by the colonies, and many states still retain the 6-year statute of limitations for property damage.13 To capture the impact of economic factors on the statutes, the ideal approach is to use state data at or near the time the statutes were adopted or developed. This eliminates historical inertia resulting from the cost of altering statutes once they are enacted. In their analysis of the economic determinants of

56

MATTHEW BAKER & THOMAS J. MICELI

adverse possession statutes, for example, Netter, Hersch & Manson (1986) used data from the year of statehood. For the purposes of our model, there is little relevant data available for the majority of states in the year of admission. In any case, dates of admission to the union cover a long period of time, inevitably raising questions about the comparability of the data. A possible solution to this problem is to restrict the analysis to a specific period of time as in Netter, Hersch & Manson (1986), but this strategy significantly reduces the number of observations. To maximize the sample size and minimize the impact of history, we therefore relied on the 1910 census for data on property values, demographics, crime rates and imprisonment; and on data on manufacturing statistics from 1909, compiled by Dodd (1993). This time frame was a natural choice for our analysis because more complete census data is available, and, because an accurate snapshot of the statutes of limitations at this time could be developed from Wood, Gould & Moore (1916). All the states but Alaska and Hawaii had entered the Union by this time, and many had just entered the union in the preceding decades.14 Even though all the statutes in the year 1916 are summarized in Wood, Gould & Moore (1916), there were still difficulties in pinning down the exact length of the statutes of limitations in some states.15 Often the statutes did not clearly indicate if they pertained to personal injury or property damage. For example, many states grouped in the same statute actions for personal injury with those for intentional injuries like libel, slander, assault, battery, and false imprisonment. Other states omitted explicit mention of accidental injuries from this list. A similar problem arose in the context of damages to property, which were often grouped with actions for intentional damage or detention of goods and chattel. Finally, a couple of states identified statute lengths for specific actions, and then noted that all other actions fell under a default statute length. In compiling our data, we used our best judgment in determining the relevant statute length. We omitted from the sample states for which it was impossible to determine a length, and we classified other states as ‘questionable’. Table 1 presents the statute lengths for personal injury and property damages for those states included in the sample. Observations deemed to be questionable are marked with an asterisk. We present results for personal injury and property damage statutes using two samples: one including all observations and one excluding questionable observations. As for the explanatory variables, we constructed several proxies from the census data to capture some of the features of our theoretical model. Summary statistics of these variables are contained in Table 2. To proxy the likelihood of legal error, we constructed two variables: the ratio of the number of judges,

57

Statutes of Limitations for Accident Cases: Theory and Evidence

justices, and lawyers in 1910 to the number of people imprisoned for crimes against property in 1904 (LJJPROP); and the ratio of the number of judges, justices, and lawyers in 1910 to the number of people imprisoned for personal injury crimes in 1904 (LJJPERS). These variables are designed to proxy the volume of cases faced by the average justice, judge and lawyer. As that volume gets smaller, fewer legal errors should occur. And since the model predicts that fewer errors should lead to a shorter statute of limitations, both variables should have negative signs. (Note that LJJPROP is only included in the property regressions, and LJJPERS is only included in the personal injury regressions.) We employed several variables to proxy the costs of litigation, the costs of precaution, and the magnitude of damages. To proxy the magnitude of Table 1. 1916 Statutes of Limitations for Property Damages and Personal Injury State

Alabama Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Mass. Michigan Minnesota Mississippi Missouri Montana

Personal Injury

Property Damage

2 2 *1 1 1 *3 1 2 *2 2 2 2 2 1 1 1 2 1 *2 2 2 1 2 *2

*6 2 3 3 6 – 3 3 *4 3 5 6 5 2 5 1 6 3 6 6 6 6 5 2

Source: Wood, Gould & Moore (1916).

State

Nebraska Nevada N. Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming

Personal Injury

Property Damage

1 2 2 2 – 3 1 2 1 1 2 2 2 2 2 1 2 1 3 *5 2 *5 *6 1

4 3 6 6 4 3 3 6 4 2 6 6 *4 6 6 3 2 3 6 5 3 5 6 4

58

MATTHEW BAKER & THOMAS J. MICELI

damages, we used the total value of personal property per capita in the state in the year 1912. This variable is referred to as PPCAP in the regressions. Our theory predicts that the sign of this variable should be negative; a higher level of damages should lead to a shorter statute of limitations, other things equal. To proxy the costs of precaution for injurers we used the average factory-wage rate, computed by dividing total wage payments in a state by the number of factory workers in that state. This variable is referred to as WAGE in the regressions below, and it should have a positive sign. Note, however, that WAGE is also likely a proxy for the litigation costs of injurers and the lost earnings of victims in personal injury cases. Since both of these variables have a negative impact on the optimal statute length (by (14) and (15) respectively), there is some ambiguity in the predicted sign of WAGE. As another measure of litigation costs, we employed the rate of illiteracy, ILLITRT. This variable, which measures the number of illiterate per hundred, is intended to reflect the likelihood that an individual is uninformed about the legal system and, thus, may capture higher costs of entering the system. In this respect, the predicted sign is negative. On the other hand, if injurers are less informed about the law, or if victims are less aware of their right to sue for Table 2. Summary Statistics Variable Name

N

Mean

Standard Dev.

Min.

Max.

PROPLIMa PROPLIM*a PERSLIMa PERSLIM*a LJJPROPb,c LJJPERSb,c TPPRCAPb,c TPPECAPb,c POLCAPb ILLITRTb PPCAPb WAGEd URBANb

47 44 45 38 48 48 48 48 48 48 48 48 48

4.30 4.32 1.93 1.66 4.02 12.9 0.41 0.18 0.50 8.36 0.98 0.49 0.40

1.579 1.548 1.095 0.582 5.364 32.144 0.170 0.077 0.325 7.323 0.494 0.129 0.241

1.000 1.000 1.000 1.000 1.293 3.119 0.042 0.007 0.111 1.700 0.483 0.267 0.109

6.000 6.000 6.000 6.000 39.144 228.167 0.777 0.368 1.533 29.000 3.346 0.838 1.000

Sources: a Wood, Gould & Moore (1916). b U.S. Bureau of the Census (1910). c U.S. Bureau of the Census (1904) Prisoners and Juvenile Delinquents in Institutions. d Dodd (1993).

Statutes of Limitations for Accident Cases: Theory and Evidence

59

damages, then deterrence is mitigated. These effects imply a longer statute and a positive sign for ILLITRT. The variables LJJPROP and LJJPERS may also capture the victim’s litigation costs. Specifically, higher values of these variables should lower litigation costs, both because potential claimants have lower search costs and because there is more competition among lawyers (Posner, 1997, p. 481). This implies a negative sign since lower victim costs make it easier to file suits, thereby increasing deterrence.16 This effect reinforces the legal error argument above. Alternatively, if these variables measure the injurer’s litigation costs, they should have a positive sign according to (14). As a final measure of litigation costs, we included the variable URBAN, which measures the percentage of population living in urban areas. Generally, one would expect urbanization to be associated with easier access to the legal system for accident victims (Danzon, 1984), implying lower litigation costs and a shorter statute of limitations. Alternatively, increased urbanization may reduce the probability that injurers will be detected or apprehended. To see the implication of this interpretation, recall that many state statutes in 1916 grouped accidental injuries with intentional injuries for both personal injury and property damage. In the theoretical model, the possibility of underdetection can be reflected by a decrease in expected damages, D, for which the injurer will be held liable.17 Thus, factors that decrease the likelihood of detection decrease deterrence, thereby causing the optimal statute length to increase according to (15). This interpretation therefore predicts a positive coefficient for URBAN. In addition to URBAN, we employed three further variables to capture the likelihood of detection. First, we included POLCAP, the number of police per thousand, in both regressions. As the number of police per capita increases, we expect the probability of detection to increase, thereby allowing the statute of limitations to decrease. Second, we included the number of prisoners per thousand jailed for crimes against property (TPPRCAP) and for crimes against person (TPPECAP), in the property and personal injury regressions, respectively. Larger values of these variables are also an indication of a greater likelihood of apprehension and, thus, should have negative signs. In summary, the two regression models we fit are: PROPLIM = ␣0 + ␣1LJJPROP + ␣2POLCAP + ␣3TPPRCAP + ␣4ILLITRT + ␣5WAGE + ␣6PPCAP + ␣7URBAN, and

(18)

60

MATTHEW BAKER & THOMAS J. MICELI

PERSLIM = ␤0 + ␤1LJJPERS + ␤2POLCAP + ␤3TPPECAP + ␤4ILLITRT + ␤5WAGE + ␤6PPCAP + ␤7URBAN.

(19)

B. Results The results of the OLS estimation are presented in Tables 3 and 4 for the statutes of limitations for property damages and personal injury, respectively. On the whole, the model performed better for property damage statutes. In particular, the results for the personal injury equation were very sensitive to specification: the model did not fit the complete sample well, but performed reasonably well when the questionable observations were excluded. In contrast,

Table 3. Results for Dependent Variable = Property Damage Statute of Limitations Sample Sample Size Ind. Variable INTRCPT LJJPROP POLCAP TPPRCAP ILLITRT WAGE PPCAP URBAN R2 Adj. R2 F Prob. > F

Full

Full

Full

Full

Ltd.

Ltd.

Ltd.

47

47

47

47

43

43

43

7.764* (6.75) –0.127* (–2.85) –0.559 (–0.59) –6.001* (–3.04) 0.071** (–2.12) –2.899 (–0.95) 0.652 (0.84) 2.917*** (1.82)

7.637* (6.86) –0.128* (–2.93) –

7.263* (7.53) –0.137* (–3.17) –

7.311* (8.21) –0.135* (–3.28) –

7.654* (6.79) –0.130* (–3.00) –

7.185* (8.16) –0.134* (–3.32) –

–5.929* (–3.10) 0.078** (–2.44) –0.997 –(0.50) –

–6.622* (–3.56) 0.072** (–2.22) –

–6.480* (–4.20) 0.074** (–2.42) – –

1.047*** (1.82)

2.041** (2.04)

–6.550* (–3.34) –0.076** (–2.23) –2.878 (–0.97) 0.702 (0.93) 3.068** (2.42)

–6.075* (–4.30) 0.078** (–2.45) –

0.718 (0.14) 2.117*** (1.84)

7.613* (6.69) –0.127* (–2.92) –0.588 (–0.63) –6.409* (–3.23) –0.072** (–2.07) –3.122 (–1.04) 0.786 (1.02) 3.760** (2.22)

0.379 0.268 3.40 0.006

0.365 0.287 4.71 0.002

0.361 0.283 4.64 0.002

0.361 0.300 5.93 0.001

0.433 0.323 3.93 0.003

0.429 0.334 4.59 0.001

0.411 0.350 6.79 0.001

* Significant at 99% level; **Significant at 95% level; ***Significant at 90% level.

– 2.687** (2.46)

61

Statutes of Limitations for Accident Cases: Theory and Evidence

the property damage model performed just as well for the full and limited samples. One reason for this difference may be that bodily injuries are not as readily measured in economic terms as damages to property. In both models, variables constructed from aggregate value measures, such as WAGE and PPCAP, were not significant. Though this was not unexpected for WAGE given the offsetting predictions, the failure of PPCAP to be significant, especially in the property equation, was disappointing because the prediction was clear. The variables constructed to proxy for legal error for the most part performed well. Specifically, the ratios of lawyers, judges, and justices to the number of prisoners for property crimes (LJJPROP), and to the number of prisoners for personal injury crimes (LJJPERS), were both significant at the

Table 4. Results for Dependent Variable = Personal Injury Statute of Limitations Sample Sample Size Ind. Variable INTRCPT LJJPERS POLCAP TPPECAP ILLITRT WAGE PPCAP URBAN R2 Adj. R2 F Prob. > F

Full

Ltd.

Ltd.

Ltd.

Ltd.

44

37

37

37

37

2.525** (2.48) –0.005 (–0.79) 0.410 (0.50) –0.086 (–0.03) –0.026 (–0.75) 0.187 (0.07) –0.335 (–0.48) –0.683 (–0.50)

1.615** (3.20) 0.005*** (–1.86) 0.774*** (1.96) –5.103* (–3.20) 0.015 (0.88) 0.263 (0.19) 0.253 (0.75) 0.230 (0.32)

1.768* (5.62) 0.005*** (–1.93) 0.853* (2.75) –4.872* (–3.33) 0.011 (0.80) –

1.882* (6.75) 0.005*** (–1.92) 0.759** (2.67) –4.39* (–3.31) –

2.042* (8.17) 0.005*** (–1.79) 0.751** (2.62) –4.09* (–3.11)

–

–

0.271 (1.47) –

0.209 (1.26) –

–

0.042 –0.140 0.23 0.976

0.330 0.173 2.11 0.074

0.325 0.219 3.08 0.022

0.312 0.228 3.73 0.013

0.279 0.215 4.38 0.010

* Significant at 99% level; ** Significant at 95% level; *** Significant at 90% level.

–

62

MATTHEW BAKER & THOMAS J. MICELI

90% or better level in nearly all regressions. The negative signs for these variables are consistent with the legal error and victim litigation cost arguments. To illustrate the impact of these variables, note that the coefficient on LJJPROP is around ⫺ 0.13 in all equations. Thus, if one state has a ratio of 1 and another 2, the second state’s statute would be about 0.13 years shorter, all else equal. The impact of LJJPERS is also negative and significant in all but one equation, but the magnitude of the effect is smaller. The variables measuring the likelihood of apprehension also performed well in both models. Specifically, the number of prisoners for property crimes per thousand (TPPRCAP) and the number of prisoners for personal injury crimes per thousand (TPPECAP) are both negative and significant at the 99% level in all but one equation. The coefficient on TPPRCAP is around ⫺ 6 in all equations. Thus, if one state has a rate of 0.1 (i.e. 1 prisoner per 10,000 in population) and another has a ratio of 0.2, the second state’s statute would be 0.6 years shorter, all else equal. The effect of TPPECAP in the personal injury model is similar, as the coefficient ranges from ⫺ 4.09 to ⫺ 5.1. The variable ILLITRT is negative and significant in the property equation, reflecting the litigation cost argument, but it was insignificant in the personal injury equation. The coefficients in the property equations are around ⫺ 0.075. Thus, if one state has an illiteracy rate of 5 and another has a rate of 10, the second state’s statute would be 0.375 years shorter, all else equal. The variable URBAN was significant at the 90% or 95% level in all property equations and positive in sign. This result is consistent with the argument that greater urbanization makes detection more difficult. The coefficient on URBAN ranges from 1.047 to 3.760. A value of 3 indicates that an increase in the urbanization rate from, say, 0.1 to 0.2 would increase the statute of limitations by 0.3 years, all else equal. URBAN was not significant in any of the personal injury equations. Finally, the variable POLCAP is not significant in the property damage regressions (though it was of the right sign), and it is significant in all but one of the personal injury regressions (but of the wrong sign). This is surprising given the significance of the other variables aimed at measuring the probability of detection.

V. CONCLUSION This paper has developed an economic analysis of statutes of limitations for accident cases by formalizing the trade-off between litigation costs and injurer precaution. Using a simple accident model, we showed that the optimal statute length balances the incentives for precaution from a longer statute, against the

Statutes of Limitations for Accident Cases: Theory and Evidence

63

increased litigation costs arising from the court’s tendency to make errors in determining negligence. As a test of the model, we looked at cross-state variation in statutes of limitation for both personal injury and property damage cases. Despite problems in constructing the data and finding adequate proxies, our results suggest that economic factors do play a significant role in determining these variations. Variables that performed especially well were those aimed at proxying the efficient operation of the legal system in terms of its tendency to make errors and the likelihood that injurers could avoid detection. In contrast, variables measuring victim damages and injurer costs of precaution did not perform well. This most likely reflects the difficulty of capturing these costs with aggregate measures, like the wage rate and per-capita value of property.

ACKNOWLEDGMENTS We acknowledge the assistance of Lewis Kurlantzick and the comments of Keith Hylton.

NOTES 1. See, for example, Epstein (1986, p. 679); Landes & Posner (1985, p. 567); and Posner (1992, p. 80). 2. See Png (1986), Polinsky & Shavell (1989), and Hylton (1990a,b). 3. In the Appendix, we develop a simple version of the model with varying injurer costs of care based on Hylton (1990b) and show that the results in the text are unaffected. In contrast to Hylton, however, we retain the assumption of a single D for all victims so that we can continue to derive a comparative static with respect to victim damages. 4. See U. S. v. Carroll Towing Co., 159 F.2d 169 (1947). 5. See, for example, Png (1986), Polinsky & Shavell (1989), and Hylton (1990b). 6. More realistically, both q1 and q2 will likely increase with time as memories fade and evidence becomes stale. The impact of this effect is described in footnote 9 below. 7. Specifically, the probability of correctly finding a negligent injurer liable, 1 ⫺ q1, exceeds the probability of finding a non-negligent injurer liable, q2. The same condition, when rearranged, implies that 1 ⫺ q2 > q1. 8. The assumption that (1 ⫺ q1)D > q2D > cv implies that no potential suits are deterred by the victim’s litigation costs given legal error. In a more general model with continuous D, the condition that 1 ⫺ q1 > q2 would result in more suits being filed when injurers are negligent than when they take care. However, in contrast with a model without error, there are now some suits in both cases, which is the key difference. 9. Note that the right-hand side of (8) is decreasing in both q1 and q2. Thus, increases in either type of error reduce deterrence. As a result, if q1 and q2 increase with time, the deterrence function of trials will fade, thereby reducing the marginal benefit of increasing L.

64

MATTHEW BAKER & THOMAS J. MICELI

10. Of course, this L* is not necessarily the same as the L* in Fig. 1(a). 11. Thus, they differ from frivolous suits that have no merit and are only filed in hopes of obtaining a settlement. 12. Incidentally, the language employed in 21 James I, c. 16 is still part of many modern, state statutes. For example, the phrase ‘. . . actions for account, debt, detinue, and replivin for goods and chattels’, is part of the original English statute, and still appears in various forms in the current statutes. 13. Kelly (1903); Angell & May (1876); and Wood, Gould, and Moore (1916) all contain information on the history of the statutes of limitations. Angell and May, and Wood, Gould, and Moore contain specific information on the state statutes of the time. 14. Two states that entered the Union in 1912, Arizona and New Mexico, are also included in our sample. 15. A modern, comprehensive summary of state statutes is in Leiter (1996). This source also contains detailed descriptions of differences in a variety of state laws. 16. Recall that in the theoretical model, we assumed that all victims filed suits so victim litigation costs did not affect the optimal statute length. 17. Even those who cause purely accidental damage may have an opportunity to avoid responsibility, as when someone sideswipes a parked car. 18. This requires a redefinition of a type 2 error to be a finding of negligence by the court when the injurer either has taken care or is not potentially negligent. 19. The assumption that all victims have the same damages, D, implies that either all victims file, or none file. In a more general model where victims have varying D, some would file and some would not, provided that w > 0. See Hylton (1990a, 1990b). 20. Note that in a world without legal error, cases (2) and (3) could not be equilibria since no injurers would ever be found negligent.

REFERENCES Angell, J. K., & May, S. W. (1876). A Treatise on the Limitations of Actions at Law (6th ed.). Boston: Little, Brown, and Co. Danzon, P. (1984). The Frequency and Severity of Medical Malpractice Claims. Journal of Law and Economics, 27, 115–148. Dodd, D. (1993). Historical Statistics for the States of the United States: Two Centuries of the Census 1790–1990. Westport, CT: Greenwood Press. Epstein, R. (1986). Past and Future: The Temporal Dimension in the Law of Property. Washington University Law Quarterly, 64, 667–722. Hylton, K. (1990a). The Influence of Litigation Costs on Deterrence Under Strict Liability and Under Negligence. International Review of Law and Economics, 10, 161–171. Hylton, K. (1990b). Costly Litigation and Legal Error Under Negligence. Journal of Law, Economics, and Organization, 6, 433–452. Kelly, J. (1903). A Treatise on the Code of Limitations of Actions. St. Paul: Frank P. Dufresne. Landes, W., & Posner, R. (1985). A Positive Economic Analysis of Products Liability. Journal of Legal Studies, 14, 535–567. Leiter, R. A. (1996). National Survey of State Laws (2nd ed.). Detroit: Gale Research. Netter, J., Hersch, P., & Manson, W. (1986). An Economic Analysis of Adverse Possession. International Review of Law and Economics, 6, 217–227.

65

Statutes of Limitations for Accident Cases: Theory and Evidence

Png, I. (1986). Optimal Subsidies and Damages in the Presence of Legal Error. International Review of Law and Economics, 6, 101–105. Polinsky, A. M., & Shavell S. (1989). Legal Error, Litigation, and the Incentive to Obey the Law. Journal of Law, Economics, and Organization, 5, 99–108. Posner, R. (1992). Economic Analysis of Law (4th ed.). Boston: Little, Brown, and Co. Posner, R. (1997). Explaining the Variance in the Number of Tort Suits Across U. S. States and Between the United States and England. Journal of Legal Studies, 26, 477–489. United States, Bureau of the Census, 13th Census. 1910. Washington, D.C.: Government Printing Office. United States, Bureau of the Census. Prisoner and Juvenile Delinquents in Institutions, 1904. Washington, D.C.: Government Printing Office, 1907. Wood, H., Gould, J., & Moore, D. (1916). A Treatise on the Limitations of Actions at Law and in Equity (4th ed). Albany: Matthew Bender.

APPENDIX: THE MODEL WITH VARYING INJURER COSTS OF CARE This Appendix generalizes the model in the text by allowing injurers to differ in their costs of care along the lines of Hylton (1990b). For simplicity, we assume that there are two types of injurers, those with low costs of care, cl, who comprise a fraction ␣ of the population; and those with high costs of care, ch, who comprise a fraction 1 ⫺ ␣. (Consideration of a continuum of cost types, as in Hylton, would greatly complicate the model without adding new insight.) Further, we assume that cl < (pn ⫺ pc)D < ch.

(A1)

Thus, using Hylton's terminology, only low-cost types are potentially negligent in the sense that a court, acting without error (and ignoring litigation costs) would find them negligent if they failed to take care. Note, therefore, that when the court acts with error, low types are susceptible to both types of errors (i.e., false ‘convictions’ and false ‘acquittals’), whereas high types are susceptible only to false convictions (type 2 errors).18 We assume, again following Hylton, that victims cannot observe either the care choice of injurers or their costs of care. Thus, we define w to be the victim's assessment of the probability that the court will find an injurer negligent, given that an accident has occurred. A victim will therefore file suit if wD ≥ cv. The value of w and the resulting number of suits will be determined in equilibrium. Assume for now that wD ≥ cv, so that all victims file suit in the event of an accident.19 (Obviously, this requires w > 0 in equilibrium, as discussed below.) Given a statute of limitations of L, low-cost injurers take care if and only if cl < c0l , where c0l is given by the right-hand side of (8). As for highcost injurers, they take care if and only if

66

MATTHEW BAKER & THOMAS J. MICELI

ch < G(L)(pn ⫺ pc)(q2D + ci) ⬅ c0h.

(A2)

Note that only the probability of a type 2 error (q2) affects the care choice of high-cost injurers because, as noted, they are never correctly found negligent. Comparison of c0l and c0h shows that c0l > c0h for all L > 0. Further, both are increasing linearly in L as shown in Figure A1. Thus, as L increases, the following three outcomes arise in sequence: (1) neither type of injurer takes care (L < L1); (2) only low-cost injurers take care (L1 ≤ L < L2); and (3) both types take care (L ≥ L2). Although victims do not observe individual injurer types or their care choices, they correctly perceive the equilibrium outcome. Thus, the probabilities of plaintiff victory in each of the three cases are as follows: (1) w = ␣(1 ⫺ q1) + (1 ⫺ ␣)q2 = q2 + ␣(1 ⫺ q1 ⫺ q2); (2) w = q2; and (3) w = q2.20 If we assume, as in the text, that q2D > cv, then all victims will file suit in each case. The optimal L is found in the above model by minimizing social costs subject to the equilibrium decisions of injurers, as described above. Social costs in each of the three cases are given by SC1 = pn[D + G(L)(ci + cv)]

(A3)

SC2 = ␣{cl + pc[D + G(L)(ci + cv)]} + (1 ⫺ ␣)pn[D + G(L)(ci + cv)]

(A4)

SC3 = ␣cl + (1 ⫺ ␣)ch + pc[D + G(L)(ci + cv)].

(A5)

Note, however, that each expression is only relevant over the range of L that induces the corresponding choice of care by injurers, as described above. Thus,

Fig. A1.

Statutes of Limitations for Accident Cases: Theory and Evidence

67

to find the optimal L, we first minimize each expression over the relevant range, and then compare the minimized expressions. Note first, that all three expressions are increasing in L, reflecting the higher litigation costs associated with a longer statute of limitations. Thus, each expression is minimized at the lowest value of L in the relevant range. For example, in case one, L should be set at zero, resulting in no care by injurers and social costs of pnD. As in the text, we ignore this case as uninteresting and focus on cases two and three. In case two, the cost minimizing statute length is L1 and, in case three, it is L2. Substituting these values into (A4) and (A5), respectively, and comparing the resulting expressions shows that case two is optimal if and only if (1 ⫺ ␣){(pn ⫺ pc)D ⫺ ch + [pnG(L1) ⫺ pcG(L2)](ci + cv)} < ␣pc[G(L2) ⫺ G(L1)](ci + cv).

(A6)

Given (A1), this expression may or may not hold. However, if we assume that ch > (pn ⫺ pc)(D + ci + cv), or that care is not desirable by high-cost injurers, even when we take account of litigation costs and assume that the statute of limitations is infinite, then the left-hand side of (A6) is negative while the righthand side is positive. Thus, the condition holds and case two is optimal. This is the most plausible case, since it implies that there are some injurers in the population (a fraction 1 ⫺ ␣) for whom care is not efficient under any circumstances. (It is, therefore, also most like Hylton's model with a continuum of injurers.) When case two is optimal, the results from the simple model in the text go through exactly. In particular, the optimal statute of limitations is given by L1 in Figure A1, which is determined by the equation cl = c0l , or condition (8) in the text written as an equality. The comparative statics in (13)–(17) therefore remain the same, except that c is replaced by cl in (13).

This Page Intentionally Left Blank

EDUCATING ALICE: LESSONS FROM THE COASE THEOREM Steven G. Medema and Richard O. Zerbe Jr. 1. INTRODUCTION The Coase theorem has fascinated, perplexed, comforted, and infuriated both economists and legal scholars for more than three decades. In ‘The Problem of Social Cost’ (1960), Ronald Coase argued that, from an economic perspective, the goal of legal decision making should be to establish a pattern of rights that would generate an efficient allocation of resources. This in and of itself is by no means non-controversial, as about twenty years of debate between law and economics scholars and those legal scholars of traditional and radical bents attests. But this point was overshadowed when Coase, using his now-famous example of crop damage caused by straying cattle, noted that once legal rights are fully specified, negotiations between affected parties will lead to an efficient outcome under the standard assumptions of competitive markets (especially, zero costs of transacting) – and that this outcome will be unaffected by the decision as to whom rights are initially assigned. In Coase’s words, it is necessary to know whether the damaging business is liable or not for damage caused since without the establishment of this initial delimitation of rights there can be no market transactions to transfer and recombine them. But the ultimate result (which maximizes the value of production) is independent of the legal position if the pricing system is assumed to work without cost (Coase, 1960, p. 8).1

The debate over the validity of this proposition has been raging ever since. The literature dealing with the Coase theorem is simply enormous, and even after more than three decades the debate over the theorem is still going strong. Research in Law and Economics, Volume 19, pages 69–112. Copyright © 2000 by JAI/Elsevier Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0308-5

69

70

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

The literature is dominated by theoretical arguments, but within the last fifteen years or so there have been a significant number of studies which attempt to assess the applicability of the theorem in experimental and empirical contexts. The purpose of this essay is to examine these three-plus decades of literature dealing with the Coase theorem – not so much with the goal of passing judgment on the various claims for and against its validity (although we shall do this as well), but, rather, in the attempt to clarify the underlying importance of the Theorem and the application of these insights to economic analysis.

2. THEOREM(S) AND IMPLICATIONS 2.1. Theorem(s) The first formal statement of the Coase theorem (and its public naming) came in 1966, when George Stigler (1966, p. 113) introduced the topic in his text on price theory and suggested that ‘The Coase theorem . . . asserts that under perfect competition private and social costs will be equal’. The theorem has been stated in numerous ways subsequent to its initial formulation, including: . . . if one assumes rationality, no transaction costs, and no legal impediments to bargaining, all misallocations of resources would be fully cured in the market by bargains (Calabresi, 1968, p. 68, emphasis in original). . . . in a world of perfect competition, perfect information, and zero transaction costs, the allocation of resources in the economy will be efficient and will be unaffected by legal rules regarding the initial impact of costs resulting from externalities (Regan, 1972, p. 427). If transaction costs are zero the structure of the law does not matter because efficiency will result in any case (Polinsky, 1974, p. 1665). . . . if there were (a) no wealth effects on demand, (b) no transaction costs and (c) rights to pollute or control pollution, the allocative solution would be invariant and optimal, regardless of the initial assignment of rights (Frech, 1979, p. 254). In a world of zero transaction costs, the allocation of resources will be efficient, and invariant with respect to legal rules of liability, income effects aside (Zerbe, 1980, p. 84). . . . a change in a liability rule will leave the agents’ production and consumption decisions both unchanged and economically efficient within the following (implicit) framework: (a) two agents to each externality bargain, (b) perfect knowledge of one another’s (convex) production and profit or utility functions, (c) competitive markets, (d) zero transactions costs, (e) costless court system, (f) profit-maximizing producers and expected utility maximizing consumers, (g) no wealth effects, (h) agents will strike mutually advantageous bargains in the absence of transactions costs (Hoffman & Spitzer 1982, p. 73). . . . when parties can bargain together and settle their disagreements by cooperation, their behavior will be efficient regardless of the underlying rule of law (Cooter & Ulen 1988, p. 105).

Educating Alice: Lessons from the Coase Theorem

71

. . . a change in [law] affects neither the efficiency of contracts nor the distribution of wealth between the parties (Schwab, 1988, p. 242).

While in many ways similar to one another, these statements of the theorem contain important differences, many of which are at the heart of the theoretical debates over the theorem.2 Nonetheless, a casual reading of these statements reveals two general claims about the outcomes. The first is the ‘efficiency hypothesis’: regardless of how rights are initially assigned, the resulting allocation of resources will be efficient. The second claim – the ‘invariance hypothesis’ – is that the final allocation of resources will be invariant under alternative assignments of rights. The debates over the correctness of the Coase theorem, and/or its proper form, have turned on both of these hypotheses, and this struggle has been manifest in the current tendency to appeal to two different versions of the theorem – the ‘strong’ version, which encompasses both the efficiency and the invariance propositions, and the ‘weak’ version, which encompasses the efficiency proposition only.3 2.2. Implication I: Externalities When Coase undertook his analysis, he was concerned with economists’ framework for the treatment of externalities and market failure. A. C. Pigou’s (1932, p. 173) original elaboration of this framework argues that a divergence between the values of marginal private and marginal social net product would not ‘make the national dividend a maximum; and consequently, certain specific acts of interference with normal economic processes may be expected . . . to increase the dividend’. It is not too much of an exaggeration to say that this approach has generated a sort of syllogism that goes, ‘externalities are the source of market failure, market failure suggests government intervention, and thus government intervention is suggested by externalities’. This syllogism is not as much in evidence now as it was when Coase was writing – due in large part to Coase, but it is still prominent. In one of the leading textbooks on the new science of policy analysis, David Weimer and Aidan Vining reach a conclusion that appears frequently in the literature: When is it legitimate for government to intervene in private affairs? In the United States, the normative answer to this question has usually been based on the concept of market failure – a circumstance where the pursuit of private interest does not lead to an efficient use of society’s resources or a fair distribution of society’s goods (Weimer & Vining, 1992, 30; MacRae & Wilde, 1985, 170).

72

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

Moreover, textbooks on microeconomics and public finance commonly present the concept of market failure as a general justification for government intervention (Browning & Browning, 1992, 657; Boardman, 1996, 99). What is the Coasean objection to the concepts of externality and to market failure? Baumol & Oates, 1975, 17 suggest that ‘[a technological] externality is present when one individual’s (say A’s) utility or production relationships include real (that is, non-monetary) variables whose values are chosen by others (persons, corporations, governments), without particular attention to the effects on A’s welfare’.4 Market failures are thought to occur through the presence of externalities, public goods, information asymmetries, monopolies, or inappropriate income distributions. All of these can be treated as externalities or similarly to externalities so our statements are meant to apply broadly to all of these. It is to this tradition of market failure that the Coase theorem is addressed. The Coase theorem shows that in a world of neoclassical competitive markets, where transaction costs are taken to be zero, Pigouvian remedies are not necessary for an efficient resolution of externality problems. All that is required is a common law or statutory rule which clearly assigns rights pertaining to the externality. From this point, bargaining or market phenomena will lead us to an efficient outcome. A corollary to this is that when rights are well defined, the observed situation will be efficient, as affected parties will have taken all Pareto-improving steps. In such a situation, any attempt to employ Pigouvian remedies will make matters worse, rather than better. Thus, the Coase theorem shows that the syllogism does not hold. One of the most important consequences of the Coase theorem is that, over time, it has tended to erode the traditional concept of market failure built around externalities and, in some quarters, caused it to be replaced with a transaction cost approach (Zerbe & McCurdy, 1997). Coase (1937) suggested that firms come into being because the use of the price system is not costless, and arise as entrepreneurs seek to reduce the transaction costs associated with using the price system. The property rights approach to transaction costs developed mainly after Coase’s 1960 article on ‘The Problem of Social Cost’ through the work of Alchain (1965), Demsetz (1967), Barzel (1985), Ostrom (1990), Allen (1991), and others. Here, transaction costs are defined as the resources necessary to transfer, establish and maintain property rights (Allen, 1991, 4). Property rights, under this approach, are defined in terms of the extent to which property may be used, alienated (i.e. sold), and non-owners excluded from its use. Associated transaction costs include the expenses associated with searching for goods, bargaining, and enforcing rights.

Educating Alice: Lessons from the Coase Theorem

73

If the costs of transferring and protecting or maintaining property are zero, and property rights are well established,5 then there will be (costless) markets for everything and the first welfare theorem holds (Arrow, 1969). Externalities then exist only because of transaction costs. Those goods that are then defined by externalities are not produced because transaction costs are too high. But for all goods more are not produced because costs are too high. So how do goods which are not produced because of an externality differ from all goods? The Coasian answer is that they do not differ. Or if they differ, it is only because the costs that prevent their production are transaction costs.6 But in this case the focus should be on transaction costs, so that when we say that market failure and externalities are the wrong story we mean to include all of these categories. One of the many implications of the Coase theorem is that the analysis of when it is efficient for government to intervene must consider the absolute advantage of government with respect to powers of coercion. Transaction costs are, for example, particularly affected by the legal system, with the result that legal changes can materially affect transaction costs. Second, government through its powers of coercion has access to means of transactions others do not have. Thus, government can, for example, tax to provide a good in which exclusion costs are high. 2.3. Implication II: Law While the Coase theorem filleted at least the standard interpretation of the received Pigouvian wisdom, it’s implications for law went to the very heart of the legal system. The theorem tells us that the form of legal rules does not matter – only their presence or absence. Thus, we will have the same amount of pollution (and thus clean air or water) and of outputs associated with the generation of pollution regardless of whether polluters or the victims of pollution are made liable for pollution damage. The same amount of effort will be devoted to precaution against causing tortuous injury regardless of whether injurers or victims are liable for harm caused. The structure of law pertaining to breach of contract will have no impact on the allocation of resources through the contracting process. Attempts by judges to engage in social engineering from the bench will be fruitless, apart from distributional (as opposed to allocational) effects. Assuming that rights are alienable, the allocation of resources will be the same regardless of the rule of the law, and that allocation will be efficient. More generally, it is a matter of indifference whether courts impose property rules or liability rules (Calabresi & Melamed, 1972), and the

74

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

entire issue of adherence to precedent becomes a moot point in terms of its effect on the allocation of resources. But beyond this, the Coase theorem and the illustrations that Coase drew from legal cases showed how the analysis of legal rules could be placed squarely within the economist’s world of determinate optimal solutions. Since any given assignment of rights expands the opportunity sets of some agents while restricting those of others, legal decision making could easily be converted into an exercise in optimal allocations, based upon which one could derive, in relatively straightforward fashion, the efficient legal rules to govern individual behavior. In essence, the Coase theorem opened the door to the analysis of rights allocation within a traditional market framework and to the asking of a relatively simple question: If rights over scarce resources are allocated through the market for all manner of goods, why not also for rights over pollution, the ability to breach contracts, tortious harms, etc.? To a mindset which finds market allocation congenial, the theorem opened up a vast new scope for the operation of markets. If market processses are able to work and allowed to work, legal outcomes will be exactly those dictated by the laws governing competitive markets. The implications of this insight are straightforward and were quickly seized upon: (i) let markets work in allocating rights; (ii) facilitate this process by removing legal impediments to the market’s operation; and (iii) when (i) and (ii) are not possible, assign rights so as to mimic the outcome of a competitive market – the outcome that would have obtained in any event had there not been impediments to the market’s operation. This argument is amazingly powerful, since it implies that the law should simply be structured to allow people to do what they would naturally agree to do if transaction costs did not preclude them from doing so. *** Neither Coase nor any of the innumerable subsequent commentators has given us a generalized, formal proof of the Coase theorem. (In Coase’s case, at least, this should come as no surprise, given that he never formally stated a theorem either.) Arguments pro and con (the latter sometimes in the guise of supposed disproofs) tend to be made via demonstrations that the theorem does or does not hold up under this or that set of assumptions or within some particular context. But all too often, the authors fail to pay sufficient attention to the assumptions set down by Coase, or, at least, to give sufficient thought to their underlying meaning. Indeed, with perhaps a single exception (income effects), all of the supposed disproofs or refutations of the theorem involve a violation of one or more of the theorem’s underlying assumptions – usually the

Educating Alice: Lessons from the Coase Theorem

75

assumptions of well-defined rights and zero transaction costs. Yet, what is interesting in this, and what we will explore here, is the extent to which these assumptions are interesting.

3. TRANSACTION COSTS 3.1. What are Transaction Costs? Coase’s own analysis, and most of the early (i.e. 1960s and 70s) literature dealing with the Coase theorem, was couched within what we shall call a quasicompetitive framework. The typical context here includes small numbers of parties, full information, competitive product and input markets, and an absence of strategic behavior among the parties to the externality – the sort of context usually posited in the standard textbook Edgeworth box model.7 Perhaps the most sticky issue in the debate over the Coase theorem is the meaning given to the assumption of zero transaction costs. The very concept of transaction costs has been so vague and ill-defined that Stanley Fischer (1977, p. 322 at note 5) was once led to remark that ‘almost anything can be rationalized by invoking suitably specified transaction costs’. Coase’s (1960, p. 15) definition of transaction costs encompasses those costs associated with search, negotiation, monitoring, and enforcement, which basically reduces to ‘resource losses incurred due to imperfect information’ (Dahlman, 1979, p. 148). A bit more specificity (although even greater breadth) is found in more recent definitions within the property rights literature such as Barzel’s (1989, p. 2) contention that transaction costs are ‘the costs associated with the transfer, capture, and protection of rights’, and Allen’s (1991, p. 3) statement that they encompass ‘the resources used to establish and maintain property rights’. 3.2. The Non-Separability of Cost Functions One of the first sophisticated attempts at refuting the Coase theorem involves the presence of a non-separable victim cost function.8 With a non-separable cost function, the level of pollution damage to B is a function not just of A’s output, but of B’s output as well, and thus a given level of output by A causes B more harm (i.e. causes a greater increase in Bs costs) the more output B produces. If A is liable for damage caused, B contributes to its own damage without having to bear the cost, since it is fully compensated. B thus has no incentive to mitigate damages and produces an inefficiently high level of output. Moreover, the associated damage liability imposes a supra-optimal cost on A, causing it to restrict its output to a sub-optimal level (Marchand &

76

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

Russell, 1973, pp. 613–615). Only when the victim is liable will it have the incentive to appropriately mitigate damages, thus allowing for efficient externality resolution through Coasean bargaining. However, it is well established that if both externality-relevant activities are controlled by a single owner, the overall allocation of resources will be efficient and invariant (Coase, 1960, p. 16). Moreover, one can imagine a contract between owners that mimics the effect of single ownership assuming costs of monitoring and negotiation are zero. This demonstration is sufficient to negate the nonseparability critique, since the inefficiency contemplated will be exploited through contracting or merger, which can be achieved costlessly, or by an entrepreneur.9 One response to this argument has been to point to the difficulties associated with carrying out a merger – that is, to introduce transaction costs (Marchand & Russell, 1975). But introducing transaction costs is no argument against the correctness of the Coase theorem.10 3.3. Non-Convexities at the Negotiation Starting Point A now classic article by Starrett (1972) on the nonconvexities in the production or consumption sets of externality victims has long been considered the final nail in the coffin of the invariance proposition. Taking the case of a producer victim, we can see the effect of the externality on the victim’s production set in Fig. 1. Here, q indicates the victim’s level of output and z the level of the externality. The nonconvexity is introduced because marginal damage will be zero beyond

Fig. 1.

Educating Alice: Lessons from the Coase Theorem

77

some level (here, z0) of externality emission; that is, beyond z0, the victim firm will shut down. If the victim firm is given the right to be free from pollution, then point A becomes the starting point for the Coasean bargain, and parties will negotiate to an efficient outcome such as point C, with output level q* and externality level z*. But suppose instead that the externality-generating firm is given the right to pollute, and that the level of the externality associated with its profitmaximizing level of output is z1, making point B the starting point for negotiations. The minimum payment that the polluting firm will accept to reduce pollution by one unit is the corresponding reduction in its profits. However, since marginal damage at and around z1 is zero, the victim firm will not be willing to offer a bribe to induce such pollution reduction. Here, the equilibrium is z = z1 > z* and q = 0 < q*, and the invariance proposition is thus negated.11 While the editors of the Journal of Economic Theory contended that Starrett’s nonconvexities argument ‘destroy[s] the validity of the Coase theorem’ (Editorial Addendum 1977, p. 222), we contend that the nonconvexities critique, too, reflects a failure to fully understand the nature of a zero transaction costs world and the information-related nature of the nonconvexities problem. There is a Pareto-better point available, but the nonconvexity means that the parties will fail to reach it because the immediate marginal adjustments are not Pareto-better. The victim will not be able to spend 15 dollars for a change that will make it 20 dollars better off because the first step along this path would involve spending a dollar with no resulting welfare gain. Not being willing to take this first step, the victim will never know that better things are on the horizon. The problem here is one of information: it is certainly the case that the victim would take this first, welfare-reducing step if it was certain that it would be better off in the end. If both victim and polluter knew of the existence of a superior position they could also merge to achieve it. The nonconvexities argument introduces imperfect information into the model, violating the zero transaction costs assumption of the Coase theorem. This critique of the theorem points to the importance of information costs, but, if these are considered as part of transaction costs, it does not point to the incorrectness of the theorem itself. 3.4. The Game-Theoretic Framework While most of the early literature debating the validity of the Coase theorem employed a quasi-competitive framework, the more recent debate has been couched largely in terms of game-theoretic bargaining. The (quite sensible)

78

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

rationale for this change in perspective is that the quasi-competitive framework is not appropriate or relevant for Coase theorem-like bargains over rights, owing to the potential for strategic behavior in small numbers bargaining situations.12 To treat small numbers bargaining without allowing for the possibility of strategic behavior assumes that individual and group rationality coincide and that the parties will reach efficient agreements simply because it is in their joint interest (Samuelson, 1985, p. 322; Regan, 1972, pp. 429– 431).13 If parties have full information about each other’s utility (or profit or production and cost) functions, the Coase theorem will hold in a noncooperative setting. Suppose that A is given the right to pollute. Then the victim, B, has an incentive to offer an alternative allocation of resources (for example, a combination of a bribe paid from B to A and a reduced level of pollution by A) to A, but B knows that A will accept this offer only if A’s utility is increased under the new allocation. Since B knows A’s utility function, he can determine the range of allocations sufficient to garner A’s acceptance. And, since B’s utility is higher with reallocation than without, he will not offer an allocation that A would reject, choosing instead that allocation from the group A will accept which maximizes his own utility. The resulting equilibrium is Pareto efficient, since, given the utility level of one player, the other player’s utility is maximized.14 The same reasoning applies to the situation where the victim is given the right to be free from pollution (Arrow, 1979, pp. 27–29; Schweizer, 1988). In fact, it is not necessary for each party to enter the bargaining process with full information, only that each party perceives that there are net gains to it from providing full information during the negotiation process and thus will reveal such information during that process (Saraydar, 1983, p. 603 at note 12). The problem is that players are unlikely to know each others’ respective utility (or profit, production or cost) functions. While it is possible that information will be revealed through the bargaining process, the application of more complex strategic thinking suggests that private information, if revealed, may be used against one’s self and thus adversely affect one’s payoff. Given this, agents have an incentive to conceal information and to expend resources both to protect the value of their own private information and to acquire information from/about others. These costs, and resultant delays and/or failures to consummate mutually-beneficial bargains are likely to preclude the attainment of efficient negotiated solutions where information is asymmetrically distributed (Cooter, 1982; Sutton, 1986; Farrell, 1987).15 Following Cooter (1982 pp. 20–24), suppose that uncertainty regarding the opponent’s response causes each player to form a rational, subjective

Educating Alice: Lessons from the Coase Theorem

79

probability distribution over his opponent’s moves. Each player then chooses for himself the strategy that maximizes his expected utility based on a comparison of the increased gains from taking a harder line in bargaining with the higher probability that this harder line will prevent an agreement from being reached. The problem is that, while each player will be playing the strategy that is optimal against the distribution of its opponent’s possible strategies, this strategy is not necessarily optimal against the particular strategy played by the opponent. The outcome will be inefficient when players err in their predictions of the moves made by their opponents. In fact, zero communication costs may actually decrease the possibility of reaching an agreement, in that this facilitates the transmission of threats and other strategic communications. Several other information-related problems can also arise. First, clear-cut rules of liability create a moral hazard problem which can preclude the attainment of efficiency. Suppose that the victims are liable for damage incurred, but that they are unable to ascertain with certainty the baseline level of the externality – against which Coasean bribe-induced reductions in the level of the externality are to be measured. In such a case, the externality-generating firm can influence the level of the bribe that it receives by making an upward adjustment in the level of the externality-generating activity at the time that the baseline level of the externality is set, or by failing to take cost-justified precautions. Conversely, if polluters are known to be liable for damages, then, in the absence of full information about actual damages and measures taken by the victim to mitigate damages, the victim’s moral hazard will result in too few resources being devoted to precaution/mitigation by the victim and too many resources being devoted to abatement by the polluter (Kamien, Schwartz & Dolbear, 1966; Tybout, 1972; Harris, 1990). Moreover, disagreements over the baseline level of pollution may result in the failure to consummate bargains when the victim is liable. In each of these situations, the outcome of the Coasean bargain will be inefficient, owing to the effects of imperfect information. With endogenous liability assignment, however, the moral hazard problems disappear, since a party does not know with certainty whether or not it will be forced to bear the costs of the externality. Thus each party will act efficiently to minimize expected costs by engaging in the appropriate level of precaution/preventive activity (Harris, 1990, pp. 701–702). Second, imperfect information opens the door to various forms of extortion. Polluters may threaten to emit higher levels of pollution in order to secure a larger bribe (Mumey, 1971). ‘Entrepreneurs’ may threaten to emit an externality in order to secure a bribe, or, potential victims may threaten to ‘come to the harm’ in order to secure a bribe (depending on the direction of

80

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

liability). If these threats necessitate the use of resources to establish credibility, the result will be inefficient (Shoup, 1971, pp. 310–12).16 Furthermore, since each individual is interested in both achieving the benefits from cooperation and getting as large a share of the benefits as possible for himself, there will be threats of noncooperation in order to increase one’s share of the gains. For these threats to be credible, however, they must occasionally be carried out, and, when this is done, the result will be sub-optimal (Regan, 1972, p. 429).17 Third, the presence of large bargaining groups raises issues related to free riding and coalition formation. Individuals will have an incentive to free ride, and thus the ability of the group to pay a bribe sufficient to induce the socially optimal level of output/pollution will be greatly reduced. Moreover, if there are differential damage effects across victims, we may observe the rise of coalitions within the victim group (for example, by level of damage), each applying pressure to encourage the result that best suits its interests. The greater is the number of coalitions, the smaller is the likelihood that the optimal solution will be reached. However, as the number of parties approaches infinity (with large numbers of right-holders and rights-seekers), the bargaining solution here will approach the efficient result of competitive equilibrium (Wellisz, 1964, p. 354; Samuelson, 1985, p. 338). Because the game-theoretic challenges rely on the presence of imperfect information, they run afoul of the zero transaction cost assumption and thus are not valid against the theorem itself.18 Defenders of the game-theoretic approach dispute this charge by pointing out that giving this type of content to the idea of zero transaction costs basically renders the idea of bargaining meaningless and detaches the theorem almost completely from reality, making it either tautological (Regan, 1972, pp. 429–30; Cooter, 1989, p. 67) or more akin to ‘astrology than . . . market analysis’ (Veljanovsk, 1982, p. 60). In fact, by this definition the theorem’s efficiency proposition must hold, since any violation of it reflects a ‘[cost] associated with the transfer, capture, or protection of property rights’. The lesson to be taken here is that transaction costs are ubiquitous (Allen, 1991, p. 4; Zerbe & McCurdy, 1997), the effect of which is indeed to make the theorem per se completely devoid of applicability to the real world. Coase (1981, p. 187) has made this point a bit more graphically, contending that analyzing a world of zero transaction costs is akin to ‘divining the future by the minute inspection of the entrails of a goose’. Properly understood, then, the information problems at the heart of the nonconvexities and game theoretic critiques go not to the correctness of the theorem, but to the importance of information costs – and transaction costs generally – in bargaining arrangements.19 That is, the important question

Educating Alice: Lessons from the Coase Theorem

81

raised by the Coase theorem has to do with the role played by transaction costs in real-world bargains. 3.5. Assessing the Role of Transaction Costs: Empirical and Experimental Studies A number of experiments have been undertaken to test the Coase theorem under conditions of full and of imperfect information. Those undertaken by Hoffman & Spitzer20 present the subjects with a range of possible outcomes, each with a different associated payoff. One party (the ‘controller’) is given the ‘property right’ and thus can determine the outcome unilaterally. For example, suppose that the set of possible dollar payoffs (PA, PB) is (5,0), (4,4), and (0,5), and that A is the controller. He will chose (5,0) unless B induces him to choose a different outcome. The Coase theorem predicts that (4,4) will be chosen, with the actual distribution of the joint payoff ($8) being a function of the negotiation process. And of course it is in B’s interest to offer A up to $4 to choose (4,4) and in A’s interest to accept any payment greater than $1 to do so. In well over 500 experiments with bargaining groups of up to twenty individuals, Hoffman & Spitzer found substantial support for the predictions of the Coase theorem. In all, the parties bargained to the efficient result 92% of the time, including 94% of the time under conditions of full information and 90% of the time under imperfect information. Moreover, the support for the theorem’s prediction was actually greatest in the large numbers bargaining, where one might expect transaction costs to be the highest. The large groups tended to devise their own informal institutional arrangements (choosing representatives from the group as bargaining agents) to overcome the large numbers problems. Coursey, Hoffman & Spitzer (1987) conducted experiments that actually introduced a discomforting externality, in order to determine whether ‘affronts to dignity’ that might be associated with an externality would influence the willingness of people to bargain. This was accomplished by introducing the possibility that the ‘victim’ would have to hold one ounce of an unpleasanttasting liquid in his or her mouth for twenty seconds. These experiments had two possible payoffs: (i) the ‘polluter’ gets $0, and the victim gets $10 and does not have to taste the liquid; (ii) the ‘polluter’ gets $20, and the victim gets $10 and does have to taste the liquid – the latter being the efficient outcome. Over the course of 40 experimental trials, the efficient outcome was chosen 95% of the time – 100% of the time when the polluter was the controller and 89% of the time when the victim could control the outcome.21 Based on their various

82

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

findings, Hoffman & Spitzer (1986, p. 162, emphasis added) contend that their results ‘produce a presumption in favor of the Coase theorem’, by which they mean that ‘a judge or a legislator should start his analysis by presuming that the parties can and will, in general, exhaust the gains from trade available through private bargaining’.22 While these experiments offer substantial support for the applicability of the Coase theorem when transaction costs are low and the stakes are relatively small, the results of a labor law-oriented experiment by Stewart Schwab (1988) are less favorable to the theorem’s applicability. The graduate student subjects in Schwab’s experiments were asked to negotiate a union contract over wages, vacation time, and the ability of the company to transfer work to its nonunion plant over the course of the three-year contract. The implications for the Coase theorem lay in the contract presumption that was said to govern labor relations in the absence of a specific contract provision: in one group of experiments, subjects were told that the legal presumption was that the company must continue to use union workers unless the contract explicitly states otherwise (i.e. includes a ‘go clause’), while the other group was told that the presumption was that the company could transfer work to the nonunion plant during the course of the contract unless the contract explicitly stated otherwise (i.e. includes a ‘stay clause’).23 The points system used in these experiments allowed for a range of efficient wage and vacation time outcomes, and each experiment had a specific efficient contract presumption regarding the transfer of labor to the nonunion plant. Analysis of the results of these experiments shows that only about 20% of the contracts were fully efficient when wage levels, vacation time, and the stay or go clause are accounted for, a vast difference from the roughly 90% efficiency of the Hoffman & Spitzer and the Harrison & McKee experiments.24 Schwab suggests that three factors may account for these differences. First, parties in these experiments were bargaining over multiple contractual terms under a binding time constraint, and thus may have found it difficult to make efficient choices on all items. Second, the subjects were not given full and perfect information, which meant both that information had to be communicated during the negotiation process and that signaling and bluffing could occur, leading to inefficient agreements. Finally, the subjects did not know what the ‘best’ outcome was from the beginning, and thus had to find their way to it and do so over a rather large bargaining range, which, of course, could easily result in inefficient outcomes. Given these factors, the rather high rate of failure to reach efficient bargains is not particularly surprising. The environment of these experiments corresponds much more closely to a natural setting – and to the setting contemplated by many of the game-theoretic challenges to the

Educating Alice: Lessons from the Coase Theorem

83

theorem – than do many of the other experimental treatments, and the factors that exist in these settings are such as to weigh heavily against the attainment of efficient results. Several empirical studies also point to the effects of transaction costs in legal-economic arrangements. Between 1850 and 1890, there were no less than 150 different laws enacted by the California legislature altering the rules that governed cattle trespass in ways that benefited farmers.25 The Coase theorem predicts that these alterations in the law will have no effect on the amount resources devoted to ranching and farming, or on ranching and farming outputs. Contrary to the theorem’s prediction, however, these legal changes resulted in an enormous increase in farm output (Vogel, 1987). Transaction-cost-induced nonconvexities explain why the Coase theorem fails here. If ranchers are not liable for trespass damages, the fact that each farmer has to negotiate with all ranchers whose cattle might stray onto his land in order to prevent damage means that the failure to contract with any given rancher whose cattle actually graze on his land will all but render useless the bribes paid to other ranchers. The farmer will thus have little incentive to initiate such negotiations, with the result that efficiency will obtain only if it is efficient for cattle to roam freely. This nonconvexity is not present when farmers are given the right, as the ranchers needs only to negotiate with those farmers onto whose land his cattle might potentially stray. Thus, we would expect that output would be lower when the ranchers were not liable (Vogel, 1987, pp. 174–76, 187). This points to the importance of assigning the legal rule to minimize transaction costs as both Coase (1960, p. 19) and Posner (1983, p. 71) suggest should be done. Further evidence on the role of information-related transaction costs can be found in John J. Donohue’s (1989) analysis of the Illinois employment experiment (Spiegelman & Woodbury, 1987), which attempted to determine whether the payment of bonuses to unemployed workers for securing employment, or to employers for hiring unemployed workers, would reduce the duration of unemployment and the costs associated with the unemployment compensation system. While the experiment was conducted to ascertain how such bonuses might affect the duration of unemployment, its application to the Coase theorem is straight-forward: mutually advantageous bargains will be struck under either scheme, all workers and employers who satisfy the eligibility requirements will collect bonuses, and members of the workerpayment-group (WPG) will find jobs and collect bonuses at the same rate as members of the employer-payment-group (EPG).26 The results of the Illinois experiment contradict the predictions of the Coase theorem on all counts: the number of bonuses paid to WPG workers was about

84

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

5 times that paid to EPG employers; many eligible workers and (especially) employers failed to collect their bonuses; and members of the WPG had a significantly greater improvement in obtaining employment, relative to the control group, than did members of the EPG (Donohue, 1989, pp. 569–577). Moreover, there was no significant difference in wages across the WPG and EPG hirees, so that aggregate (wages plus bonus) compensation was higher for WPG workers than for EPG workers. That is, it appears that workers did not bargain with employers over the wage or the bonus as a result of this program (Donohue, 1989, pp. 586–590). Donohue (1989, pp. 591–601) asserts that transaction costs are extremely low here, and thus that the theorem fails in a case most favorable to its success. However, it appears that he greatly underestimates the effect of transaction costs within this experiment. At the most basic level, if transaction costs were zero, the entire premise of the experiment would disappear, since job search is a positive transaction cost phenomenon. Furthermore, acquiring this so-called ‘free’ bonus money actually involves a substantial number of steps (and even more for employers than for workers), the aggregate effect of which is to make the process rather costly, relative to the size of the bonus.27 Indeed, since program-related transaction costs are higher under the EPG system than under WPG, we would expect more EPG participation and bonus collection – expectations that are borne out by the results of the experiment (Lindgren, 1990, p. 583). Of course the Coase theorem has implications for all manner of legal rules, including, as Peters (1986, 1992) points out, those governing divorce, which establish property rights with respect to dissolution of marriage. Under unilateral divorce law, the spouse seeking divorce has property rights with respect to dissolution while, under mutual consent, the right rests with the spouse who does not wish to see a divorce occur (Peters, 1992, p. 690). The conventional wisdom was that the shift toward unilateral divorce laws would make divorce easier, thus increasing the divorce rates in states that adopted such laws. However, the Coase theorem predicts that if bargaining costs are minimal and information is symmetric across parties, divorces will only be undertaken when they are efficient (that is, joint benefits exceed joint costs), regardless of the law governing divorce, and that the legal rule will have no impact on the divorce rate. The evidence regarding the effect of the legal regime on the divorce rate is the subject of some dispute. Peters (1986) finds that both the divorce rates and the distribution of compensation under the alternative legal rules are consistent with the Coase theorem’s predictions. However, a number of subsequent studies find that the evidence tends to support the conclusion that divorce rates

Educating Alice: Lessons from the Coase Theorem

85

are in fact higher in unilateral-divorce states than in mutual-consent states.28 But even if one is willing to accept the result that divorce rates are not impacted by the legal rules governing divorce, it seems unlikely that these results actually reflect the working of Coase-theorem-type mechanisms. First, there is no evidence to suggest that ‘unilateral’ divorces are undertaken only when they are efficient. Peters infers that the efficiency proposition holds based on a questionable claim that transaction costs are low and the fact that the data confirm the model’s invariance and distribution predictions. Allen (1992, 1995), for example, argues that, by working an uncompensated transfer of wealth from wives to husbands, the move to no-fault divorce violates the zero transaction costs/fully-specified rights condition assumed by the theorem, and thus that the rise in the divorce rate does not constitute a legitimate argument against the theorem. Legal costs (from attorney fees to asset searches) and psychic costs associated with divorce only add to this.29 Furthermore, the Coase theorem predicts not just an invariant divorce rate, but an invariant allocation of household resources as well – just as Coase’s farmer-rancher example predicts an invariant allocation of resources devoted to farming and ranching. The fact that female labor force participation is higher in states with unilateral divorce rules thus speaks loudly against the claim of invariance. Some have inferred that the high rate at which suits are settled prior to trial supports the Coase theorem (Hoffman & Spitzer, 1986, pp. 168–169). Glanter (1983, pp. 28–30) finds that roughly 90% of all lawsuits are settled before they go to trial, and that, when they do not settle, it tends to be due to (i) cases that require a judicial decree to be settled; (ii) cases that are not costly to litigate, which reduces the incentive to settle; (iii) the placing by one or more parties of special value on having a judicial decree for reasons including, inter alia, precedent and reputation; (iv) cases that involve an issue that is not easily negotiated over, such as a ‘fundamental value’; and (v) the high transaction costs associated with settlement as compared to going to trial – factors that lie outside of the bounds of the Coase theorem. However, there is no way to infer from the settlement data whether the parties have bargained to the socially optimal result contemplated within a zero transaction costs world, or if they simply have realized some of the potential gains from negotiation but hit a point where the transaction costs from further negotiation exceed the expected gains, and choose to settle at a ‘sub-optimal’ outcome because this settlement, although not optimal, is still better than going to trial.30 The pervasiveness of transaction costs – particularly as illuminated by the many supposed disproofs of the Coase theorem and in the empirical evidence just touched upon – cast substantial doubt on any conclusion of a general presumption in favor of the theorem’s realistic applicability. It is one thing to

86

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

show that the Coase theorem is largely confirmed in a laboratory setting that attempts more or less to mimic the zero transaction costs world, but it is quite another to say, with Hoffman & Spitzer, that the results thus generated establish ‘a presumption in favor of the Coase theorem’ for efficiently resolving realworld externality problems with twenty or fewer parties. What may be established is that externalities will be resolved efficiently through negotiations in many, and perhaps even most instances where transaction costs are very low and the stakes are very small. But to move from this to claims of wide-spread applicability and presumptions in favor of the theorem in real-world cases is a somewhat different matter, requiring both caution and a great deal of further study as to the propensity of people to bargain over legal rights, the role of transaction costs in this process, and the outcomes of such bargains. Indeed, the empirical evidence points to the importance of the potential difficulties for the applicability of the theorem that are raised by the theoretical challenges discussed above, such as the effects of imperfect information, the potential for strategic behavior, and nonconvexities, and, as such, begin to get at the role played by transaction costs in legal-economic bargains.

4. INCOME, TASTE, AND PREFERENCE EFFECTS A further complication is introduced when one or both parties to the externality are consumers, for, at this point, we are forced to take into account effects on demand that attend alternative assignments of rights.31 These effects arise, first, from differences in tastes between parties, and, second, from differences between willingness to pay (WTP) and willingness to accept (WTA). The general rule is that any source of diverence between WTP and WTA for the same persons will lead to an exception to the invariance part of the Coase theorem as long as tastes are different. If tastes are different (that is, as long as indifference curves are not homothethic), alternative distributions of income resulting from alternative specifications of rights will affect the pattern of demand and therefore the pattern of resource allocation, though not its efficiency. For example, suppose that an economy produces only beef and fish, and that fertilizer runoff used to produce grass, an input into beef production, reduces fish production. Those that produce beef prefer to eat beef and those that produce fish prefer fish. A change in the liability rule from one in which beef producers are liable to one in which they are not will increase incomes of beef producers and reduce incomes of fish producers. This will then increase the demand for beef and reduce the demand for fish; thus, the relative production of beef and fish will not be invariant to the liability rule.

Educating Alice: Lessons from the Coase Theorem

87

Legal change can alter the sense of ownership and thus change the measure of value from the WTP to WTA or vice versa. Following Willig (1976), economists have tended to assume that any differences between WTA and WTP owing to a price change are small. This is now recognized as untrue in important cases. For environmental goods, for example, researchers have demonstrated repeatedly that WTA questionnaires generate values from two to nineteen times greater than those elicited by WTP questions (Levy & Friedman, 1994, p. 495 at n. 6; Hoffman & Spitzer, 1993, pp. 69–85). There are three reasons for the difference: income effects, substitution possibilities, and loss aversion.32 Let us first consider the implications of income effects. If people experience diminishing marginal utility of income, the utility loss resulting from a given reduction in income is greater than the utility gain associated with an equivalent increase in income. Thus, if individuals bargain over utility, rather than over wealth per se, we would expect to see differences between WTA and WTP, and hence negotiated solutions that vary with the initial assignment of rights (Hovenkamp, 1990). Since the maximum amount a person is willing to pay to avoid damages is a function of his budget constraint, while there is no such constraint on the amount that the individual is willing to accept, we will see (as long clean air is a normal good) a difference in the amount of pollution, and thus in pollution-related output, depending on the initial assignment of rights (Mishan, 1971, p. 19). The invariance proposition will hold only when income effects are not present or when all relevant income elasticities of demand are zero. The income effects critique has led some commentators to restate the Coase theorem with an income effects qualification, while others use the weak rather than the strong version of the theorem for this same reason.33 A further argument against the invariance proposition comes from the influence that alternative assignments of rights may have on WTA v. WTP through effects on consumer tastes and preferences. Here, the assertion is that consumer tastes and preferences are influenced by the structure of legal rules (Zerbe, 1996). Prospect theory posits that individuals have a value (rather than utility) function which is convex for gains and concave for losses, and that the degree of concavity is greater than the degree of convexity, so that losses of a given size are felt more acutely than gains of that same size (Kahneman & Tversky, 1979). The link between this idea and the WTA v. WTP argument is that the offer of money to relinquish a right would be treated as a loss, whereas the purchase of a right would be regarded as a gain. Since losses count more than gains, the minimum amount that an individual would be willing to accept to relinquish a right will exceed the amount that he is willing to pay to acquire it (Kahneman, Knetsch & Thaler, 1990). One reason for this loss aversion is

88

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

suggested by ‘endowment effects’ – people valuing ‘received income’ more highly than ‘opportunity income’. Because of this, people will be willing to forego more opportunity income to retain a right than they would spend in received income to acquire it, and thus WTA will exceed WTP (Thaler, 1980; Kelman, 1979). While the evidence regarding the existence of a divergence between WTA and WTP is not uniform, it is by no means clear that the divergence is trivial. The survey evidence shows a substantial difference between WTA and WTP, and recent experimental studies support the contention that WTA may be substantially greater than WTP – often more than twice as great (Levy and Friedman, 1994; Hoffman & Spitzer, 1993, pp. 69–85). An experimental study undertaken by Kahneman, Knetsch & Thaler (1990) probes this issue in the context of the Coase theorem and, in a number of different types of experiments, finds significant endowment effects. Using items such as mugs, pens, binoculars, and chocolate bars, they find that individuals, when given the opportunity to exchange these items for cash, exhibit a strong reluctance to part with entitlements and thus that, contrary to standard assumption of economic theory, preferences are apparently not independent of entitlements (Kahneman, Knetsch & Thaler, 1990, p. 1339). The value that the subjects place on these objects ‘appears to increase substantially as soon as the individual is given the object’ (Kahneman, Knetsch & Thaler, 1990, p. 1342). Moreover, the resulting disparity between WTA and WTP does not dissipate in repeated trials (i.e. with market experience). Consider next the issue of substitution possibilities. Recently, Hanneman (1991) showed that the poorer the substitutes for the good, the greater the divergence between the WTP and the WTA. Put another way, the divergence will be greater the more unique the good. For most people, the WTA to allow degradation of the Grand Canyon will be much greater than the WTP to prevent degradation. Thus, a change in the law that results in a change from a WTP to a WTA criterion will have dramatic effects on the measure of value for the good. The experimental findings of Kahneman, Knetsch & Thaler provide support for this hypothesis, suggesting that endowment effects are most likely to occur for items that are not easily replaceable. This makes the endowment effect particularly important for the Coase theorem, since things like a nice view, or clean air or water, are not easily replaced, and it thus can be expected that people will refuse to sell such goods even at prices somewhat greater than their reservation price for buying them. How important these forces are in creating a wedge between the WTP and the WTA is as yet uncertain. At this point, however, it seems reasonable to say that income and substitution effects and loss aversion are sufficient to invalidate

Educating Alice: Lessons from the Coase Theorem

89

the invariance, although not the efficiency, claim of the Coase theorem when consumers are party to an externality. A further complication arises because endowment effects reduce the perceived gains from trade as compared with a world in which preferences are independent of endowments. Since fewer mutually advantageous exchanges are possible, the volume of trade is lower than it otherwise would be (Kahneman, Knetsch & Thaler, 1990, p. 1344).34 Given the size of the potential disparity between WTA and WTP, one can conceive of situations where each party’s WTA is greater than the other party’s WTP, so that no trade would occur. Indeed, experiments run to test this implication in a Coase theorem context revealed substantial under-trading relative to the theorem’s predictions (Kahneman, Knetsch & Thaler, 1990, pp. 1339–1341). Zerbe (1996) and Knetsch (1997) have argued that the correct measure of damages as between the WTP and WTA is a function of the psychological reference point, which may not correspond with the legal reference point.35 For example, Ellickson’s (1986) study of the response to alternative range laws shows such a difference. In one half of Shasta County, California, open range was the legal rule, and in the other half closed range was the rule. Yet in both parts of the county, in spite of opposite assignments of liability, people expected and provided similar remedies. Cattle owners took responsibility for the damages in all cases, and this responsibility existed for many years and was enforced through social norms. The reference state was one of crops not being damaged by straying cattle. One of the more intriguing cases is suggested by the recent literature that considers a situation in which legal ownership is different from psychological ownership. (Zerbe, 1997, 1998). Property rights are fully specified, but there is a divergence between the legal specification and the psychological sense of ownership or entitlement. Evidence suggests that a sense of ownership attends certain environmental goods even if there is no individual ownership. A decision to cut down the last remaining stand of privately owned redwood trees, the Headwater Grove, may create a sense of loss among some that are not legal owners of the grove. This loss is correctly measured by the WTA. If, however, a decision to measure the value of the grove is based on legal ownership, the value to the public will be based on the WTP for preservation. Since for a normal good the WTP will be less than the WTA, the grove may be cut when it should not be. Although property rights in this example are fully specified, they are specified inefficiently. ‘Efficiency’, says Zerbe (1998), ‘requires that the legal measure of property and damage correspond to psychological reference points’. A heuristic proof of this can be made by imagining that the condition

90

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

for efficiency is not met. Imagine that Ronald believes he owns a right or a property, B, and that George also believes that Ronald owns property B (perhaps the right to call a certain theorem their own). They discover that the law, however, holds that George, not Ronald, owns B. Ronald suffers a loss of B psychologically and therefore economically, while George gains B´. Since losses are, on the average, worth more than equivalent gains (due to income effects, substitution effects, and loss aversion), on the average George will gain less than what Ronald loses – that is, on average B > B´. This is perfectly general. The application of law to effect a legal ownership different from psychological ownership must, on average, impose net losses. Whether or not this sort of case constitutes a violation of the efficiency part of the Coase theorem rests again on whether or not information costs are regarded as part of transaction costs. In so far as information costs are assumed to be zero, this case will not arise. Of course if information-cost-induced inefficiencies exist, once the information is revealed trade in rights at zero costs will restore the correspondence between legal and psychological rights so that there is no further violation of the Coase theorem.36 That is, Ronald will buy the right to the property from George at a price between B and B´. Yet, this does not obviate the fact that a net loss previously occurred in violation of a Coase theorem that assumes positive information costs.

5. PROPERTY RIGHTS While disagreement and misunderstanding over what is meant by transaction costs is at the heart of many of the debates over the Coase theorem, an important related issue concerns unspecified property rights. Under the Barzel/ Allen definition of transaction costs, zero transaction costs implies complete property rights (Allen, 1991, 1995; Cheung, 1992).37 A way to understand this reasoning is that property rights under this approach are defined in terms of the extent to which property may be used, alienated (i.e. sold), and non-owners excluded from its use. Associated transaction costs include the expenses that attend searching for goods, bargaining, and enforcing rights. Property rights become more extensive and complete as transaction costs approach zero and are fully specified only when transaction costs are zero. 5.1. Information Costs One of the benefits of Coase theorem discussions is that they help to spell out the range of potential property rights specifications and the issues thereby raised. There is an unspecified property right that lies at the heart of the private

Educating Alice: Lessons from the Coase Theorem

91

information examples. For example, Farrell (1987) constructs a problem in which parties A and B care about a time that is set – for example, the time for beginning a race. He notes that King Solomon would have little trouble finding the optimum value since he charges each party an amount equal to their effect on each other, and that a ‘bumbling bureaucrat’ can under some circumstances find a superior (second best) result to that achieved under zero cost negotiations. The demonstration that a centralized authority can achieve an optimum where bargaining cannot is, however, equivalent to showing that there is an unspecified property right. Efficiency is achieved because ‘the authority’, rather than the bargaining parties themselves, has control over the property right. When A is allowed to set the time (about which both A and B care), however, B will prefer to not participate in bargaining since his gain is greater when A just sets the time unilaterally, and B pays nothing. The advantage that Solomon has is that he can force A and B to participate. Note that in this example, B gets to use the time that A sets. Farrell’s example assumes that B is nevertheless allowed to participate in the race – that is, the race is not owned. But, if the race is owned the owner will in fact charge both A and B, yielding the Solomon solution. Another analogy can be made to ownership of a lake. The owner will charge each user at least the cost they impose on other users. Farrell implicitly assumes that government is the only possible owner of the lake. It is true that we can consider other examples in which the ownership of the resource seems more foreign to our usual thinking, as when the time represents a curfew or a time after which noise must be reduced – but there is a lack of ownership nonetheless.38 5.2. Income and Wealth Effects All of the wealth effects-related objections to the Coase theorem that rest on consumer preferences rest on a change in the distribution of wealth. Yet, these arguments at base reflect property rights that are not fully specified or are inefficiently specified. These conditions then violate the Coase theorem’s assumption that property rights are fully specified. Consider a change in the rule of liability. If property rights are fully defined (in the sense of complete ownership), this alteration of liability cannot take place without compensation; if it does, the right was not fully defined in the first place, in violation of the theorem’s assumptions. Thus, owing to the compensation, the distribution of wealth will be unaffected (Allen, 1995, p. 10). Of course, this rebuttal does not go to the case where non-existent rights are subsequently defined. However, in a world of zero transaction costs the definition of rights would be perfectly anticipated and thus reflected in resource values (Allen, 1995, pp. 10–11).

92

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

5.3. Entry One of the most discussed challenges to the theorem concerns the entry effects induced by liability or bribe payments.39 If polluters are made liable for damages, the flow of liability payments into the victim industry will increase the rate of return in that industry. If one assumes that firms entering the market are also eligible for compensation, then entry will occur in the long run, leading to an increase in the output of the victim industry. When victims are liable, in contrast, the flow of bribe payments from victims to polluters raises the rate of return in the polluting industry, leading to entry into that industry and a corresponding increase in output. In either case, the entry results in a supraoptimal level of output; and, of course the invariance proposition fails to hold because of the disparate entry effects of alternative legal rules.40 The issue of property rights specification provides the key to distinguishing between the competing claims regarding invariance in the presence of entry. Specifically, the entry argument violates the assumption of fully-specified rights. Suppose that ranchers are liable for damage done by their cattle. If rights are fully specified, the flow of liability payments will be capitalized into the value of farmland that adjoins ranching property, and there will be no incentive for entry into farming in order to secure the bribe. In analogous fashion, any bribes that result from farmer liability will be capitalized into the value of ranch land that adjoins farms, and there will be no incentive to enter ranching. Given this, the long-run entry effects that are said to invalidate the invariance proposition will not occur (Demsetz, 1972a; Frech, 1979). Entry will occur only when potential entrants are able to secure a valuable right without paying for it, but, by definition, such a right is not fully defined (Holderness, 1989, pp. 183–84).41 In the end, invariance turns on the issue of whether rights are assigned to open or closed classes of individuals or entities (Holderness, 1989).42 Consider first the assignment of rights within closed classes, such as landowners. The assignment of rights to one class of owners creates at once a windfall gain for those having the right and a windfall loss for those not having it. However, in a competitive system these windfall gains and losses are immediately capitalized into the value of the land so that both types of land yield a normal rate of return. Since the rate of return for each of these types of land is unaffected by the assignment of rights, there are no incentives for entry or exit. Thus, the invariance proposition holds for closed classes. The invariance claim does not hold for open classes. The distinction between open and closed classes calls attention to a broad category of spurious objections, all based on incomplete property right specifications. Rights in open

Educating Alice: Lessons from the Coase Theorem

93

classes are not delimited to the extent necessary to make market transactions possible. Here, those who are not parties to the lawsuit through which the initial assignment of rights is generated can acquire that right costlessly merely by entering, and this valuable right will not be capitalized into the price of any resource. Entry will indeed result. Similarly, the absence of a right reduces the returns to that activity, thereby inducing exit. The asymmetric entry/exit effects across alternative assignments of rights will thus result in different long-run outputs under alternative assignments of rights, thereby negating the invariance proposition. This distinction illuminates the divergent results obtained by many of those offering support for or claiming to refute the invariance proposition. Those who have found the invariance proposition to be valid in this context have either explicitly or implicitly assumed or worked with examples constituting closed classes, while those finding against invariance have analyzed the problem in open-class contexts – primarily situations with two industries where entry is possible. The invariance proposition is thus applicable to, for example, externalities affecting land values, but is inapplicable to tort situations, such as accident law, where there is free entry into one or both classes, and to assignments of rights that cover all (current and future) entrants into an industry.

6. BEHAVIORAL IMPLICATIONS Still another set of implications raised by the Coase theorem literature arises from the questions regarding individual behavior that emerge from the experimental and empirical studies. We have noted above the implications for the stability of tastes and preferences that flow out of the experiments on WTA versus WTP, but other considerations – particularly regarding the influence of norms on legal-economic bargains – also emerge from this literature. One of the interesting issues raised by the early Hoffman and Spitzer experiments comes from the nearly equal division of the payoffs ( ± $1) in the vast majority of the cases. Individual rationality suggests that when the payoff set (PA, PB) is (5,0), (4,4), (0,5), as in the above example, the controller will not settle for less than $5 and could potentially induce the other party to agree to a $7.99–$0.01 split of the $8 payoff. The fact that so many controllers settle for less than what should be their reservation price suggests either altruism or a potential problem with the experimental environment. It may be that participants in these experiments did not understand the full meaning and import of having a unilateral property right as controllers, or that they did not feel a morally justified right to be the controller, since that position was

94

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

determined on the basis of a coin flip rather than being, in some (for example, Lockean) sense, earned. But even after measures were implemented to control for these issues, 20–30% of the experiments generated less than individuallyrational outcomes, suggesting that the subjects behave more like Lockeans than like utilitarians or egalitarians.43 The results from the experiments with a discomforting externality further call into question the assumption of individual rationality (or the inducement thereof within the experimental environment), since, in half of the experiments, the ‘polluters’ paid the victims to taste the liquid even when the polluters had a unilateral right to force them to do so. There are also several empirical studies that illustrate the ability of parties to work out solutions to disputes along quasi-Coasean lines, but which have behavioral implications at odds with the standard theory. Ellickson’s (1986, 1991) study of the effects of open- versus closed-range laws44 on cattle trespass disputes in Shasta County, California, finds that cattlemen and their neighbors do in fact cooperate to resolve their disputes, regardless of who is liable. However, the evidence also suggests that it is not Coase-theorem-type mechanisms at work here; rather, individuals seem to rely on community norms to determine their behavior. For example, while the theorem predicts that the cattleman would install a fence if he were liable (closed range), and that the neighboring farmer would do so if he were liable (open range), it is almost always the cattleman who installs the fence because both cattlemen and their neighbors believe that the cattleman is morally obligated to do so, since his cattle cause the damage. Moreover, in this and other legal-economic contexts, the citizens seem to be very ignorant of the relevant law and ignore those aspects of the law that conflict with their view of the world. As such, they do not bargain ‘in the shadow of the law’,45 but beyond it; community norms seem to have much more force than the legal rule in place. Ellickson suggests that this may be due to the fact that relations among the neighbors are both complex and continuing, because of which the transaction costs associated with acquiring information and litigating disputes are high, and reliance on norms offers a lower-cost way of resolving these disputes. In a study that has interesting commonalties with that of Ellickson, Nick Hanley & Charles Sumner (1995) examine an externality situation owing to the roaming of red deer in the Scottish Highlands which cause damage to growing trees and, in the process, impose substantial costs on the owners of these forests, the value of the timber from which is diminished. In addition, the wandering deer may destroy growing crops on farmland, and, when they stray onto sheep grazing land, reduce the forage for sheep, thus imposing costs on both farmers and sheep ranchers. The beneficiaries of the red deer population

Educating Alice: Lessons from the Coase Theorem

95

are estate owners, who derive substantial income and estate value from the presence of red deer on their estates (Hanley & Sumner, 1995, pp. 88–91). Given the level of damage, the small number of parties, the ease of quantifying damage to forests, and the relative ease with which estate owners could reduce the size of their herds, the situation seems to reflect an inefficiently-high deer population and a fertile ground for the working of Coase-theorem-type mechanisms. Even so, an extensive study by Sumner (1993) failed to turn up any instances of Coasean bargaining between owners of deer estates and neighboring landowners. What one does observe, however, are ‘Deer Management Groups’ which neighboring landowners have established to coordinate the management of the deer population on estates, forests, and farmland. These groups internalize, and thus reduce the level of, the externality, and effectively avoid the third-party effects that can result with bilateral bargaining. It is interesting to note the parallel between the rise of the cooperative Deer Management Groups and the behavior of neighbors revealed in Ellickson’s study of cattle ranching in Shasta County. While the law offers a low-cost option (free government culling) for dealing with red deer damage, groups of neighboring landowners in essence ignore the law and work out a solution amongst themselves, perhaps because it is the case that the transaction costs associated with the cooperative efforts of the Deer Management Groups are lower than those that would attend bilateral negotiations of the Coasean variety. Like the Ellickson study, the Donohue study also seems to point to a significant role for norms in legal-economic bargains. Donohue offers two possible explanations, both behaviorally based, for why the Illinois employment experiment failed to satisfy the predictions of the Coase theorem. The first is ignorance on the part of workers, who seemed not even to realize that bargaining was possible and may not have understood that the bonuses could aid them in gaining employment. Second, there may have been an authoritarian effect which caused the workers to believe that the bonus was in effect inalienable – that the party designated to receive the bonus was in fact entitled to its full value, an effect that would also discourage the type of bargains envisioned by the theorem. Beyond these, the presence of stigma and institutional rigidity effects would lead one to predict minimal bargaining over wages and bonuses (in an amount different from the control group), a result supported by the Illinois experiment. In fact, Ellickson (1989 pp. 627–28) finds the results of this experiment very consistent with his own study of Shasta County – that people often tend to rely on norms rather than legal rules to govern their behavior, particularly when the stakes are low and there is an

96

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

expectation of a continuing relationship, as among neighbors in Shasta County and between employers and employees in the Illinois experiment. There is even evidence to suggest that norms can function to facilitate bargaining in the absence of a legal framework. Janet Landa’s (1981, 1994) studies of Chinese middleman traders show that these traders form ethnically homogeneous middleman groups which share a like Confucian code of ethics that promotes trust among the traders and facilitates contracting. While there is no formal system of contract law to govern these relationships, the group imposes informal social sanctions, such as ostracism from the group, on those who fail to meet their contractual obligations, and, through this process, is able to achieve a relatively fluid system of contract and exchange.46 Taken as a group, these studies point to the importance of coming to grips with how transaction costs might induce behavior that deviates from the traditional rationality assumption. The studies of Ellickson, Landa & Donohue especially suggest that agents fall back on other factors, such as norms, to govern their actions when transaction costs, and especially information costs, are high. The counter that falling back on these norms constitutes a rational response to the presence of transaction costs misses the point, in that, while such may indeed be the case (or very well may not), rational choice theory provides us with little insight into the problem – that is, into why people do what they do. Rational choice theory may provide an ex post justification for reliance on norms to govern individual relations or decision making, but in and of itself it does not, a priori, enlighten us as to how agents will behave. This is not to say that high transaction costs are the sole rationale for the existence of or reliance upon norms, as witnessed by the Lockean behavior exhibited by subjects in the Hoffman and Spitzer experiments, where transaction costs were very low (and a continuing relationship was not expected). However, the evidence does seem to suggest a greater propensity to rely on norms or rules of thumb to govern decision-making when transaction costs are high – for bounded rationality reasons or otherwise. The explanation for and prediction of such behavior would seem to necessitate the bringing in of insights from other disciplines, as Ellickson and Landa have done. The extent of the implications of this for the behavioral foundations of law and economics remains unclear, although some prominent members of the law and economics community have suggested that certain of the behavioral concepts underlying law and economics should be revised (Ellickson, 1989, 1991; Jolls, Sunstein & Thaler, 1998). Even more unclear are the implications for the behavioral underpinnings of economics generally. While deviations from homo economicus outside of the traditional ‘economic’ realm would come as no surprise to many, it bears keeping in mind that standard economic

Educating Alice: Lessons from the Coase Theorem

97

exchanges are, at their heart, transactions over rights. As such, the questions are whether people respond differently to issues of rights over air, water, etc. than they do to rights over apples, clothing, and automobiles, and whether the level of transaction costs influences the processes by which decisions are made. A negative answer to the former question has been presupposed by the various efforts associated with economic imperialism, but the assumption has been that we can take homo economicus on the road. While not denying the robustness of the extension of the economic model to other disciplines, there seems to be more that a little evidence that we should consider importing as well as exporting here.

7. THE IMPORTANCE OF THE COASE THEOREM In the end, whether the various challenges to the Coase theorem go to its correctness or its relevance comes down to ‘how one interprets the almost mystical world of zero transactions costs’ (Zerbe, 1980, p. 85). But we would argue that the importance of the theorem lies not in whether it is correct, but in the detailed nature of the assumptions required to make it correct. The interesting question about the theorem then is the nature of the transaction costs and the specification of property rights associated with those situations in which the theorem is thought to not be correct. The legacy of the theorem lies in the subsequent work attempting to detail the nature of transaction costs and their effect on the workings of the economic system. The richness and variety of types of economic arrangements can now be seen in the richness and variety of transaction costs and in mechanisms for reducing them. In this sense the importance of the Coase theorem lies not in its supposed correctness or incorrectness and the corresponding policy relevance or lack thereof, but, rather, in the positive transactions cost propositions that flow from it. These include both normative and positive propositions. Examples are Posner’s (1992, p. 52) normative suggestion that at law, ‘[s]ince transactions are never costless in the real world, efficiency is promoted by assigning the legal right to the party who would buy it . . . if it were assigned initially to the other party’, and the positive prediction of Lesser et al. (1996) that suits at law in situations where negotiation costs are low will involve considerations of distribution not efficiency. The Coase theorem has helped to give rise to a extensive body of work, much of it summarized by Eggertsson (1990), concerned with economic behavior and institutions and to a more detailed and useful sense of what is meant by property. The Coase theorem has made clearer the relationship between transaction costs and property rights and in doing so has begun to give a much

98

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

stronger basis for understanding how legal regimes change in response to changes in constraints (North, 1981). One can now define the strength of property rights in terms of lower transaction costs for the exclusion, exchange, and use of property. In part of the economics literature at least (Eggertsson, 1990), the transaction cost approach has replaced the market failure model of public intervention that is expressed by Weimer & Vining (1992, p. 30): When is it legitimate for government to intervene in private affairs? In the United States, the normative answer to this question has usually been based on the concept of market failure – a circumstance where the pursuit of private interest does not lead to an efficient use of society’s resources or a fair distribution of society’s goods.

When the costs of transaction mechanisms are introduced there are departures from perfect markets. These departures are in fact externalities because they represent effects not taken into account in the decision-making process. Externalities, then, are found everywhere there are transaction costs, and are ubiquitous. Since the concept of market failure rests on externalities that are defined by transaction costs, the concept of market failure (and the concept of externality) does no work for us that is not already done by transaction costs (Zerbe & McCurdy, 1996). Externalities are in fact an unnecessary complication in the theory of government intervention. Public goods represent a useful example of a situation in which the market failure model can be, and to some extent is being, replaced by a transaction cost model. The older market failure approach is represented by Samuelson (1954), who saw public goods as a class of market failures. For example, in analyzing the classic case of the lighthouse, Samuelson (1964, p. 45) writes: Here is a later example of government service: lighthouses. These save lives and cargoes; but lighthouse keepers cannot reach out to collect fees from skippers. So, says the advanced treatise, ‘we have a divergence between private advantage and money cost . . . and true social advantage and cost . . . Philosophers and statesmen have always recognized the necessary role of government in such cases of ‘external-economy divergence between private and social advantage’.

The transaction cost–property rights approach appears to provide a richer vehicle of analysis, as shown by Cheung, North, and others. For example, Coase (1974) shows that the British lighthouse system was once a wellfunctioning private system and that, in general, the system was more complex than that suggested by the simplistic market failure diagnostic. The triumph of the transaction cost approach shows that the true legacy of the Coase theorem lies not in its correctness, but in drawing attention to the role played by transaction costs within the economic system.

Educating Alice: Lessons from the Coase Theorem

99

8. CONCLUSION In light of the foregoing discussion, three things can be said about the Coase theorem. First, it is correct, in the sense that it has withstood all of the challenges mounted against it to date. Efficiency will obtain, regardless of the initial assignment of rights, and the result will be invariant keeping in mind that even income effects are irrelevant if one accepts the Barzel/Allen definition of zero transaction costs. Second, the theorem, although correct, is unrealistic, as Coase recognized.47 The latter point, of course, should have been obvious from the beginning, which raises the question as to why the debate over the theorem has been so intense. A small part of the answer, within the economics profession at least, may lie in the interesting theoretical puzzle that the theorem poses. Dwarfing this, however, is the normative debate that, implicitly or explicitly, pervades nearly all aspects of the theorem’s discussion. Three prescriptions for legal-economic policy are said to flow from the Coase theorem. (1) Rights-cum-market solutions are said to be preferable to Pigouvian remedies for the resolution of externality problems. (2) Property and contract are efficient; any interference with the outcomes so generated will make matters worse rather than better. It is this implication that makes the Coase theorem, in the minds of some, ‘the cornerstone of a laissez-faire legal and economic policy regarding contract and property law’ (Hoffman & Spitzer, 1986, p. 151). (3) When transaction costs are positive, rights should be assigned to those who would possess them in the end-state if transaction costs were zero, as seen in the prescription of wealth maximization, or ‘mimic the market’. But the Coase theorem says none of these things. The theorem is a positive statement with no normative implications; it is an ‘is’ statement, not an ‘ought’ statement. Each of the above propositions rests on the assumption that efficiency is the goal of legal-economic policy. But the Coase theorem goes merely to the presence absence of efficiency; it does not tell us that it is all that matters or even that it matters at all. It is this normative leap that seems to underlie most of the hostility to the Coase theorem – and, by extension, to law and economics generally.48 Furthermore, even if one takes efficiency to be the goal of legal-economic policy, the Coase theorem does nothing to establish the sanctity of property and contract or the superiority of the market over Pigouvian remedies, owing to the ubiquitous nature of transaction costs. If coordination is costless, the market will optimally allocate rights and resources, but so too will Pigouvian remedies.

100

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

On the other hand, the Pigouvians fare no better in this debate since, after waving away the Coase theorem on the grounds that transaction costs are positive, they tend to immediately fall back on the demonstration that Pigouvian remedies generate socially optimal outcomes, using models in which government is assumed to operate with full information and without cost.49 Here, we come to the true import of the Coase theorem. The theorem is not, in the end, about markets or about costless bargaining; rather, it is about the costs of coordination. If coordination is costless, markets function perfectly; but so does government. If coordination is costly, markets function imperfectly; but so does government. The task for legal-economic policy thus becomes that of ascertaining the magnitude and influence of these costs and the resulting implications for alternative institutional-policy arrangements. The true and valuable legacy of the theorem is all of the subsequent work on transaction costs that explores the costs of coordination under different regimes and in different situations. Like the Coase theorem itself, this, too, is without direct normative implications. However, it has led to certain normative claims, as noted above. One, from an analytical perspective, is that the received conception of externalities should be abandoned. Another, this time from a policy perspective, is that judgments as to the appropriate form of government intervention should be made on the basis of what institutional arrangement produces the lowest combination of coordination costs. In this regard, it is interesting to note that much of the normative debate and the propositions we noted above in that regard can be turned into a series of positive predictions about which arrangements will promote economic efficiency. Where the affected parties could reach a solution through negotiation but choose litigation or regulation, the real issue is likely to be who is to be assigned property rights rather than how to realize gains from trade.50

ACKNOWLEDGMENTS We would like to thank Douglas Allen, James Buchanan, Robert Cooter, Robert Ellickson, Elizabeth Hoffman, Richard Posner, Warren Samuels, and participants in the Workshop in the History of Economic Thought and Methodology at Michigan State University for comments and suggestions on previous drafts of this essay and for helping to clarify our thinking on certain points raised herein. The excellent research assistance of Mary Therese Cogeos is also gratefully acknowledged. Of course, the authors are solely responsible for any remaining errors.

Educating Alice: Lessons from the Coase Theorem

101

NOTES 1. This is as close as Coase comes in his essay to stating what has come to be known as the Coase theorem – an idea actually first put forth in Coase (1959). For useful treatments of ‘The Problem of Social Cost’ as a whole, see Zerbe (1976), Schlag (1986), and Medema (1996); for context, see Coase (1937, 1959). 2. See Medema & Zerbe (2000) for a reasonably exhaustive survey of this literature. 3. The strong version of the theorem is reflected in the above-quoted statements of the theorem by Regan, Frech, Zerbe, and Hoffman & Spitzer, while the weak version is reflected in the statements of Calabresi and Cooter and Ulen. 4. In a simple form, an externality can be expressed by considering two firms producing goods X and Z. The production of Z is said to have an external effect on the production of X if the production of X depends in part on Z. Suppose the production of X depends on labor and on the production of Z. We can write X = f(Lx; Z). The symbol Lx denotes the amount of labor used in producing good X. Good Z appears on the right side of the semicolon denoting that it is not under the control of the producer of good X. Firm X may be in the business of cleaning clothes and firm Z may pollute the air as a byproduct of producing Z, thereby increasing cleaning costs for firm X. We are concerned only with technological externalities. When wages of programmers rise generally because Microsoft expands production, this adversely effects other companies that use programmers, but is a pecuniary externality because it is an effect transmitted through the price system. Such an effect calls for no intervention on efficiency grounds since the price system assures that the benefits of Microsoft’s action exceeds the social costs. 5. Of course if the costs of establishing property rights are seen as part of transaction costs, and therefore are zero, the Coase theorem does not need to assume that property rights are well defined. 6. Consider a public good for which price discrimination costs are high owing to difficulties in determining individual levels of demand. One might not wish to call this an externality but we would say that the transaction cost approach would regard this as a transaction, cost problem similar to those created by externalities. 7. There is another class of competitive Coase theorem models which assumes perfectly competitive markets in the rights over the externality. As the theorem’s correctness in this case is assured by the first optimality theorem (Arrow, 1970), we can safely pass on to the more standard small-numbers competitive analysis. Empirical evidence supporting the efficient internalization of externalities in competitive markets can be found in Steven Cheung’s classic studies of share contracting in China and Taiwan and of relations between beekeepers and apple orchard owners in the northwestern U. S. See Cheung (1969a, 1969b, 1973, 1978, 1980). 8. It is well established that, if the victim’s cost function is additively separable (that is, if CB = C(qA, qB) = C(qA) + C(qB)), then the Coase theorem holds – the outcome is both efficient and invariant under alternative assignments of rights. See Marchand & Russell (1973) and Gifford & Stone (1973). 9. See, for example, Nutter (1968), Coelho (1975), and Zerbe (1980, pp. 87–88). 10. A further argument that can be raised against the nonseparabilities critique is that it violates the assumption of fully-specified property rights, since the victim is able to

102

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

procure revenues from the polluter without giving up anything in return. See the discussion of property rights, below. 11. Note that both outcomes are efficient, in the Paretian sense. Cooter (1980) deals with the design of legal rules to circumvent the nonconvexities problem. 12. See, for example, Davis & Whinston (1962), Samuelson (1966, p. 1141), Shoup (1971, p. 310), Regan (1972, p. 428), and Cooter (1982, pp. 16–17). 13. Indeed, the quasi-competitive framework is rather akin to a cooperative game model of the Coase theorem. Here, of course, efficiency (although not invariance) is assured by definition in a two-person game. And, while the situation is complicated by the presence of three or more players (Aivazian and Callen, 1981), it has been shown that here, too, the efficiency claim holds up (Coase, 1981; DeBornier, 1986). For analyses of the Coase theorem in a cooperative game context, see Davis & Whinston (1965); Arrow (1979); Aivazian & Callen (1981); Samuelson (1985); and Aivazian, Callen & Lipnowski (1987). 14. As Arrow (1979, p. 29) points out, however, this is not the competitive equilibrium. 15. Note that if neither party’s utility function is a function of the other’s private information, then an efficient result will be reached. However, efficiency cannot be guaranteed if either party’s utility function is a function of the other’s private information (Schweizer, 1988, pp. 259–63). See also Arrow (1979 pp. 29–31), Cooter (1982, pp. 20–24), Samuelson (1985), and Illing (1992). 16. It should be noted, however, that Coase (1959, p. 27 at note 54) recognized that the employment of resources ‘solely to establish a claim’ could preclude the attainment of the efficient result, and he assumed away such behavior. Demsetz (1972a, p. 23) has argued that competition for these gains will drive the price of extortion to zero, so that extortion is not a barrier to the attainment of the efficient equilibrium. 17. One can see illustrations of what Regan (1972) has called the ‘a priori argument’ for the theorem in the challenges to the extortion argument. For example, it is argued that the limits of extortion are set by the size of the available rents: if one party tries to extort from the other an amount greater than this, the other party could simply transfer its resources into their next-best use. The extorting agent, being unwilling to forego the potential gain, will thus agree to a solution which garners for it an amount not in excess of the other party’s rents. Thus, the only differential impact of alternative legal rules will be on the distribution of rents; the final allocation of resources will be unaffected (Boyd & Mohring, 1971; Demsetz, 1972a; Feldman, 1974). These rebuttals essentially involve using the quasi-competitive framework to argue against the results derived within a game-theoretic context. 18. On this point see also Allen (1995, pp. 12–13), Dahlman (1979, pp. 158–159 at note 26), Hovenkamp (1990, p. 787), Illing (1992), Samuelson (1985, p. 323), Saraydar (1983), and Zerbe (1980, pp. 85–86). It also bears noting that Coase (1960, pp. 31–33) discusses the problem of moral hazard (along the lines of Harris, 1990), and that he does so in the context of positive transaction costs. 19. This definition of transaction costs also invalidates other challenges to the theorem which were not discussed here. For example, Jung et al. (1995) have shown that rent seeking behavior negates the theorem’s predictions. Yet, the investment of resources to establish a claim violates the zero transaction costs assumption (Coase, 1959; Medema, 1996). Daniel Posin (1994, 1995) has demonstrated that Coase theorem

Educating Alice: Lessons from the Coase Theorem

103

fails in the presence of risk. But as risk, by definition, assumes imperfect information, this involves the introduction of transaction costs into the picture (Medema, 1995). 20. See Hoffman & Spitzer (1982, 1985, 1986), Coursey, Hoffman & Spitzer (1987), and Harrison, Hoffman, Ruström & Spitzer (1987). 21. The willingness of individuals to ‘sell their dignity’ in this context raises the whole issue what sorts of dignity-related activities ought to be salable in the market – for example, prostitution and pornography. While Hoffman & Spitzer are not willing to jump to the conclusion that a willingness to sell the right to avoid tasting bitter liquid implies that prostitution and pornography ought to be legalized, they do not entirely close the door on such matters. Instead, they choose to ‘express no opinion’ as to whether such activities should be allowed, and adopt a more modest position in favor of allowing individuals to transfer moderate amounts of dignity and/or moderate amounts of danger. 22. Hoffman & Spitzer (1986, pp. 163–68) also suggest that the strength of the evidence for the optimality of the bargaining outcomes establishes a presumption in favor of injunctive over damages remedies because of the high probability that the parties will bargain to the efficient result. For discussions of the relative efficacy of property rules and liability rules under varying transaction cost scenarios, see Calabresi & Melamed (1972); Polinsky (1979; 1980); Ayres & Talley (1995a,b); and Kaplow & Shavell (1995,1996). 23. These contract presumptions have actual counterparts in labor law the Milwaukee Spring cases. See Milwaukee Spring Div., 265 N.L.R. B. 206 (1982), Milwaukee Spring Div., 268 N.L.R. B. 601 (1984), and UAW v. NLRB, 765 F.2d 175 (D. C. Cir. 1985). 24. Out of 108 contracts, all but two had efficient wage levels, but only 31% had efficient vacation levels and only about 65% had a stay clause where it was efficient or a go clause where it was efficient. 25. At the time when, in 1850, California joined the Union, its principal industries were mining and cattle raising, and, reflecting the importance of the cattle industry to the state, California had what was, in essence, strict nonliability for cattle trespass. This rule clearly (and seemingly intentionally) favored ranchers, and it played a major role in hindering the development of agriculture in the state (Vogel, 1987, pp. 163–64, 167). 26. The WPG workers would be expected to have lower wages than EPG workers, reflecting a bargaining away of a share of their bonus, as compared to EPG workers’ higher wages as employers bargained away a share of their bonus. 27. Lindgren (1990, pp. 581–582, 585) lists the steps that participants in the WPG and EPG programs must go through in order to collect bonuses and offers several reasons why one would not anticipate an invariant distribution of income or bargaining over the bonuses. See also Ellickson (1989). 28. See Allen (1992) and Peters’ (1992) reply; Zelder (1993a,b); Friedberg (1995); and Brinig & Buckley (1995). These works also contain numerous citations to literature on both sides of the argument. 29. For contrasting views on the potential magnitude of transaction costs here, see Peters (1992, p. 690) and Allen (1992, p. 684). 30. See also Coursey & Stanley (1988). 31. This isssue was first hinted at by Buchanan and Stubblebine (1962) and Turvey (1963), and later more explicitly elaborated by Dolbear (1967) and Mishan (1965, 1967,

104

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

1971) within the economics literature and by Kelman (1979) within the legal literature. 32. Hoffman & Spitzer (1993) present an excellent survey of the WTA v. WTP issue, much of the evidence regarding which comes from the experimental literature. 33. See the above-quoted statements of the theorem for illustrations. 34. This does not imply that endowment effects reduce welfare; it simply means that, because of the value placed on ownership, fewer welfare-improving trades are available. 35. See also the references cited in Knetsch (1997). 36. Underlying this proof is the notion that we cannot speak of it being efficient to change preferences to be in accord with the law since this violates the proper context for benefit cost analysis – which requires that preferences be taken as they lie – and the very concept of efficiency. 37. See also Schlag (1989). 38. In a related vein, in the case of Starrett’s fundamental non-convexities one would ask: who owns the information that a fundamental non-convexity exists? If no such information exists, in what sense then is the result inefficient? 39. See Calabresi (1965), Bramhall & Mills (1966), Tybout (1972), Baumol (1972), Schultze & d’Arge (1974), and Frech (1979). 40. The efficiency critique involved in the entry argument is easily disposed of, as it involves the introduction of transaction costs into the picture. If transaction costs are zero, agents will exploit all potential gains from exchange, which means that the longrun, entry-induced inefficiency will be cured through the same type of bargaining transactions that were employed to resolve the original externality problem (Calabresi, 1968, p. 67). Alternatively, any long-run misallocation will be cured by a single owner who will enter the market in order to exploit the potential for gain (Nutter, 1968). Thus, long-run entry effects do not invalidate the efficiency argument. 41. This is consistent with Barzel’s (1989, p. 2) definition of property rights as ‘the powers to consume, obtain income from, and alienate . . . assets’, and Allen’s (1995, p. 2) definition of an ‘economic property right’ as ‘one’s ability, without penalty, to exercise a choice over a good, service, or person’ (emphasis in original). 42. An open class is defined as one into which entry is unrestricted, while a closed class is one which can be entered only if the right is purchased from a current class member. See also Demsetz (1972b, pp. 229–231). 43. See Harrison & McKee (1985) and Hoffman & Spitzer (1985, 1986). Even so, there was not a significant drop-off in the rate at which efficient bargains were achieved, leading Hoffman & Spitzer (1986, pp. 159–160) to suggest that, rather than calling the Coase theorem into question, these results speak to the robustness of the theorem across alternative hypotheses regarding individual behavior. 44. Under open-range laws, cattlemen are not usually responsible for accidental trespass damage, whereas they are strictly liable under closed range laws. 45. See Mnookin & Kornhauser (1979) and Cooter, Marks & Mnookin (1982). 46. Cheung (1973) finds somewhat similar behavior among apple orchard owners and beekeepers in the state of Washington. 47. See Coase (1960, p. 15; 1981, p. 187; 1988, pp. 174–179). 48. See, for example, Baker (1975); Kelman (1979); Symposium on Efficiency as a Legal Concern (1980); A Response to the Efficiency Symposium (1980); Schlag (1986); Gjerdingen (1986); Johnston (1990); and Crespi (1991).

Educating Alice: Lessons from the Coase Theorem

105

49. But see Baumol (1972) for a more judicious evaluation of Pigouvian remedies. 50. See, for example, Lesser, Dobbs & Zerbe (1996).

REFERENCES Aivazian, V. A., & Callen, J. L. (1981). The Coase theorem and the Empty Core. Journal of Law and Economics, 2(41), 175–81. Aivazian, V. A., Callen, J. L., & Lipnowski, I. (1988). The Coase theorem and Coalitional Stability. Economica, 54(216), 517–20. Alchian, A. A. (1965). Some Economics of Property Rights. Il Politico, 30, 816–29. Allen, D. W. (1991). What are Transaction Costs? Research in Law and Economics, 14, 1–18. Allen, D. W. (1992). Marriage and Divorce: Comment. American Economic Review, 82, 679–685. Allen, D. W. (1995). Property Rights, Transaction Costs, and Coase: One More Time. In: S. G. Medema (Ed.), Coasean Economics. Boston: Kluwer. A Response to the Efficiency Symposium. 1980. Hofstra Law Review, 8. Arrow, K. J. (1969). The Organization of Economic Activity: Issues Pertinent to the Choice of Market versus Nonmarket Allocation. In: The Analysis and Evaluation of Public Expenditures: The PPB System, edited by Joint Economic Committee, 91st Congress of the United States, 1st Session. Washington, DC: US Government Printing Office. Arrow, K. J. (1979). The Property Rights Doctrine and Demand Revelation Under Incomplete Information. In: M. J. Boskin (Ed.), Economics and Human Welfare: Essays in Honor of Tibor Scitovsky (pp. 23–29). New York: Academic Press. Ayres, I., & Talley, I. (1995a). Solomonic Bargaining: Dividing a Legal Entitlement to Facilitate Coasean Trade. Yale Law Journal, 104, 1027–1117. Ayres, I. and Talley, E. (1995b). Distinguishing Between Consensual and Nonconsensual Advantages of Liability Rules. Yale Law Journal, 105(1), 235–53. Baker, C. E. (1975). The Ideology of the Economic Analysis of Law. Philosophy and Public Affairs, 5, 3–48. Barzel, Y. (1985). Transaction Costs: Are They Just Costs? Journal of Institutional and Theoretical Economics, 141(1), 4–16. Barzel, Y. (1989). Economic Analysis of Property Rights. Cambridge, Cambridge University Press. Baumol, W. J. (1972). On Taxation and the Control of Externalities. American Economic Review, 62, 307–322. Baumol, W. J., & Oates, W. E. (1975). The Theory of Environmental Policy. New York: PrenticeHall. Boardman, A., Greenberg, D. H., Vining, A., & Weimar, D. L. (1996). Cost-Benefit Analysis: Concepts and Practice. Upper Saddle River, NJ: Prentice Hall. Boyd, J. H., & Mohring, H. (1971). Externalities: ‘Direct Interaction’ vs. ‘Asset Utilization’ Frameworks. Economica, (n.s.) 38, 347–361. Bramhall, D. F., & Mills, E. S. (1966). A Note on the Asymmetry Between Fees and Payments. Water Resources Research, 2, 615–616. Brinig, M. F., & Buckley, F. H. (1995). Divorce, Legal Rules, and Free Spirits. Mimeo: George Mason University School of Law. Browning, E. K., & Browning, J. M. (1992). Microeconomic Theory and Applications. (4th ed.). New York: HarperCollins.

106

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

Buchanan, J. M. (1986). Rights, Efficiency and Exchange: The Irrelevance of Transaction Costs. In: J. M. Buchanan (Ed.), Liberty, Market and State: Political Economy in the 1980s (pp. 92–107). New York: Buchanan, J. M., & Stubblebine, W. C. (1962). Externality. Economica, 29, 371–384. Calabresi, G. (1961). Some Thoughts on Risk Distribution and the Law of Torts. Yale Law Journal, 70, 499–553. Calabresi, G. (1965). The Decision for Accidents: An Approach to Nonfault Allocation of Costs. Harvard Law Review, 78, 713–745. Calabresi, G. (1968). Transaction Costs, Resource Allocation and Liability Rules: A Comment. Journal of Law and Economics, 11, 67–73. Calabresi, G. (1992). The Pointlessness of Pareto: Carrying Coase Further. Yale Law Journal, 100, 1211–1237. Calabresi, G., & Melamed, D. A. (1972). Property Rules, Liability Rules, and Inalienability: One View of the Cathedral. Harvard Law Review, 85, 1089–1128. Cheung, S. N. S. (1969a). The Theory of Share Tenancy. Chicago: University of Chicago Press. Cheung, S. N. S. (1969b). Transaction Costs, Risk Aversion, and the Choice of Contractual Arrangements. Journal of Law and Economics, 12, 23–42. Cheung, S. N. S. (1973). The Fable of the Bees: An Economic Investigation. Journal of Law and Economics, 16, 11–33. Cheung, S. N. S. (1978). The Myth of Social Costs: A Critique of Welfare Economics and the Implications for Public Policy. London: Institute of Economic Affairs. Cheung, S. N. S. (1980). The Myth of Social Cost. San Francisco: Cato Institute. Cheung, S. N. S. (1992). On the New Institutional Economics. In: L. Werin, & L. and H. Wijkander (Eds), Contract Economics (pp. 48–65). Oxford: Basil Blackwell. Coase, R. H. (1937). The Nature of the Firm. Economica, 4, 386–405. Coase, R. H. (1959). The Federal Communications Commission. Journal of Law and Economics, 2, 1–40. Coase, R. H. (1960). The Problem of Social Cost. Journal of Law and Economics, 3, 1–44. Coase, R. H. (1974). The Lighthouse in Economics. Journal of Law and Economics, 23(2), 357–376. Coase, R. H. (1981). The Coase theorem and the Empty Core: A Comment. Journal of Law and Economics, 24(1), 183–87. Coase, R. H. (1988). The Firm, the Market and the Law. Chicago: University of Chicago Press. Coelho, P. R. P. (1975). Externalities, Separability and Resource Allocation: Comment. American Economic Review, 65, 721–723. Cooter, R. D. (1980). How the Law Circumvents Starrett’s Nonconvexity. Journal of Economic Theory, 22, 499–504. Cooter, R. D. (1982). The Cost of Coase. Journal of Legal Studies 111. 1–33. Cooter, R. D. (1987). The Coase Theorem. In: J. Eatwell, M. Milgate & P. Newman (Eds), The New Palgrave (pp. 457–460). Stockton Press: New York. Cooter, R. D. (1989). The Coase Theorem. In: J. Eatwell, M. Milgate & P. Newman (Eds), The New Palgrave: Allocation, Information, and Markets (pp. 64–70). New York: W. W. Norton. Cooter, R., & Ulen, T. (1988). Law and Economics. Homewood, IL: HarperCollins Publishers. Cooter, R. D., Marks, S., & Mnookin, R. (1982). Bargaining in the Shadow of the Law: A Testable Model of Strategic Behavior. Journal of Legal Studies, 11, 225–251.

Educating Alice: Lessons from the Coase Theorem

107

Coursey, D. L., & Stanley, L. R. (1988). Pretrial Bargaining Behavior within the Shadow of the Law: Theory and Experimental Evidence. International Review of Law and Economics, 8, 161–179. Coursey, D. L., Hoffman, E. and Spitzer, M. L. (1987). Fear and Loathing in the Coase Theorem: Experimental Tests Involving Physical Discomfort. Journal of Legal Studies, 16, 217–248. Crain, W. M., Saurman, D. and Tollison, R. D. (1978). The Coase Theorem and Quasi Rents: Correcting the Record. Public Finance Quarterly, 6(2), 259–62. Crespi, G. S. (1991). The Mid-Life Crisis of the Law and Economics Movement: Confronting the Problems of Nonfalsifiability and Normative Bias. Notre Dame Law Review, 67(2), 231–252. Dahlman, C. J. (1979). The Problem of Externality. Journal of Law and Economics, 22, 141–162. Davis, O. A., & Whinston, A. B. (1962). Externalities, Welfare and the Theory of Games. Journal of Political Economy, 70, 241–262. Davis, O. A. and Whinston, A. B. (1965). Some Notes on Equating Private and Social Cost. Southern Economic Journal, 32, 113–126. de Bornier, J. M. (1986). The Coase Theorem and the Empty Core: A Reexamination. International Review of Law and Economics, 6(2), 265–71. Demsetz, H. (1967). Toward a Theory of Property Rights. American Economic Review, 57, 347–59. Demsetz, H. (1972). When Does the Rule of Liability Matter? Journal of Legal Studies, 1, 13–28. Demsetz, H. (1972b). Wealth Distribution and the Ownership of Rights. Journal of Legal Studies, 1, 223–232. Dolbear, F. T. Jr. (1967). On the Theory of Optimum Externality. American Economic Review, 57, 90–103. Donohue, J. J. III. (1989). Diverting the Coasian River: Incentive Schemes to Reduce Unemployment Spells. Yale Law Journal, 99, 549–609. Editorial Addendum to ‘A Note on Rent and the Coase Theorem’ by David Shapiro. 1977. Journal of Economic Theory, 14, 221–222. Eggertsson, T. (1990). Economic Behavior and Institutions. Cambridge: Cambridge University Press. Ellickson, R. C. (1986). Of Coase and Cattle: Dispute Resolution among Neighbors in Shasta County. Stanford Law Review, 38, 623–687. Ellickson, R. C. (1989). The Case for Coase and Against ‘Coaseanism’. Yale Law Journal, 99, 611–630. Ellickson, R. C. (1991). Order Without Law: How Neighbors Settle Disputes. Cambridge: Harvard University Press. Farrell, J. (1987). Information and the Coase Theorem. Journal of Economic Perspectives, 1(2), 113–29. Feldman, R. (1974). Liability Rules and the Transfer of Economic Rent. Journal of Legal Studies, 3, 499–508. Fischer, S. (1977). Long Term Contracting, Sticky Prices, and Monetary Policy. Journal of Monetary Economics, 3, 317–323. Frank, R. H. (1991). Microeconomics and Behavior. New York: McGraw Hill. Frech, H. E. III. (1973). Pricing of Pollution: The Coase Theorem in the Long Run. Bell Journal of Economics, 4(1), 316–19.

108

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

Frech, H. E. III. (1979). The Extended Coase Theorem and Long Run Equilibrium: The Nonequivalence of Liability Rules and Property Rights. Economic Inquiry, 17(2), 254–68. Friedberg, L. (1995). Did No-Fault Divorce Raise Divorce Rates? Evidence from Panel Data.. Working Paper. Massachusetts Institute of Technology. Gifford, A. Jr., & Stone, C. C. 1973. Externalities, Liability and the Coase Theorem: A Mathematical Analysis. Western Economic Journal, 11(3), 260–69. Gjerdingen, D. H. (1986). The Politics of the Coase Theorem and Its Relationship to Modern Legal Thought. Buffalo Law Review, 35, 871–935. Glanter, M. (1983). Reading the Landscape of Disputes: What We Know and Don’t Know and Think We Don’t Know) about Our Allegedly Contentious and Litigious Society. UCLA Law Review, 31, 4–71. Hanley, N., & Sumner, C. (1995). Bargaining Over Common Property Resources: Applying the Coase Theorem to Red Deer in the Scottish Highlands. Journal of Environmental Management, 43, 87–95. Hanneman, W. M. (1991). Willingness to Pay and Willingness to Accept: How Much Can They Differ? American Economic Review, 81, 635–47. Harris, F. H. deB. (1990). Economic Negligence, Moral Hazard, and the Coase Theorem. Southern Economic Journal, 56(3), 698–704. Harrison, G. W. et al. (1987). Coasian Solutions to the Externality Problem in Experimental Markets. Economic Journal, 97, 388–402. Harrison, G. W., & McKee, M. (1985). Experimental Evaluation of the Coase Theorem. Journal of Law and Economics, 28(3), 653–70. Hoffman, E., & Spitzer, M. L. (1982). The Coase Theorem: Some Experimental Tests. Journal of Law and Economics, 25(1), 73–98. Hoffman, E., & Spitzer, M. L. (1985). Entitlements, Rights, and Fairness: An Experimental Examination of Subjects’ Concepts of Distributive Justice. Journal of Legal Studies, 14, 259–297. Hoffman, E., & Spitzer, M. L. (1986). Experimental Tests of the Coase Theorem with Large Bargaining Groups. Journal of Legal Studies, 15, 149–171. Hoffman, E., & Spitzer, M. L. 1993. Willingness to Pay vs. Willingness to Accept: Legal and Economic Implications. Washington University Law Quarterly, 71, 59–114. Holderness, C. G. (1989). The Assignment of Rights, Entry Effects, and the Allocation of Resources. Journal of Legal Studies, 18, 181–189. Hovenkamp, H. J. (1990). Marginal Utility and the Coase Theorem. Cornell Law Review, 75, 783–810. Illing, G. (1992). Private Information as Transaction Costs: The Coase Theorem Revisited. Journal of Institutional and Theoretical Economics, 148(4), 558–76. Jaffe, J. F. (1975). The ‘Coase Theorem’: A Reexamination – Comment. Quarterly Journal of Economics, 89(4), 660–61. Johnston, J. S. (1990). Law, Economics, and Post-Realist Explanation. Law & Society Review, 24(5), 1217–1254. Jolls, C., Sunstein, C. R., & Thaler, R. (1998). A Behavioral Approach to Law and Economics. Stanford Law Review, 50, 1471–1550. Jung, C., Krutilla, K., Viscusi, W. K., & Boyd, R. (1995). The Coase Theorem in a Rent-Seeking Society. International Review of Law and Economics, 15(3), 259–68. Kahneman, D., & Tversky, A. (1979). Prospect Theory: An Analysis of Decision Under Risk. Econometrica, 47, 263–291.

Educating Alice: Lessons from the Coase Theorem

109

Kahneman, D., Knetsch, J. L., & Thaler, R. H. 1990. Experimental Tests of the Endowment Effect and the Coase Theorem. Journal of Political Economy, 98(6), 1325–48. Kamien, M. I., Schwartz, N. L., & Dolbear, F. T. (1966). Asymmetry Between Bribes and Charges. Water Resources Research, 2, 7–157. Kaplow, L., & Shavell, S. (1995). Do Liability Rules Facilitate Bargaining? A Reply to Ayres and Talley. Yale Law Journal, 105(1), 21–33. Kaplow, L., & Shavell, S. (1996). Property Rules versus Liability Rules: An Economic Analysis. Harvard Law Review, 1094, 13–790. Kelman, M. (1979). Consumption Theory, Production Theory and Ideology in the Coase Theorem. Southern California Law Review, 52, 669–698. Knetsch, J. L. (1997). Reference States, Fairness, and Choice of Measure to Value Environmental Changes. In: M. H. Bazerman et al. (Eds), Environment, Ethics, and Behavior: The Psychology of Environmental Valuation and Degradation. San Francisco: The New Lexington Press. Landa, J. T. (1981). A Theory of the Ethnically Homogeneous Middleman Group: An Institutional Alternative to Contract Law. Journal of Legal Studies, 10, 34–62. Landa, J. T. (1994). Trust, Ethnicity, and Identity: Beyond the New Institutional Economics of Ethnic Trading Networks, Contract Law, and Gift-exchange. Ann Arbor: University of Michigan Press. Lesser, J., Dobbs, D., & Zerbe, R. O. Jr. (1996). Environmental Economics, Boston, Addison Wesley. Levy, D. S., & Friedman, D. (1994). The Revenge of the Redwoods? Reconsidering Property Rights and the Economic Allocation of Natural Resources. University Of Chicago Law Review, 61, 493–526. Lindgren, J. (1990). ‘Ol’ Man River . . . He Keeps Rollin’ Along: A Reply to Donohue’s Diverting the Coasean River. Georgetown Law Journal, 78, 577–591. MacRae, D., & Wilde, J. A. (1985). Policy Analysis for Public Decisions. Lanham, MD: University Press of America. Marchand, J. R., & Russell, K. P. (1973). Externalities, Liability, Separability, and Resource Allocation. American Economic Review, 63, 611–620. Marchand, J. R., & Russell, K. P. (1975). Externalities, Liability, Separability, and Resource Allocation: Reply. American Economic Review, 65, 730–732. Medema, S. G. (1994). R. H. Coase. London: Macmillan. Medema, S. G. (1995). Through A Glass Darkly, or Just Wearing Dark Glasses? Posin, Coase, and the Coase Theorem. Tennessee Law Review, 62, 1041–1056. Medema, S. G. (1996). Of Pangloss, Pigouvians, and Pragmatism: Ronald Coase on Social Cost Analysis. Journal of the History of Economic Thought, 18, 96–114. Medema, S. G. (1996). Comment: The Coase Theorem, Rent Seeking, and the Forgotten Footnote. International Review of Law and Economics, 17, 177–78. Medema, S. G. (1997). The Trial of Homo Economicus: What Law and Economics Tells Us about the Future of Economic Imperialism. New Economics and Its Writing: History of Political Economy Annual Supplement 27, 122–142. Medema, S. G., & Zerbe, R. O. Jr. (2000). The Coase Theorem. In: B. Bouckaet, & G. De Geest (Eds), Encyclopedia of Law and Economics, Aldershot: Edward Elgar Publishing. Mishan, E. J. (1965). Reflections on Recent Developments in the Concept of External Effects. Canadian Journal of Economics and Political Science, 31, 3–34. Mishan, E. J. (1967). Pareto Optimality and the Law. Oxford Economic Papers, 1, 255–287.

110

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

Mishan, E. J. (1971). The Post-war Literature on Externalities: An Interpretative Essay. Journal of Economic Literature, 9, 1–28. Mnookin, R., & Kornhauser, L. (1979). Bargaining in the Shadow of the Law: The Case of Divorce. Yale Law Journal, 88, 950–997. Mumey, G. A. (1971). The Coase Theorem: A Reexamination. Quarterly Journal of Economics, 85(4), 718–23. North, D. (1981). Structure and Change in Economic History. New York: W. W. Norton. Nutter, W. G. (1968). The Coase Theorem on Social Cost: A Footnote. Journal of Law and Economics, 11, 503–507. Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge: Cambridge University Press. Parisi, F. (1995). Private Property and Social Costs. European Journal of Law and Economics, 2, 149–73. Peters, H. E. (1986). Marriage and Divorce: Information Constraints and Private Contracting. American Economic Review, 76, 437–454. Peters, H. E. (1992). Marriage and Divorce: Reply. American Economic Review, 82, 686–693. Pigou, A. C. (1932). The Economics of Welfare (3rd ed.). London: Macmillan. Polinsky, A. M. (1974). Economic Analysis as a Potentially Defective Product: A Buyer’s Guide to Posner’s Economic Analysis of Law. Harvard Law Review, 87, 1655–1681. Polinsky, A. M. (1979). Controlling Externalities and Protecting Entitlements: Property Right, Liability Rule, and Tax-Subsidy Approaches. Journal of Legal Studies, 8, 1–48. Polinsky, A. M. (1980). Resolving Nuisance Disputes: The Simple Economics of Injunctive and Damage Remedies. Stanford Law Review, 32, 1075–1112. Posin, D. Q. (1994). The Coase Theorem: Through a Glass Darkly. Tennessee Law Review, 61, 797–868. Posin, D. Q. (1995). Risk and the Error of the Coase Theorem. Tennessee Law Review, 62(4), 1057–1072. Posner, R. A. (1983). The Economics of Justice. Cambridge: Harvard University Press. Posner, R. A. (1992). Economic Analysis of Law (4th ed). Boston: Little Brown and Company. Regan, D. H. (1972). The Problem of Social Cost Revisited. Journal of Law and Economics, 15, 427–437. Samuels, W. J. (1974). The Coase Theorem and the Study of Law and Economics. Natural Resources Journal, 14(1), 1–33. Samuelson, P. A. (1954). The Pure Theory of Public Expenditure. Review of Economics and Statistics, 36, 386–89. Samuelson, P. A. (1964). Economics: An Introductory Analysis (6th ed.). New York: McGrawHill. Samuelson, P. A. (1966). Modern Economic Realities and Individualism. In: J. E. Stiglitz (Ed.), The Collected Scientific Papers of Paul A. Samuelson (pp. 1407–1418). Cambridge: MIT Press, 1407–1418. Samuelson, W. (1985). A Comment on the Coase Theorem. In: A. E. Roth (Ed.), Game-Theoretic Models of Bargaining (pp. 321–339). Cambridge: Cambridge University Press. Saraydar, E. (1983). Bargaining Power, Dissimulation, and the Coase Theorem. Zeitschrift fur die gesamte Staatswissenschaft, (JITE) 139(4), 599–611. Schlag, P. (1986). An Appreciative Comment on Coase’s The Problem of Social Cost: A View from the Left. Wisconsin Law Review 1986, 919–962. Schlag, P. (1989). The Problem of Transaction Costs. Southern California Law Review, 62, 1661–1699.

Educating Alice: Lessons from the Coase Theorem

111

Schulze, W. D., & d’Arge, R. C. (1974). The Coase Proposition, Information Costs and Long-run Equilibrium. American Economic Review, 64, 763–772. Schwab, S. J. (1987). Collective Bargaining and the Coase Theorem. Cornell Law Review, 72, 245–287. Schwab, S. J. (1988). A Coasean Experiment on Contract Presumptions. Journal of Legal Studies, 17, 237–268. Schweizer, U. (1988). Externalities and the Coase Theorem: Hypothesis or Result? Journal of Institutional and Theoretical Economics, 144, 245–266. Shapiro, D. L. (1974). A Note on Rent and the Coase Theorem. Journal of Economic Theory, 7(1), 125–28. Shapiro, D. L. (1978). The Coase Theorem and Quasi Rents: Correcting the Record: Reply. Public Finance Quarterly, 62, 262–64. Shoup, D. C. (1971). Theoretical Efficiency in Pollution Control: Comment. Western Economic Journal, 9, 310–313. Spitzer, M. L., & Hoffman, E. (1980). A Reply to Consumption Theory, Production Theory and Ideology in the Coase Theorem. Southern California Law Review, 53, 1187–1223. Starrett, D. A. (1972). Fundamental Nonconvexities in the Theory of Externalities. Journal of Economic Theory, 4, 180–199. Stigler, G. J. (1966). The Theory of Price (3rd ed.). New York: Macmillan. Sumner, C. (1993). Red Deer Management Problems in Strathyre Forest: Ecology, Economics and Land Use, M.Sc. dissertation, Department of Environmental Science. Stirling: University of Stirling. Sutton, J. (1986). Noncooperative Bargaining Theory: An Introduction. Review of Economic Studies, 53, 709–724. Symposium on Efficiency as a Legal Concern. Hofstra Law Review, 8, 1980. Thaler, R. (1980). Toward a Positive Theory of Consumer Choice. Journal of Economic Behavior and Organization, 1, 39–60. Turvey, R. (1963). On Divergences Between Social Cost and Private Cost. Economica, 30, 309–313. Tybout, R. A. (1972). Pricing Pollution and Other Negative Externalities. Bell Journal of Economics, 3, 252–266. Tybout, R. A. (1973). Pricing of Pollution: The Coase Theorem in the Long Run: Reply. Bell Journal of Economics, 4(1), 320–21. Veljanovski, C. G. (1977). The Coase Theorem: The Say’s Law of Welfare Economics? Economic Record, 53, 535–541. Veljanovski, C. G. (1982). The Coase Theorems and the Economic Theory of Markets and Law. Kyklos, 35, 53–74. Vogel, K. R. (1987). The Coase Theorem and California Animal Trespass Law. Journal of Legal Studies, 167, 149–187. Weimer, D. L., & Vining, A. R. (1992). Policy Analysis: Concepts and Practice, 2nd edition. Englewood Cliffs, NJ: Prentice Hall. Wellisz, S. (1964). On External Diseconomies and the Government-Assisted Invisible Hand. Economica, 31, 345–362. Willig, R. D. (1976). Consumer’s Surplus Without Apology. American Economic Review, 66, 589–597. Woodbury, S. A., & Spiegelman, R. G. (1987). Bonuses to Workers and Employers to Reduce Unemployment: Randomized Trials in Illinois. American Economic Review, 77, 513–530.

112

STEVEN G. MEDEMA & RICHARD O. ZERBE JR.

Zelder, M. (1993a). The Economic Analysis of the Effect of No-Fault Divorce Law on the Divorce Rate. Harvard Journal of Law and Public Policy, 16, 241–267. Zelder, M. (1993b). Inefficient Dissolutions as a Consequence of Public Goods: The Case of NoFault Divorce. Journal of Legal Studies, 22, 503–520. Zerbe, R. O. Jr. (1976). The Problem of Social Cost: Fifteen Years Later. In: S. A. Y. Lin (Ed.), Theory and Measurement of Economic Externalities (pp. 29–36). New York: Academic Press. Zerbe, R. O. Jr. (1980). The Problem of Social Cost in Retrospect. Research in Law and Economics, 2, 83–102. Zerbe, R. O. Jr. (1998). Is Benefit Cost Analysis Legal? Journal of Policy Analysis and Management, 17(3), 419–456. Zerbe, R. O. Jr., & McCurdy, H. (1999). The Failure of Market Failure. Policy Analysis Management, 18(4), 558–578. Zerbe, R. O. Jr., & Urban, N. (1988). Including the Public Interest in Theories of Regulation. Research in Law and Economics, 11, 1–23.

BARS TO THE PAYMENT OF PREJUDGMENT INTEREST: THE SUPREME COURT AND THE FEDERAL EMPLOYERS’ LIABILITY ACT David E. Ault and Gilbert L. Rutman In Monesson Southwestern Railway Co. v. Morgan,1 the U.S. Supreme Court, hereafter referred to as the Court, stated clearly that prejudgment interest was not permitted in cases brought under the Federal Employers’ Liability Act (FELA). The Court’s opinion rests upon its interpretation of common law on personal injury cases and Congressional intent when FELA was first enacted. In its opinion, the Court states that common law did not permit the payment of prejudgment interest in personal injury cases at the time that FELA became law. Further, had Congress wished to do so, it could have done so by explicitly permitting the payment of prejudgment interest in the legislation or in any of the subsequent amendments to the original 1908 law. The fact that the law is silent on this matter indicates that Congress did not wish to make the payment of prejudgment interest possible when awarding damages in FELA cases. The effect of the Court’s refusal to permit the payment of prejudgment interest, in effect, creates an unrequited loan that is made by the plaintiff to the defendant. Regardless of the amount of time that lapses between the date upon which the plaintiff’s damages occurred and the date upon which judgment as to damages is awarded, the Court does not permit the payment of interest to compensate for the loss of the use of these funds by the plaintiff during the prejudgment period. The Court, however, has never made it clear as to the

Research in Law and Economics, Volume 19, pages 113–137. Copyright © 2000 by JAI/Elsevier Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0308-5

113

114

DAVID E. AULT & GILBERT L. RUTMAN

interest concept that is being considered in these cases. Is the Court referring to nominal rates of interest or ‘real’ rates of interest, that is, nominal rates of interest adjusted for expected inflation? In this paper, we will examine the Supreme Court’s arguments with respect to the payment of prejudgment interest within the context of precedents established in U.S. common law. FELA addresses the recovery of damages as the result of injuries incurred while participating in interstate commerce. While the law itself deals only with injuries incurred while working on the railroads, the law and the body of decisions that have resulted from the interpretation of the law have influenced the entire body of law regarding the recovery of damages as the result of personal injury or death.2 While FELA was enacted in 1908 and involved railroads, at that time one of the most important, if not the most important, employer in the U.S., the law remains one of the most important guides to the recovery of damages resulting from most forms of personal injuries. Recent precedent decisions, for example, have involved the calculation of damage awards when retirement benefits and contributions are not directly tied to one another.3 The Court’s position that prejudgment interest is prohibited in FELA cases rests on the nature of common law and an understanding of the development of the concepts of tort, negligence, ‘nominalism’, and usury. The transition of the U.S. economy from one based primarily upon agriculture to one based on its industrial sector also appears to play a significant role in the Court’s decisions with respect to the issue of the payment of interest in damage cases. This paper begins with a discussion of the Federal Employers’ Liability Act, analyzes the Supreme Court’s position on prejudgment interest in light of common law and the role of nominalism in personal injury cases, and concludes with some remarks on the rationale for this exclusion.

THE FELA The Federal Employers’ Liability Act was enacted in 1908. The act repealed the provisions of the Masters and Servants Acts and the body of common law associated with these acts and restructured the conditions under which damages could be awarded as the result of death or injury while working on a railroad.4 At the time the law was enacted, Congress could, had it desired to do so, have permitted the payment of prejudgment interest. The Court has argued that common law prohibited the payment of prejudgment interest at the time FELA was passed; and the fact that other aspects related to common law were changed is an indication that Congress never intended for prejudgment interest to be paid. The Court’s position that Congress never intended for prejudgment

Bars to the Payment of Prejudgment Interest

115

interest to be paid as part of damage awards is further strengthened by the fact that Congress, when amending FELA in 1910 and 1939, made no attempt to address the issue of prejudgment interest.5 In the same year that FELA became law, Congress created legislation that established a Workers’ Compensation system for federal employees.6 The essential difference between the two systems is that, under workers’ compensation, the individual was not required to prove employer negligence to receive compensation. This ‘no fault’ approach provided predetermined compensation for injuries regardless of who was at fault and whether or not claims arose as a result of negligence on the part of the employee.7 Under workers’ compensation systems, employers are required to provide insurance through either a self-insurance program, a private carrier, or a stateprovided program. While workers’ compensation programs vary among the states, two features are common to all: (1) limits are imposed upon the types of damages that are covered, and (2) benefits to be paid are prescribed explicitly.8 A plausible explanation for the decision by Congress to place the railroad industry under a negligence-based, trial-driven court system rests in the special relationship that Congress has always had with the railroad industry. This special relationship had its beginnings in the 1880s with the creation of the Interstate Commerce Commission and its power to regulate interstate commerce. At the time that FELA was enacted, the railroad industry was: . . . the premier industry in the United States and working for a railroad was especially hazardous. The enactment of FELA strengthened the railroad workers’ rights to sue their employers for all damages related to on-the-job injuries without limit or restriction . . . . To collect the injured worker has to demonstrate negligence on the part of the employer, and awards can be reduced depending on the degree of employee negligence through a ‘comparative negligence’ standard. An injured worker can receive no compensation at all if the injury is judged to be entirely the worker’s fault.9

The data presented in Table 1 support this view of the hazardous nature of railroad work. Between 1890 and 1916, the accident rate increased from 3.0% to a peak of 10.4%, an increase of nearly 350%. After 1916, the accident rate declined until it reached 1.7% in 1939, the year in which FELA was last amended. The hazardous nature of railroad work in the late nineteenth and early twentieth-centuries is even more apparent when one examines the number of accidents between 1890 and 1939. In 1890, there were 22,396 accidents reported. The number rose steadily until 1916 when 176,923 accidents were reported, an increase of nearly 700% in 26 years. The relatively greater increase

116

DAVID E. AULT & GILBERT L. RUTMAN

Table 1. Accidents and Deaths Among Railroad Employees Year

Number of Employees

Number of Accidents

Accident Rate (Percent)

Number of Deaths

1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931

749,000 784,000 821,000 874,000 780,000 785,000 827,000 823,000 875,000 929,000 1,018,000 1,071,000 1,189,000 1,313,000 1,296,000 1,382,000 1,521,000 1,672,000 1,436,000 1,503,000 1,699,000 1,670,000 1,716,000 1,815,000 1,710,000 1,548,000 1,701,000 1,786,000 1,892,000 1,960,000 2,076,000 1,705,000 1,670,000 1,902,000 1,795,000 1,786,000 1,822,000 1,776,000 1,692,000 1,694,000 1,517,000 1,283,000

22,396 26,140 28,267 31,729 23,422 25,696 29,969 27,667 31,761 34,923 39,643 41,142 50,524 60,481 67,067 66,833 76,701 87,644 82,487 75,006 95,671 126,039 142,442 171,417 165,212 138,092 176,923 174,247 156,013 131,018 149,414 104,530 117,197 152,678 125,319 119,224 111,903 88,223 70,873 60,739 35,872 23,358

3.0 3.3 3.4 3.6 3.0 3.3 3.6 3.4 3.6 3.8 3.9 3.8 4.2 4.6 5.2 4.8 5.0 5.2 5.7 5.0 5.6 7.5 8.3 9.4 9.7 8.9 10.4 9.8 8.2 6.7 7.2 6.1 7.0 8.0 7.0 6.7 6.1 5.0 4.2 3.6 2.4 1.8

2,451 2,660 2,554 2,727 1,823 1,811 1,861 1,693 1,958 2,210 2,550 2,675 2,969 3,608 3,632 3,361 3,929 4,534 3,405 2,610 3,382 3,602 3,635 3,715 3,259 2,152 2,941 3,199 3,419 2,138 2,578 1,446 1,657 2,026 1,543 1,599 1,672 1,576 1,329 1,428 977 677

117

Bars to the Payment of Prejudgment Interest

Table 1. Continued Year

Number of Employees

Number of Accidents

Accident Rate (Percent)

1932 1933 1934 1935 1936 1937 1938 1939 1940

1,052,000 991,000 1,027,000 1,014,000 1,086,000 1,137,000 958,000 1,007,000 1,046,000

17,742 15,932 17,338 16,742 22,409 24,114 16,569 17,383 18,350

1.7 1.6 1.7 1.7 2.1 2.1 1.7 1.7 1.8

Number of Deaths 579 533 556 600 720 712 513 536 583

in the number of accidents, as compared to the accident rate, is due to the significant increase in the number of railroad employees between 1890 (749,000) and 1920 (2,076,000), the year in which railroad employment peaked. This is an increase of 277%. In the five years (1904–8) prior to the enactment of FELA, 18,861 railroad employees were killed while working. The number of railroad employees killed on the job peaked in 1907 at 4,534 and continued to average more than 3,000 per year until 1918 when the number of accidents began to decline significantly. The most frequent and serious injuries were related to boiler explosions and to other accidents that occurred in the railyards. These accidents included falling from cars as the result of loose handgrips, poorly attached ladders, and slippery surfaces, especially in wet and wintry weather.10 During the period 1904–14, 39,064 workers lost their lives while employed in the railroad industry. By 1939, the number of workers killed while on the job had declined almost 722% to 536 from 1907, the peak year for number of deaths and the year before FELA was enacted. By the latter part of the twentieth-century, in the period 1981–90, the accident rate in the railroad industry was below that of 10 other major industries: construction, trucking, air transportation, water transportation, agriculture, manufacturing, local transit, mining, other transportation, and private industry.11 Congress was aware of the hazardous nature of railroad employment when debating the passage of FELA and clearly expected the passage of FELA would lead to a reduction in the number of accidents and the accident rate.12 The effect of Congressional action was to pressure the railroad industry into

118

DAVID E. AULT & GILBERT L. RUTMAN

creating a safer work environment by establishing a negligence-based, trialdriven tort system rather than a no-fault workers’ compensation system. The amount of damages could be tailored specifically to the nature of the worker’s injuries and the circumstances that led to those injuries. The evidence summarized in Table 1 indicates that accidents and deaths began to decline immediately after the passage of FELA.13 While direct causation between the passage of FELA and the decline in the accident rate and the number of employment-related deaths of railroad employees is only suggested by the data, the Transportation Research Board concluded that the FELA process generally provides a higher and a broader range of benefits to injured workers than workers’ compensation programs.14 The higher awards made possible through litigation in a fault system act as an incentive to provide a safer work environment.

THE SUPREME COURT, COMMON LAW, AND PREJUDGMENT INTEREST In arriving at its decision in Monessen, the Court cited the work of C.T. McCormick & T. Sedgwick15 as the foundation for its opinion that common law prohibits the payment of prejudgment interest. Each devotes a chapter solely to the issues related to the payment of prejudgment interest in legal disputes. In neither of these chapters is there a definitive statement concerning the payment of prejudgment interest in cases of tort. In Laycock v. Parker,16 for example, the Wisconsin Supreme Court stated, ‘. . . it may be safely said that the tendency has been in favor of allowing interest rather than against it, and that the degree of certainty of ease with which the approximate amount can be ascertained has grown less and less stringent’.17 In Depalma v. Weinman and Pierce v. Lehigh Valley,18 the Supreme Courts of New Mexico and Pennsylvania argued that it was at the discretion of the jury as to whether or not prejudgment interest should be paid in cases of tort.19 More specifically, in Pierce, the Pennsylvania Supreme Court stated: The right to compensation for delay in the payment of damages arising out of tort depends upon the circumstances of the case. It is therefore usually a question for the jury under the evidence submitted. If the fault of nonpayment of the claim rests with the defendant, he cannot complain if he is required to compensate for the delay. If, on the other hand, the fault lies with the plaintiff by reason and excessive and unconscionable demand, one which the defendant is required to protect himself against by litigation, he should not be penalized for the unwarranted conduct of the plaintiff, and required to pay damages for the delay in the settlement of the claim.20

Bars to the Payment of Prejudgment Interest

119

McCormick did, however, distinguish between liquidated and unliquidated damages. In Cochran v. City of Boston,21 the court noted that ‘. . . where it was held that in an action for injury to property the jury in their [sic] discretion, and as incident to determining the amount of the real loss, may consider the delay caused by the defendant’.22 The court went on to state that injury to property was different than injury to persons. In personal injury cases, the damages continue to accrue after the date upon which the injury took place. The extent of the injury may not be determined until sometime after the accident or incident took place. Further, the permanence of personal injuries may not be known until some time has passed. Events that occur after the actual date upon which the injury occurred may be essential to the determination of the damages that resulted.23 In cases of personal injury, the court’s opinion rests upon the calculation of damages. The amount that the negligent party is required to pay for pain and suffering, lost earnings, or expenses does not become a ‘definitive obligation’ until a verdict has been finally entered.24 Regardless of the merits of the argument, McCormick’s presentation does not support the conclusion that common law prohibits the payment of prejudgment interest in cases of personal injury. In Sedgwick’s chapter on interest, he quotes extensively from the opinions and writings of Lords Mansfield and Ellenborough. In general, Mansfield and Ellenborough note that, when money is due without a definite time of payment, and no contract states either implicitly or explicitly that interest should be paid, the precedent in English common law is that interest should not be paid.25 Mansfield further notes that before the reign of Henry VIII in the sixteenthcentury, all payment of interest on debt was prohibited by common law.26 Similar to McCormick, Sedgwick notes that the rule prohibiting interest on debt was modified in the United States early in the twentieth-century.27 He goes on to state that the exceptions under which interest could be paid on damages were gradually being extended. The first exception was the payment of interest when specified by contracts. The courts then extended this exception by permitting the jury to determine when to award interest on damages. The next step was to apply the exception to an increasing number of cases.28 Sedgwick, like McCormick, traces the extension of the rule regarding the payment of interest on damages to cases of tort in which the demand is unliquidated.29 He argues that, ‘It sufficiently appears from what has already been said that there is no general principle which prevents the recovery of interest in actions of tort. In fact that the demand is unliquidated has been shown to be insufficient to exclude interest’.30

120

DAVID E. AULT & GILBERT L. RUTMAN

In actions to recover damages for personal injury, Sedgwick argues explicitly and directly in unequivocal language that the payment of prejudgment interest should be excluded: ‘It is certainly not allowed in such actions as . . . for personal injury by negligence’.31It is equally clear that Sedgwick’s exclusion is based on a principle of ‘fairness’ to the defendant. The confusion over the payment of prejudgment interest appears to lie in the nature of personal injury torts, for the judge’s decision must award a specific sum and determine a date prior to judgment from which interest is to accrue.32 It is not that United States common law in the early twentieth-century prohibited prejudgment interest, but rather that a definitive statement as to how to compute such interest could not be made fairly.

THE ASSESSMENT OF DAMAGES IN COMMON LAW: THE IMPORTANCE OF PREJUDGMENT INTEREST The body of common law gives rights to individuals and offers remedies for any violation of those rights. Whenever common law finds in favor of the plaintiff, there is a remedy in the form of compensation for damages associated with the injury. The object of such awards is to compensate the plaintiff for any monetary losses, such that the plaintiff is returned, to the extent that monetary damage awards can, to the state that he occupied prior to the tort.33 The principle that requires compensation for damages to be equal to the pecuniary loss sustained by the plaintiff as a result of action by the defendant is unique to English common law. The principle that the plaintiff can recover only what he has lost does not permit arbitrary awards, such as those under workers’ compensation where the loss of a finger or other limb is set regardless of the circumstances under which the loss occurred or its effect upon the injured party. Nor is the matter left solely to the discretion of the judge as is permitted under Code Napoleon.34 The central principle of common law with respect to damages is that the individual is to be made whole. Although this principle has remained unchanged, the rules for assessing damages have been refined, examined, and changed as decisions have been handed down from one generation of judges to the next. The rules for assigning responsibility and assessing damages have changed as the socio-political-economic system has changed. Judges, while adhering to the precedents of their eras as the basis for most decisions, do respond to the pressures of changes in the patterns of litigation.35 The development of common law with respect to actions of tort is, however, anything but linear. Charles Gregory notes that the law student who searches for a body of literature to predict the outcomes in tort cases, ‘. . . must be

Bars to the Payment of Prejudgment Interest

121

impressed by the unintellectual nature of legal thinking, by the repeated inconsistencies occurring at the principle and theory level, and by the shabbiness of explanation given for decisions in judicial opinions’.36 With respect to the development of a consistent theory of liability, he goes on to state: ‘the result makes the traditional seamless web of the common laws look like a patched and moth-eaten old mantle. Naturally, it is impossible to read these cases and retain the illusion that judges do not make and unmake law’.37 The development of actions of tort in cases involving negligence is a relatively modern phenomenon. In early English common law, there was no action for negligence. S.F. C. Milsom believes that the concept of torts did not really appear until the eighteenth century.38 Richard Posner states that negligence did not become the dominant standard in civil liability cases in the United States until the beginning of the nineteenth century39 while Lawrence Friedman argues that the law of torts was insignificant before 1900.40 Actions for negligence seem to have originated in the common law concept of trespass. By the twelfth-century, English common law recognized protection for private property. Damages could be sought as compensation for wrong that others had committed against his person or property.41 In cases of trespass, a person was held strictly liable for his actions. Sir W. Holdsworth notes: The general rule is that a man is liable for the harm which he has inflicted upon another by his acts, if what he has done comes within some one of the forms of actions provided by the law, whether that harm has been inflicted intentionally, negligently, or accidentally. In adjudicating upon questions of civil liability the law makes no attempt to try the intent of a man, and the conception of negligence has as yet hardly arisen. A man acts at his peril.42

During the thirteenth and fourteenth-centuries, the concept of trespass, which involved only those incidents of direct contact between plaintiffs and defendants, was expanded to include trespass in a similar case or action that involved damages that occurred consequentially from the actions of the defendant where the defendant and the plaintiff had no direct contact.43 There is a presumption that, during Middle Ages, common law prohibited the payment of prejudgment interest in cases of trespass.44 McCormick notes that, in medieval Europe, interest was called usury, which was not only not enforceable if included in a contract but subject to criminal action if one party sought to collect such payments. The rationale for this prohibition appears to be that those with wealth should not attempt to extract additional payments from the unfortunate.45 Given the economic conditions and nature of contracts in the Middle Ages, it is quite possible that common law prohibited interest payments because such payments could be abused.

122

DAVID E. AULT & GILBERT L. RUTMAN

At the same time, in the feudal economies of the Middle Ages, where damages were set by custom, the payment of prejudgment interest may never have been an issue when awarding damages in personal injury cases. Rather than barring prejudgment interest, the courts could have been silent on the issue because it never arose. In England, unless an issue such as the payment of prejudgment interest in cases of damages had been addressed through legislation enacted by Parliament, the issue would never have been addressed unless it was the subject of litigation. English common law courts do not issue hypothetical opinions. The courts confine themselves to settling actual disputes. All important questions do not, however, result in litigation and, under common law, new law can be made only when a dispute related to the question is litigated.46 During the thirteenth and fourteenth-centuries in England, the interval between the date upon which the tort occurred and judgment was rendered was quite short. The issue of prejudgment interest may, therefore, have been unimportant. In 1272, the Common Pleas Court received its own chief justice and was separated completely from the other courts.47 As a result, common law courts were royal courts under the jurisdiction of the King who often sat as the presiding judge. This practice lasted until the fifteenth-century. The judges of these courts were, in effect, royal justices.48 At this stage in the development of common law, there was no adequate appeals process, a process that has played a prominent role in the recent development of common law. This lack of a well-defined appeals process persisted until 1875.49 The appeals process evolved over time. By the fourteenth-century, the Common Pleas Court clearly recognized that it had the power and authority to order a new trial if there were irregularities in a lower court. The number of cases, however, in which such action was taken appears to have been small until the seventeenth-century.50 The lack of an adequate appeals process in medieval England is made even more evident when one examines the operations of the manorial courts, which handled the largest volume of cases. Cases brought before these courts involved issues, such as unwarranted fishing in the manor’s ponds, allowing too many head of cattle to graze on the estate, or selling sour beer.51 With respect to the right of appeal to the King’s Court, Sir F. Pollock and F.W. Maitland note: ‘The right of appeal to King’s Court existed but was restricted to an appeal or complaint of false judgment. If upheld, . . . a lord may lose forever the right to hold court, a penalty that most assuredly reduced the number of complaints or appeals’.52

Bars to the Payment of Prejudgment Interest

123

Between the seventeenth and eighteenth-centuries, common law in the United States evolved with respect to the manner in which fault was assigned in injury cases. The defendant no longer was absolutely liable for his actions. Negligence on the part of the plaintiff could lead to a complete bar to the recovery of damages.53 The motivating force behind this fundamental change in actions related to damages appears to have been the fear of impeding the Industrial Revolution. Judges learned that lawsuits and damage awards could diminish the rate of economic growth which, at the time, was the highest priority given its potential for the greater good of society.54 Within the promise that the Industrial Revolution would promote the general welfare, in the United States, the railroad was considered to be the premier industry that led the way. The railroad was literally the Industrial Revolution’s engine of growth and change: In the first generation of tort law, the railroad was the prince of machines, both as symbol and as fact. It was the key to economic development . . . . It bound cities together, and tied the farms to the cities and seaports. Yet . . . boilers exploded; trains hurtled off tracks; bridges collapsed; locomotives collided in a grinding scream of steel. Railroad law and tort law grew up, then, together. In a sense, the two were the same.55

In the nineteenth-century, judges responded to the economic sentiment of the day in their assigning of fault and damages. While the Industrial Revolution began in England, the pursuit of economic growth had ascended to the highest priority in the United States. Absolute liability was rejected because its application could stunt economic growth. The capital in any enterprise, including the railroads, could be drained if all damages caused by an ‘accident’ could be recovered through litigation. The judges, therefore, sought to mitigate the adverse effect that damage awards could have upon the capital available to fuel economic growth and expansion.56 Given the history of common law’s treatment of interest and the relationship between Congress and the railroads, why did the Supreme Court of the United States interpret Congressional silence on the payment of prejudgment interest as a prohibition against such awards even after the decline in the relative importance of the railroad industry? Posner has argued that common law evolves as economic conditions change.57 Why, then, has the United States Supreme Court held steadfastly for 87 years that common law prohibits the payment of prejudgment interest even after the pro-business sentiment of the nineteenth-century had waned? Did the opinions of the Court in the early twentieth-century reflect the pro-business sentiment of the era? Is the Court’s position that the only remedy for this prohibition is further legislation explained by a judicial bias in favor of nominalism?

124

DAVID E. AULT & GILBERT L. RUTMAN

NOMINALISM AND THE COMMON LAW McCormick provides a simple explanation for the Court’s behavior: The question of interest is one much more often passed upon than carefully considered by courts. It is usually presented only incidentally to much more important issues, and often decided one way or the other at the close of exhaustive investigation of other questions, and with the perhaps unconscious feeling that it is not of sufficient magnitude to justify further serious labor.58

Another plausible explanation for the failure of the Court to change its stance on this issue is the principle of nominalism that characterizes English common law and is the basis for common law in the United States. A bias towards nominalism would add to the caution of the Court in allowing prejudgment interest because a nominal interest rate includes a return for expected inflation. Nominalism refers to the expression of value in terms of current prices regardless of the changes in purchasing power over time. This conviction in English common law that a ‘pound is a pound’ appears to be so firmly rooted in England that no one has questioned the concept.59 Nineteenth-century English attempts to address the issue of nominalism by legislation and litigation were unsuccessful. The principle that a debt expressed in pounds was to be repaid in the same number of pounds regardless of the rate of inflation between the date that the debt was incurred and the date it was repaid was so strong that there was no rule or procedure that would permit this concept to be revised.60 In the twentieth-century, changing economic conditions in England and the United States did lead to modification of the principle of nominalism. In 1945, English judges, in a Scottish case,61 stated that the courts were not bound by this principle in awarding damages. The judges argued that changes in the ‘value of money’, that is, its purchasing power, should be taken into consideration when measuring damages.62 In 1948, English common law included compensation for the loss of purchasing power in awarding damages in a personal injury case.63 In the United States, compensation for the effects of inflation on purchasing power began to evolve in the mid-nineteenth-century. According to J.P. Dawson & F.E. Cooper,64 the principal case in which the Supreme Court recognized that applying the principle of nominalism would lead to an inequitable outcome was Willard v. Taylor.65 In assessing the importance of the opinion in this case, Dawson & Cooper argue that it is the leading case in the United States on the subject of specific performance. This case has been cited as the basis for opinions, not only with respect to specific performance, but also for its use of the conditional decree, a means of revising prices indirectly.66

Bars to the Payment of Prejudgment Interest

125

In Willard v. Taylor, the Court recognized the difference between relative price changes and absolute increases in the price level which affect purchasing power. The Court attempted to take into account the effects of inflation on purchasing power as well as provide relief for the effects of inflation on the value of property. In cases, however, where changes in the value of property were due to changes in supply relative to demand, no relief was granted.67 During the latter part of the nineteenth-century, the courts did not follow the reasoning set forth in Willard with respect to the granting of relief from the effects of inflation. Dawson & Cooper argue that the reason for this behavior was the difficulty that the courts had in separating the change in the value of property caused by inflation from changes in value caused by market conditions.68 Dawson & Cooper, however, assert: The doctrine announced in Willard v. Taylor still survives. It is still true that an unforeseen change in circumstances, defeating the purposes for which the contract was made or destroying the intended proportion between price and other performance, can lead to a refusal of specific performance or a ‘conscientious modification’ of the contractual terms.69

Another explanation for the courts not granting relief from the effects of inflation is that the issue was irrelevant in the latter part of the nineteenthcentury. Between 1870 and 1900, there was price deflation in 21 of the 31 years in this period. Inflation occurred in only 10 years; however, it was less than 2.0% in six and less than 3.0% in three of these 10 years.70 During the twentieth-century, the United States courts began to allow for the effects of inflation by assessing damages in tort actions for personal injury. When decisions resulted in awards that were deemed excessive, the appellate courts ruled that judges could take into consideration the fact that the purchasing power of money had declined. Verdicts with regard to damages in periods of the nineteenth and early twentieth-centuries characterized by little or no inflation, therefore, were not necessarily valid guidelines for determining appropriate damages in the WWI and post-WWI periods.71 There is little doubt, however, that, by the first third of the twentieth-century, United States courts recognized the significance of the effects of inflation on damage awards. The problem lay in measuring the effects of these changes on the size of damage awards. One specific problem arose in cases where estimates of an individual’s earning power or capacity could have been affected by changes in purchasing power of money. If, for example, a person’s injuries occurred in 1985 and the case came to trial in 1995, what is the proper method for determining the individual’s earning capacity in 1995? Dawson & Cooper refer to such problems as ‘inconveniences’ which did not alter the position of the courts that a change in the value of money had occurred and that an

126

DAVID E. AULT & GILBERT L. RUTMAN

appropriate remedy must take this fact into account.72 Based on this analysis, there is little reason to believe that the Supreme Court’s decision in Monessen to disallow the payment of prejudgment interest was based on the principle of nominalism.

THE NATURE OF THE DEMAND AND PREJUDGMENT INTEREST The inconveniences discussed by Dawson & Cooper appear to be particularly vexing in the area of unliquidated demands. In cases in which the claim is liquidated, courts appear to have fewer problems determining the exact date upon which the total debt can be established. In such cases, the defendant can be charged interest, that is, prejudgment interest on the debt. Because a liquidated claim represents an expressed contract to pay a certain sum on a given date or within a given period of time, both parties know how much and when it is to be paid. The court can, therefore, determine the sum of money, the use of which the plaintiff was deprived, and the exact period over which that use was denied.73 While McCormick argues that measurement problems prevent the payment of prejudgment interest on an unliquidated claim, such as that which arises in a personal injury case, Sedgwick is more flexible.74 He does not divide claims into liquidated and unliquidated demands arguing that there is no reason why an unliquidated claim should exempt the defendant from paying interest. His review of cases does not indicate that the existence of unliquidated demands clearly demarcates those cases in which prejudgment interest is barred from those cases in which it is granted.75 Sedgwick arrayed cases along a continuum using two criteria to determine if prejudgment interest was warranted. The first criterion was whether or not the claim was of such nature that an exact and specific amount of money could be determined by relatively simple calculations or access to readily obtainable information, such as a market price. The second criterion was that a definite time period must have been established over which interest was to be paid.76 Sedgwick is very clear that common law does not prohibit the payment of prejudgment interest in actions of tort or unliquidated damages. Sedgwick, quoting from McMahon v. New York and Erie Railroad, notes that, ‘The old common law rule which required that a demand be liquidated, or its amount in some way ascertained before interest could be allowed, has been modified by general consent, so as to hold that if the amount is capable of being ascertained by mere computation, then it shall carry interest’.77 Sedgwick further notes that, in Van Renseelaer v. Jewett, the United States Second Circuit Court of Appeals

Bars to the Payment of Prejudgment Interest

127

specifically permitted prejudgment interest on an unliquidated claim because the claim met the two criteria that he had outlined.78 In personal injury cases, Sedgwick, similar to McCormick’s view on the matter, argued that in tort actions, for personal injury because of negligence, the payment of prejudgment interest should not be allowed as part of damages. His rationale for this position was not that prejudgment interest was prohibited by common law in such cases but that the payment of prejudgment interest did not meet the criteria that he had established.79 This is contrary to the Supreme Court’s opinion in Monessen, where the majority opinion did not permit the payment of prejudgment interest on the grounds that common law did not permit such payments. The Court used Sedgwick and McCormick as sources to support this position.

NOMINALISM, PREJUDGMENT INTEREST, AND DAMAGES UNDER FELA In Jones and Laughlin v. Pfeifer,80 the Court dealt directly with the principle of nominalism when estimating damages in FELA cases, as well as the appropriate interest rate to be used when discounting future damages to present value. In examining the methodology to be used when calculating future damages, the Court took, as its point of departure, a hypothetical, inflation-free economy. In so doing, the Court recognized explicitly that inflation alters the estimates of future losses, such as earnings, in a manner that must be recognized when calculating the value of damages.81 In Pfeifer, the Court addressed the inflation issue by discussing the criteria for selecting an appropriate rate of interest to be used as the discount factor. The Court further noted that the use of a market, or nominal rate of interest, would create an inequity that benefited the defendant. If the plaintiff were to invest his award at a market rate of interest, ‘. . . the stream of income that he received would provide him with only enough dollars to maintain his existing nominal income; it did not provide him with a stream comparable to what his lost wages would have been in an inflationary economy’.82 The Court went on to note that there was a consensus among the courts that, given the present rate of inflation, this inequity could not be tolerated.83 In order to estimate damages that are adjusted for the effects of inflation on purchasing power, nominal interest rates must be adjusted to provide estimates of the ‘real’ rate of interest. The real rate of interest is defined as: R = i ⫺ P where i = the market or nominal rate of interest; and P = the rate of inflation. The real rate of interest is, therefore, the difference between the market, or

128

DAVID E. AULT & GILBERT L. RUTMAN

nominal rate, and the rate of change in the price level, that is, the rate of inflation. The use of the real rate of interest as the discount factor avoids unwarranted speculation about the future rate of inflation. Estimates of the real rate have consistently produced figures of about 2%.84 In Pfeifer, the Court noted that any discount factor between one and three % would, depending upon the rationale, be acceptable.85 In O’Shea v. Riverway Towing Co.,86 Posner seemed to suggest a real rate of 2% as acceptable. The Court, in examining the procedures to be used in calculating the present value of lost future earnings, appears to be concerned with the effect of the loss of purchasing power created over time by inflation on the plaintiff’s standard of living. The use of the real rate of interest when calculating the present value of a future stream of earnings or damages would permit the Court to set an award that would offset the effects of future inflation on the plaintiff’s standard of living. The damage award would be set in real terms, that is, dollars of constant purchasing power rather than nominal dollars. The Court expressed its concern as follows: ‘Inflation’s current magnitude and impredictability create a substantial risk that the damages award will prove to have little relation to the lost wages it purports to replace’.87 When discussing the calculation of past losses, the Court, however, has not expressed the same concern with respect to nominalism. This concern about the effects of inflation on purchasing power has not been discussed because the Court has held that the payment of prejudgment interest is prohibited in FELA cases because common law prohibits such payments and Congress has been silent on the issue. The Court, in Monessen, notes that neither FELA nor the Federal Interest Statute mention prejudgment interest. Monessen began as a case in Pennsylvania State Court in which the judge applied a Pennsylvania statute that permitted damages to be increased by 10% for ‘damages of delay’. The case was appealed to the Pennsylvania Supreme Court which affirmed the original award of $26,712.50 in prejudgment interest. The case was appealed to the United States Supreme Court on the grounds that Pennsylvania law was superseded in FELA cases by Federal law.88 The Court’s opinion was 5–4 to reverse the original state court decision and to remand the case for further review. Justices Blackmun and Marshall agreed in part, but dissented in part, as did Justices O’Connor and Rhenquist. The Blackmun dissent is particularly noteworthy and will be discussed later. The Supreme Court’s position is based on its perception that Congress, by its failure to include the issue in the legislation, provided a clear indication that prejudgment interest was to be prohibited. In the 1908 version of FELA, Congress did, however, change other aspects of common law, such as the

Bars to the Payment of Prejudgment Interest

129

defense of contributory negligence, in order to liberalize the awards that could be made to injured workers under FELA. FELA was debated in the House on April 6, 1908 and in the Senate on April 9.89 The debate in both houses centered on making certain that the bill (HB 20310) met the concerns raised by Justice White in his opinion overturning the 1906 version of the Federal Employers’ Liability Act. The majority held that the 1906 law attempted to regulated intrastate as well as interstate commerce. The House Bill faced little opposition and was passed 302 yeas, 1 nay (Charles E. Littlefield of Maine), 7 present, and 79 not voting. Most of the remarks were made by representatives from Western and Southern States, states with large farm populations and towns that were dependent upon the railroads for connection to the major urban centers in the East.90 The debate in the Senate was more intense, with several amendments proposed and rejected. One of the proposed amendments would have extended the law to cover all interstate commerce regardless of the mode of transportation.91 There was also some discussion of using this opportunity to bring the law concerning recovery of damages incurred while engaged in interstate commerce into conformity with maritime law.92 Senator Augustus O. Bacon of Georgia raised the strongest objection to the bill, arguing that the proposed legislation would replace state law that made it easier for individuals injured while working for the railroad to recover damages. His concerns were (1) state courts were more convenient than Federal courts for plaintiffs. Transferring jurisdiction to Federal courts could make the plaintiff’s costs of suing so high as to discourage the injured from seeking recovery of damages; and (2) The railroads were more likely to try to modify Federal legislation than they were to try to alter legislation at the state level.93 Those speaking in favor of the bill tended to be from the West with most opposition from the South and from those states that had laws similar to the law proposed in HB20310. While Democratic Senators had consulted railroad employees as to their position on the bill, by 1908, there seemed to be little organized opposition to the bill on the part of the railroads. After a third reading, the bill passed the Senate on a voice vote. The issue of how damages were to be determined was barely mentioned in the debate over HB20310. The bill simply states that the determination of damages is to be left to the jury. Further, Congress has never amended FELA to provide for the payment of prejudgment interest when it has had the opportunity to do so. The Court concluded by stating that if prejudgment interest is to be made available under FELA, Congress must provide so explicitly.94

130

DAVID E. AULT & GILBERT L. RUTMAN

If taken literally, this emphatic and definitive statement with respect to the payment of prejudgment interest in FELA cases could create confusion in adjusting damages for past losses for the effects of inflation on purchasing power between the time the tort occurred and the date upon which the verdict was given. To adjust awards for the effects of past rates of inflation, that is, to compensate the plaintiff in real, rather than nominal dollars, interest must be paid. The interest rate must be greater than or equal to the rate of inflation, for any time the market or nominal rate of interest is less than the rate of inflation, the real rate of interest is negative. The plaintiff, therefore, not only is denied interest on his past losses but receives compensation with less purchasing power than actual damages incurred. The Supreme Court’s position on the payment of prejudgment interest is puzzling given the opinion of the Court of Appeals in Poleto v. Consolidated Rail Corp.95 The Appeals Court did not interpret Congress’s silence in FELA as an unwillingness to permit the payment of prejudgment interest. On this issue, the Appeals Court stated: Both the general federal interest statute . . . and the FELA are silent on the issue of prejudgment interest. The silence on neither statute is controlling, however. . . . The absence of a FELA provision concerning prejudgment interest . . . does not end our inquiry. As the Supreme Court has instructed, the failure to mention interest in statutes which create obligations has not been interpreted by this court as manifesting an unequivocal congressional purpose that the obligation shall not bear interest. For in the absence of an unequivocal prohibition on interest on such obligations, this court has fashioned rules which granted or denied interest on particular statutory obligations by an appraisal of the congressional purpose in imposing them and in the light of general principles deemed relevant by the court.96

While the Third Circuit Court upheld the prohibition of the payment of prejudgment interest, the opinion went on to state that the purposes of FELA would be enhanced by permitting the payment of such interest. In the court’s view, the payment of prejudgment interest would further address the intent of Congress to permit ‘liberal recovery’ in FELA cases, a position taken by the Supreme Court in Atchison, Topeka, and Santa Fe v. Buell.97 Such payments would promote the full recovery of damages suffered by railroad workers.98 In the end, the Third Circuit Court opinion seems to support McCormick’s contention that the courts simply do not treat interest as a substantive issue and, therefore, simply dispose of the point summarily by noting that the issue was settled in a previous case.99 In his minority opinion concurring in part, while dissenting in part with the majority opinion in Monessen, Justice Blackmun, joined by Justice Marshall, stated, ‘Plainly, the ‘congressional purpose’ in providing a damage award under

Bars to the Payment of Prejudgment Interest

131

FELA was to compensate a worker for his injury. Nothing in the legislative history or in our decisions suggests that the compensation was intended to be anything less than what was required to compensate him fully’.100 Blackmun went on to argue that there was sufficient precedent to permit the payment of prejudgment interest if such payment would make the injured party whole. Blackmun further argues that the reason that prejudgment interest is not awarded may lie in the fact that damage awards may include payment for pain and suffering for which interest is ‘usually not awarded’,101 as well as payment for actual economic losses. As Blackmun and the Third Circuit Court judges note, the fact that Congress said nothing about prejudgment interest does not lead to the conclusion that such payments are barred. The legislation itself simply states that the determination of damage awards should be left to the jury.102 Blackmun further notes that the ‘unavailability of interest in FELA actions is particularly curious’ given previous decisions, especially Pfeifer and Southwestern v. Dickinson.103

SUMMARY AND CONCLUSIONS The purpose of awarding damages is to make the plaintiff ‘whole’, that is, to place the plaintiff in approximately the same position with respect to his standard of living, earnings, or value of his property that he would have been in had the tort not occurred. The prohibition of the payment of prejudgment interest is, therefore, curious, for the failure to pay such interest leaves the plaintiff less than whole. The degree to which an award without prejudgment interest fails to fully compensate the plaintiff depends upon the rate of inflation and the length of time between the occurrence of the tort and the verdict. The Court’s rationale for barring prejudgment interest appears to be rooted in precedents in English common law that date back to the twelfth-century; the principle of nominalism that has dominated damage awards in England; the problems of measuring damages in personal injury and wrongful death cases; the role of the railroads in the industrialization of the United States, especially in the late nineteenth- and early twentieth-centuries; and the special relationship that Congress had with the railroads between the Civil War and WWII. Given the exemption of railroad workers from Workers’ Compensation, it seems clear that Congress intended that railroad workers who were injured on the job be given the opportunity to sue and recover damages even if the damages may have been due, in part, to the worker’s own negligence; and to recover fully, regardless of the worker’s own negligence, if the injuries were the result of a coworker’s negligence or an employer’s violation of the law. This approach seems to have been used because working for a railroad was

132

DAVID E. AULT & GILBERT L. RUTMAN

dangerous. The differential treatment accorded plaintiffs in FELA cases seems to have achieved its purpose of creating a safer work environment, for the number of accidents and the accident rate in the railroad industry fell dramatically within a decade following the passage of FELA. The railroad industry is now as safe, if not safer, than most other heavy industries in the United States. Given the Court’s argument in Pfeifer on the determination of the discount rate to be used when calculating future damages, it is difficult to explain the Court’s continued refusal to permit prejudgment interest until Congress creates legislation that explicitly permits such payments. The courts in England and the United States have recognized for a least a century that inflation reduces nominal awards in a manner that adversely affects the plaintiff’s award for future damages. This has led the courts to argue that some approximation of the real rate of interest should be used when discounting future damages to present value. The failure of the Court to recognize that past damages are also adversely affected by inflation in a manner that makes the plaintiff less than whole is even more puzzling in light of the courts’ recognition of the effects of inflation on future damages. The Court may have passed on this issue in earlier periods for one or more of the following reasons: (1) The length of time between the occurrence of the tort and the date that a verdict was handed down was so short that the effect of inflation on the purchasing power of the award for past damages was so small that it could be ignored. (2) At the time that FELA was enacted, the concept of paying interest was unusual. (3) In personal injury and other cases in which damages are unliquidated, the courts could not find a method that would lead to a ‘fair’ determination of the amount of prejudgment interest to be awarded. (4) Personal injury awards can be complex with some parts eligible for prejudgment interest and other parts ineligible. As long as the Court rests its opinions on the grounds that common law prohibits the payment of prejudgment interest in FELA and similar cases, prejudgment interest will be barred in federal court. Because FELA cases result in unliquidated claims, the Court could prohibit the payment of prejudgment interest on grounds of fairness, that is, because it is difficult to ascertain when the claim became a debt owed by the defendant. The Court, however, might clarify its position with respect to inflation and interest. The Court could, for example, state that the real rate of interest on past damage awards must be zero

Bars to the Payment of Prejudgment Interest

133

which would permit the nominal rate of interest of such awards to exceed zero. This position would be consistent with its opinion in Pfeifer on the calculation of future damages. Another approach that the Court might take would be to permit the jury to decide whether or not the payment of prejudgment interest is appropriate, a position held by some courts in the past when dealing with unliquidated claims. A similar approach would be for judges to instruct the jury as to when prejudgment interest can be paid and the type of damages to which prejudgment interest may be applied. Any of these approaches would permit damage awards in FELA cases to better approximate the goal of making the injured party whole.

ACKNOWLEDGMENT The authors wish to thank an anonymous referee for his comments on an earlier version. Any errors that remain are, of course, the sole responsibility of the authors.

NOTES 1. Monessen Southwestern Railway Co. v. Morgan (1988). 2. In September 1998, search of Shepard’s yielded 161 citations of Monessen. Of these, 36 were Federal court cases heard at the Appellate level. Of the 36, 26 did not involve a railroad. 3. Rachel v. Consolidated R. Corp. (1995) concerns how the loss of pension benefits is determined when the length of actual service is less than the length of service that would have determined pension benefits had the individual retired voluntarily. In calculating losses, the Tier II contribution rate may not be used to calculate losses in earnings plus fringe benefits. Prior to Rachel, estimates of losses in fringe benefits could include Tier II contributions. 4. The 1908 legislation was actually the second attempt to create a Federal statute governing damages suffered while employed by a railroad in interstate commerce. The original legislation was passed in 1906 but found to be unconstitutional by the Court for attempting to regulate intrastate as well as interstate commerce. 5. U.S. Code, 1988 Ed., vol. 18, Table 43–7. 6. National Transportation Research Board (1994, p. 2). 7. National Transportation Research Board (1994, p. 3). 8. National Transportation Research Board (1994, p. 2). 9. National Transportation Research Board (1994, p. 2). 10. Accidents were such that Congress asked for the following information in the first session of the Sixtieth Congress, the session in which the 1908 version of FELA was passed: Senate Doc. 682 and House Res. 539 asked that the Interstate Commerce Commission provide a list of all persons killed or injured by boiler explosions since July 31, 1903 (See United States Cong., Congressional Record (1908, pp. 1352 and 1967). Senate Res. 267 asked the ICC to report on the number injured or killed as the result

134

DAVID E. AULT & GILBERT L. RUTMAN

of faulty sill steps, hand brakes, ladders, roof holds, grab irons, or running boards beginning in 1901. 11. National Research Transportation Board, Appendix A (1994, Table A–2, p. 175). 12. Early in the first session of the Sixtieth Congress, a joint resolution, Senate Res. 80 and House Res. 177, was presented calling for a Constitutional amendment to address the issue of employers’ liability (Congressional Record (1908, p. 5514). Rep. Adolph J. Sabath (Illinois) delivered a speech on June 6, 1910, in which he compared the death and injuries while employed on the railroad to the casualties of several Civil War battles, as well as to the death and injuries on the railroads in several European countries. He predicted that FELA would reduce the number of deaths and injuries by 50%. See Congressional Record (1910, pp. 303–4, 307). 13. There were several state laws similar to FELA in effect prior to 1908. As the figures in Table 1 indicate, these laws seemed to have little effect upon the number of accidents in the industry. 14. National Research Transportation Board (1994, p. 3). 15. See C.T. McCormick, Cases and Materials on the Law of Damages (1935); and T. Sedgwick, A Treatise on the Measure of Damages (1920). 16. Supreme Court of Wisconsin (1899). 17. McCormick (1935, p. 62). 18. Depalma v. Weinman (Supreme Court of New Mexico, 1909); and Pierce v. Lehigh Valley ( Supreme Court of Pennsylvania, 1911). 19. McCormick (1935, pp. 70–2). 20. McCormick (1935, p. 72). 21. Supreme Judicial Court of Massachusetts (1912). 22. McCormick (1935, pp. 78–9). 23. McCormick (1935, p. 79). 24. McCormick (1935, pp. 79–80). 25. Sedgwick (1920, pp. 553–5). 26. Sedgwick (1920, p. 553). 27. Sedgwick (1920, p. 555). 28. Sedgwick (1920, pp. 567–8). 29. Sedgwick (1920, pp. 564–637). 30. Sedgwick (1920, p. 626). 31. Sedgwick (1920, p. 626). 32. Sedgwick (1920, pp. 555–630). 33. Sedgwick (1920, pp. 24–5). 34. Sedgwick (1920, pp. 20–3). 35. Friedman (1985, p. 21). 36. Gregory (1951, pp. 359–60). 37. Gregory (1951, p. 360). 38. Milsom (1969, pp. 249, 261–5, especially, p. 265). 39. Posner (1972, pp. 29–96). 40. Friedman (1985 , p. 467). 41. Pollock & Maitland (1968, p. 537). 42. Holdsworth (1966, 3, p. 375). 43. Gregory (1951, p. 363). 44. McCormick (1935, p. 60); and Holdsworth (1966, 2, p. 469).

Bars to the Payment of Prejudgment Interest

135

45. McCormick (1935, p. 60); and Holdsworth (1966, 2, p. 469). 46. Friedman (1985, p. 22). 47. Holdsworth (1966, 1, p. 194). 48. Holdsworth (1966, 1, p. 194). 49. Pollock & Maitland (1968, 2, p. 664); and Holdsworth (1966, 1, p. 214). 50. Holdsworth (1966, 1, p. 225). 51. Pollock & Maitland (1968, 1, pp. 590–1). 52. Pollock & Maitland (1968, 1, p. 666). 53. Butterfield v. Forrester, King’s Bench (1809). This, however, was not the general rule in the United States, see Gregory (1951). 54. Friedman (1985, p. 468); and Gregory (1951, p. 382). 55. Friedman (1985, p. 468). 56. Friedman (1985, pp. 468–9). 57. Posner, (1977, pp. 22–3). 58. McCormick (1935, p. 59). 59. Mann (1953, p. 78). 60. Mann (1953, pp. 77–80, 83). Mann notes that there may have been certain qualifications to the principle of nominalism when awarding damages but that such exceptions were rare. 61. Sands v. Devon (1945). 62. Mann (1953, p. 95). 63. Hart v. Giffeth-Jones (1948). See Mann (1953, p. 96), for further discussion. 64. Dawson & Cooper (1935, pp. 863–70). 65. Willard v. Taylor (1869). 66. Dawson & Cooper (1935, pp. 867–8). 67. Dawson & Cooper (1935, p. 886). 68. Dawson & Cooper (1935, pp. 869–72). 69. Dawson & Cooper (1935, pp. 872–3). 70. Friedman & Schwartz (1982, pp. 121–37). Data derived primarily from Table 4.8. 71. Dawson & Cooper (1935, p. 886). 72. Dawson & Cooper (1935, p. 888). 73. Sedgwick (1920, p. 570). 74. McCormick (1935, pp. 79–80). 75. Sedgwick (1920, pp. 569–70). 76. Sedgwick (1920, p. 571). 77. Sedgwick (1920, pp. 616–7). Sedgwick refers to McMahon v. New York and Erie Railroad, 20 N.Y., 463, 469; and Van Renseelaer v. Jewett, 2 N.Y., 135. 78. Sedgwick (1920, p. 617). 79. Sedgwick (1920, p. 626). 80. Jones and Laughlin Steel Corp. v. Pfeifer (1983). 81. Pfeifer (1983, p. 532). 82. Pfeifer (1983, p. 540). 83. Pfeifer (1983, pp. 540–1). 84. Irving Fisher, in his work on interest rates, found that his estimates, using data particularly from England in the latter half of the nineteenth-century, consistently produced a real rate of 2.0%. Subsequent estimates have been unable to dispute this

136

DAVID E. AULT & GILBERT L. RUTMAN

finding of Fisher that, in other than relatively short periods of time, the real rate of interest is two %. See Fisher (1930). 85. Pfeifer (1983, p. 549). 86. O’Shea v. Riverway Towing Co. (1982). 87. Pfeifer (1983, pp. 546–7). 88. The history of Morgan v. Monessen Southwestern Railway Co. prior to its appeal to the U.S. Supreme Court is: 339 PA. Super. 465; 489 PA. 2d 254; 513 PA. 86; 518 PA. 2d 1171 See, Monessen, p. 2. The majority opinion was written by Justice White with Justices Brennan, Kennedy, Scalia, and Stevens concurring. 89. Congressional Record (1908, pp. 4427–43, for the House, and pp. 4526–51, for the Senate. 90. Congressional Record (1908, p. 4439). 91. Congressional Record (1908, p. 4532). 92. Congressional Record (1908, p. 4536). 93. Congressional Record (1908, pp. 4546–7). 94. Monessen, p. 7. 95. Charles E. Poleto v. Consolidated Rail Corporation v. Hammermill Paper Company v.A. E. Staley Manufacturing Co., and the Baltimore and Ohio Railroad Company (1987). 96. Poleto, pp. 10–11, emphasis added. 97. Atchison, Topeka, and Santa Fe Railway Co. v. Buell (1987). 98. Poleto, p. 12. 99. Poleto, p. 2–3. 100. Monessen, pp. 8–9. 101. Monessen, p. 9. 102. Monessen, p. 9; and Poleto. See also USC, FELA, sec. 2. 103. Monessen, p. 9; and St. Louis Southwestern Railway Co. v. Dickinson (1985). Justice O’Connor, with Chief Justice Rhenquist joining, while also dissenting in part to the majority opinion, agreed with the majority that prejudgment interest is impermissible in FELA cases. See Monessen, p. 11.

REFERENCES Dawson, J.P., & Cooper, F.E. (1935). The Effect of Inflation on Private Contracts: United States, 1862–79, Michigan Law Review, 33, 863–70. Fisher, I. (1930). The Theory of Interest Rates. New York: MacMillan and Co. Friedman, L. (1985). A History of American Law (2nd ed.). New York: Simon and Schuster. Friedman, M., & Schwartz, A. (1982). Monetary Trends in the United States and the United Kingdom. Chicago: National Bureau of Economic Research. Gregory, C. (1951). Trespass to Negligence in Absolute Liability. Virginia Law Review, 37, 359ff. Holdsworth, W. Sir (1966). A History of English Law. Vols. 1, 2, and 3. London: Vanhusen and Co., Ltd., 1903. Reprinted by Sweet and Maxwell, Ltd. Mann, F.A. (1953). The Legal Aspect of Money (2nd ed.). Oxford: The Clarendon Press. McCormick, C.T. (1935). Cases and Materials on the Law of Damages. Chicago: The Foundation Press. Milsom, S. F. C. (1969). Historical Foundations of the Common Law. London: Butterworth and Co., Ltd.

Bars to the Payment of Prejudgment Interest

137

National Transportation Research Board (1994). Compensating Injured Railroad Workers Under the Federal Employers’ Liability Act. Special Report 241. Washington, D.C.: National Academy Press. Pollock, F. Sir, & Maitland, F. W. (1968). The History of the English Common Law (2nd ed.). Vol. 1 and 2. Cambridge: Cambridge University Press. Posner, R. (1977). Economic Analysis of the Law (2nd ed.). Boston: Little, Brown, and Co. Posner, R. (1972). The Theory of Negligence. Journal of Legal Studies, 1, 29–96. Sandler, J. R. (1992). Prejudgment Interest in Personal Injury Claims: A Proposal for the Illinois General Assembly. The John Marshall Law Review, 25, 595–627. Sedgwick, T. (1920). A Treatise on the Measure of Damages (9th ed.), Vol. 1. New York: Baker, Voohis, and Co. United States Congress. (1908). Congressional Record. Sixtieth Congress. First Session. Washington, D.C.: United States Government Printing Office. United States Congress. (1910). Congressional Record. Sixty First Congress. First Session. Washington, D.C.: United States Government Printing Office.

LEGAL CASES Atchison, Topeka, and Santa Fe Railway Co. v. Buell (1987). No. 85–1140. Butterfield v. Forrester (1809). King’s Bench. 11 East Go. Cochran v. City of Boston (1912). Supreme Judicial Court of Massachusetts. 211 Mass. 171. Dipalma v. Weinman (1909). Supreme Court of New Mexico. 15 N. Mexico 68, 103 Pac. 782. Hart v. Giffeth-Jones (1948). 2 All E.R. 729. Jones and Laughlin v. Pfeifer (1983). 462 U.S. 523. Laycock v. Parker (1899). Supreme Court of Wisconsin. 103 Wis. 161. Monessen Southwestern Railway Co. v. Morgan (1988). 486 U.S. 330. O’Shea v. River Towing Co. (1982). 667 F. 2d. CA7. Pierce v. Lehigh Valley (1911). 232 PA 170. 81 ATL 142. Charles E. Poleto v. Consolidated Rail Corporation v. Hammermill Paper Co. v.A. E. Staley Manufacturing Co., and the Baltimore and Ohio Railroad Co. (1988). Nos. 86–5249, 86–5250. 826 F 2s 1270. Rachel v. Consolidated Railway Corp. (1995). 891 F. Supp. 428 (N.D. Ohio). St. Louis Southwestern Railway Co. v. Dickinson (1985). 470 U.S. 409. Sands v. Devon (1945). S.C. 380. United States Code (1988). Vol. 18. Willard v. Taylor (1869). 75 U.S. 557.

This Page Intentionally Left Blank

OF DISTRIBUTIVE JUSTICE AND ECONOMIC EFFICIENCY: AN INTEGRATED THEORY OF THE COMMON LAW Brian N. Wasankari, Richard O. Zerbe Jr., and Steven Vinyard ABSTRACT Zerbe has elsewhere developed a concept of economic efficiency that has implications for common law efficiency. Here we explore aspects of this concept of efficiency for the relationship between common law efficiency and considerations of distributive justice. In particular, we consider examples from criminal law – the law of rape, from contract law– exculpatory clauses, and from tort law–contribution. We find that including considerations of distributive justice better explains common law efficiency than traditional Kaldor-Hicks effficiency.

I. GENERAL INTRODUCTION Recently, Richard O. Zerbe, Jr. (due to be published 2001) developed an axiomatic system to provide grounding for the application of efficiency analysis to normative issues. This paper is an application of several of those axioms. The application primarily involves the concept, created by Zerbe, of

Research in Law and Economics, Volume 19, pages 139–222. 2000 JAI/Elsevier Inc. ISBN: 0-7623-0308-5

139

140

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

the regard for others, which refers to the value people place on projects that do not directly effect them but which do effect others whom they care about. Microeconomic analysis1 relies on the formulation of models to analyze its subject phenomena (Mansfield, 1994). Thus, microeconomic analysts of the common law2 have constructed models, or more precisiely theories,3 of the common law itself, by which, ultimately, they hope to explain its doctrine in terms of a ‘handful of equations’ (Landes & Posner, 1987, pp. 312–13).4 The most successful economic theory extant is that which will here be termed the traditional Kaldor-Hicks efficiency theory of the common law.5 That theory holds that the common law is best described as a system designed to promote the traditional notion of Kaldor-Hicks efficiency (Posner, 1992, Economic Analysis, p. 23).6 Although the traditional theory has met with significant descriptive success (Bruce, 1984), it has also ‘aroused considerable antagonism’ (Posner, 1992, Economic Analysis, p. 25). Yet despite suffering such ‘slings and arrows’, (Shakespeare Hamlet, Act 3, Sc. 1) its underlying validity remains (Posner, 1992, Economic Analysis p. 26).7 That validity is vulnerable, however, since it may be invalidated through the introduction of an alternative theory that better explains common-law reality (Kuhn, 1970; Posner, 1992, Economic Analysis, pp. 17–18; Beutel, 1957).8 Two facts make the possibility of such invalidation particularly likely. First, the conception of Kaldor-Hicks efficiency, upon which the traditional theory is built, can be improved (Zerbe, 1998, ‘Three Rules’ and Zerbe, 2001, Chapter 2). Indeed, Zerbe (2001, Chapter 8) claims that the analysis he develops explains the conditions under which the common law will be efficient, as well as the conditions under which it is likely to be inefficient. The traditional analysis erroneously mandates that the economic efficiency of an allocative change be measured independently of the effect of that change on the distribution of wealth and the fact of compensation (Zerbe, 1998, ‘Three Rules’, p. 426f; 2001, Chapter 3).9 This, in turn, results in either the neglect or abrogation of distributive justice concerns.10 The second fact is that judges seek not only an efficient, but a just result, where justice is both corrective and distributive.11 Taken together, these facts suggest that an efficiency theory which properly considers not only traditional efficiency, but also distributive justice, will provide a better description of the common law than does the traditional theory. Indeed, that is the premise of this paper, which first proposes an alternative efficiency theory of the common law, deemed the integrated or KHZ theory, and then tests the descriptive capacity of that theory against that of the traditional theory.

Of Distributive Justice and Economic Efficiency

141

Section II provides the starting point by describing and logically evaluating both the traditional KH theory and KHZ, and suggesting that, because only the latter considers distributive justice, it will produce a better description of the common law than does the former.12 Section III tests this suggestion in two steps. First, it sets forth ‘descriptive capacity’ as the basis for the evaluation of theory. Second, it attempts to measure the descriptive capacity of both theories through a three-part survey of problematic common-law doctrine, which broaches the fields of crimes, contracts, and torts. That survey begins with criminal law, and, specifically, the law of rape. It finds that although certain rapes should be legal under the traditional KH theory, all should be illegal under KHZ given modern social norms. Since all rapes are actually illegal today, rape law is better described by KHZ. Next, contract law, and particularly the law of exculpatory clauses, is considered. Again, it is found that the integrated theory provides a better description than does its traditional counterpart, for although exculpatory clauses seeking to exclude liability for negligence are efficient under the latter, they are inefficient under the former. Last, the tort law movement toward contribution among negligent joint tortfeasors is analyzed. That movement is found to be inefficient under the traditional theory, but efficient under the integrated theory, and therefore better described by the latter. Given that the law of rape, exculpatory clauses, and contribution are better described by KHZ than by the traditional efficiency theory, it is cautiously concluded that, ceteris paribus, the former provides a better description of the common law than does the latter. If a sufficient quantum of future research should confirm or expand this result, we argue that the traditional theory must be supplanted by the integrated theory.13

II. THE THEORIES A. Introduction Before theory can be analyzed, however, it must be enunciated (Byrns & Stone, 1992, p. 15). Accordingly, this section will provide a basic description, and some theoretical analysis, of both the traditional KH efficiency theory and KHZ. First, KHZ will be described as stating that the common law, in the aggregate, is best explained as a system designed to promote traditional Kaldor-Hicks efficiency. Second, it will be noted that this theory has elicited considerable criticism. Although much of this criticisim is easily dismissed,

142

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

what will be termed the distributive justice critique ultimately suggests the demise of the traditional theory. Third, with that suggestion in mind, the integrated theory will be described as holding that the common law, as a whole, is best explained as a system designed to promote KHZ efficiency. Last, it will be shown that not only is the distributive justice critique not a threat to KHZ, but that it actually suggests that theory’s descriptive superiority. B. The Traditional Kaldor-Hicks Efficiency Theory of the Common Law 1. Exposition Economic theory of law, in general, is of relatively recent vintage (Cunningham, 1992, p. 141). Nevertheless, it has become profoundly influential.14 Indeed, what began as an adjunct and limited analytical tool of the law has been transformed into a holistic theory of law itself.15 An element of that theory, what is here termed the ‘traditional Kaldor-Hicks efficiency theory of the common law’,16 has had a particularly profound impact on legal discourse (Landes & Posner, 1987, pp. 24–25). That theory, the focus of this subsection, will be described in three steps. First, its theoretical composition will be defined. Second, its fundamental proposition will be recounted. Third, the twin concepts implicit in that proposition, ‘the common law’ and ‘efficiency’, will be explained. To properly define the composition of the traditional Kaldor-Hicks efficiency theory of the common law, it is necessary to consider the more general ‘efficiency theory of the common law’ (Posner, 1992, Economic Analysis, pp. 23, 25). The latter is actually an amalgam of subordinate theories, one positive and one normative.17 The positive theory ‘hypothesizes that common law [sic] rules and decisions are best explained. . . ‘as if’. . . judges [were] consciously trying to promote’ economic efficiency (Posner, 1981, p. 775).18 The normative version is both practically valuable19 and theoretically relevant.20 However, this paper focuses solely on the positive version.21 Thus, what is here termed the traditional Kaldor-Hicks efficiency theory of the common law is synonymous with what is elsewhere called the positive efficiency theory of the common law.22 The fundamental proposition of that theory, regardless of its name, ‘is not that every common-law doctrine and decision is efficient’ (Posner, 1992, Economic Analysis, p. 23),23 but ‘that the common law is best. . . explained as a system [designed]. . .. to promote [economic] efficiency’ (Posner, 1992, Economic Analysis, p. 23).24 As such, the theory does not purport to explain ‘why the common law has become concerned with efficiency’,25 but merely to explain the common law itself (Posner, 1981, p. 779).

Of Distributive Justice and Economic Efficiency

143

Implicit within the theory’s fundamental proposition are two key concepts which demand definition: ‘the common law’ and ‘efficiency’. Although the term ‘common law’ is subject to a multitude of definitions (Calleros, 1994) within the context of the traditional efficiency theory of the common law, it is generally defined as ‘th[ose] fields of law that have been created largely by judges as a by-product of deciding cases’ (Posner, 1992, Economic Analysis, p. 779). There is authority for such a definition, and, indeed, it shall be adopted here as well.26 ‘Efficiency’, within the context of the positive efficiency theory of the common law, has come to mean Kaldor-Hicks efficiency,27 or potential Pareto superiority (Cooter & Ulen, 1997, p. 41; Posner, 1992, Economic Analysis, p. 14).28 Under the Kaldor-Hicks standard,29 a change in the allocation of goods is efficient if those who gain from the change gain more than those who lose.30 It is not necessary that the gainers actually compensate the losers, but only that the gainers could compensate the losers.31 Hence, it is possible, and indeed, likely, that Kaldor-Hicks efficient allocative changes will produce both gainers and losers (Cooter & Ulen, 1997, p. 4).32 Within the context of the efficiency theory of the common law, Kaldor-Hicks efficiency is more often referred to as ‘wealth maximization’, but the terms are, purportedly, synonymous (Posner, 1992, Economic Analysis, pp. 13–14).33 To summarize, what is here termed the traditional Kaldor-Hicks efficiency theory of the common law is what has traditionally been referred to as the positive efficiency theory of the common law. That theory holds that those fields of law which have been created largely by judges as a by-product of deciding cases are, in the aggregate, best explained as a system designed to promote traditional Kaldor-Hicks efficiency. 2. Evaluation a. Introduction The traditional efficiency theory of the common law has evoked a substantial amount of criticism (Landes & Posner, 1987, pp. 9, 17).34 Although much of this criticism is misguided,35 at least one critique36 approaches validity37 and ultimately suggests the abrogation of the traditional theory. That critique holds that judicial behavior, and hence, the common law itself, cannot be described solely in terms of traditional Kaldor-Hicks efficiency, because such efficiency either ignores38 or undercuts distributive justice.39 This subsection will first, briefly recapitulate what may rightly be regarded as ineffectual criticism of the traditional theory, and second, present the distributive justice critique which suggests the ultimate demise of that theory.

144

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

b. Ineffectual Criticism Among the ineffectual criticisms are those pertaining to the theory’s (1) underlying assumptions, (2) incomplete explanatory power, (3) purported conservative political bias, (4) lack of causal explanation, and (5) empirical procedure.40 A frequent criticism of the traditional efficiency theory is that it rests on unrealistic assumptions (Polinsky, 1989, p. 16–17; Landes & Posner, 1987, p. 9). That criticism is answered in the simple proposition that ‘abstraction is of the essence of scientific theory’ (Posner, 1992, Economic Analysis, p. 17). To be valuable, ‘a model must in general simplify and abstract from the real situation’ (Mansfield, 1994, p. 13).41 To this basic truth, Landes & Posner (1987, p. 22) add another, noting that ‘[o]ne can complicate a theory to the point where any empirical observation is consistent with it’, but when this point is reached, the theory can neither be refuted nor confirmed. Thus, ‘lack of realism’, is in effect, ‘a precondition of theory’(Posner, 1992, Economic Analysis, p. 17). It is also argued that the traditional theory is flawed because it cannot explain every common law doctrine (Posner, 1992, Economic Analysis, p. 26). This criticism is adequately met in a two-fold response. First, on a general level, as long as the predictions derived from one model are superior to those obtained from competing models, the former must be retained, for ‘[i]f one is interested in predicting the outcome of a particular event, one [is] forced to use the model that predicts best, even if this model does not predict very well’ (Mansfield, 1994, pp. 15–16).42 The traditional theory was, and perhaps remains, ‘the most promising theory of law extant’ (Posner, 1992, Economic Analysis, p. 26).43 Second, a theory that most rules are traditionally Kaldor-Hicks efficient can only be refuted by evidence (1) that most are traditionally Kaldor-Hicks inefficient or (2) that another theory explains more common-law rules than does the traditional efficiency theory (Landes & Posner, 1987, p. 24). The critical literature in this regard supports neither proposition, and must accordingly be dismissed. It has also been argued that economics itself is founded upon, and saturated by, a conservative political bias (Buchanan, 1974). Aside from being irrelevant (Mansfield, 1994, p. 14), this criticism ignores several liberal economic findings (Posner, 1992, Economic Analysis, p. 26). It is further argued that the traditional efficiency theory is unsatisfactory because it does not set forth the mechanism by which efficient common-law rules actually evolve (Landes & Posner, 1987, p. 14). To this, at least two responses are possible. First, it is not incumbant upon the traditional theory to offer such an explanation, for the theory need not explain ‘why the common law has become concerned with efficiency’ to assert that the common law may

Of Distributive Justice and Economic Efficiency

145

be described as if it were designed to promote efficiency (Posner, 1981, p. 776). Second, even if such an explanation were necessary, it has already been given (Landes & Posner, 1987, p. 19).44 Another refutable criticism holds that the empirical procedure used to test the theory is flawed (Landes & Posner, 1987, pp. 22–23). This criticism proceeds in two steps. First, it is argued that the method of comparing theoretically efficient rules and outcomes to the rules and outcomes of reported appellate decisions is ‘rationalization rather than prediction’ and hence, of no empirical value (Landes & Posner, 1987, p. 21). Second, it is argued that because law and economics commentators have relied only on the ratio decidendi of opinions to the exclusion of other language, they have ignored potentially conflicting, and, thus, invalidating data (Landes & Posner, 1987, p. 22). This second argument usually focuses on the fact that the language of opinions is not explicitly economic (Posner, 1992, Economic Analysis, p. 23).45 In response to the first argument, it has been noted that much of legitimate science is engaged in explaining the known rather than in predicting the unknown (Landes & Posner, 1987, p. 22). With respect to the second, it is sufficient to note that no economic theory requires that its subjects be aware of such theory. All that is required is that they behave in a manner approximated by that theory (Posner, 1992, Economic Analysis, p. 23).46 Even the traditional efficiency theory largely meets this test (Posner Economic Analysis, p.26). c. The Distributive Justice Critique That theory encounters difficulty, however, in overcoming the distributive justice crtique. That critique proceeds from two fundamental premises. First, it assumes that judges seek to do justice (Zerbe, 2001, Chapter 8),47 where justice is both corrective and distributive (Aristotle Ethics, Bk. IV; Englard, 1993, pp. 11–20; Calnan, 1997, pp. 77–104). Second, it notes that Kaldor-Hicks efficiency, at least as traditionally construed,48 either ignores49 or undercuts distributive justice.50 Taken together, these premises entail that judicial behavior, and thus the common-law itself, cannot be completely described in terms of traditional Kaldor-Hicks efficiency (Landes & Posner, 1987, p. 9). This argument will be explored, first, through exposition and evaluation of the distributive justice critque itself, and, second, through discussion of its implications for the validity of the traditional efficiency theory. A description and evaluation of the distributive justice crtique is best approached through consideration of its underlying premises, which determine both its form and legitimacy. The first premise, that judges seek to do justice through the decision of controversies, contains both normative and positive elements.51 Although only the latter must be proven to prove the premise, there

146

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

seems to be adequate support for both.52 With respect to the normative element, that is, that judges should seek to do justice when deciding cases, there is both scholarly53 and popular consensus (Kaufman, 1980, p. 83).54 With respect to the positive element,55 that judges actually seek to do justice when deciding cases, there is, unfortunately, a paucity of empiricism. Nevertheless, it has been noted that if research reveals that ‘principles of justice’ run ‘consistently through. . . recorded decisions’, then the positive element must be upheld. Furthermore, it is hypothesized that this is the likely result (Kaufman, 1980, p. 87).56 Moreover, what empirical evidence there is suggests that judges actually do seek to do justice in deciding cases (Glick, 1990, pp. 261–302). Therefore, given the existing evidence, there seems to be sufficient support for the proposition that judges seek to do justice. Although this statement may be an adequate evaluation of the first premise’s validity, it is not a sufficient description of the premise itself, for it fails to define ‘justice’.57 Although there is no consensus as to the precise nature of justice,58 the basic Aristotelian conception of justice as an amalgam of corrective and distributive components (Aristotle Ethics, Bk. V)59 seems to enjoy nearly universal support (Fuller, 1949, p. 30).60 Indeed, it is one of ‘the major. . . premises of our legal and moral order’ (Fuller, 1949, p. 30). Thus, even if nothing more can be said, ‘justice’ may be said to be the synergistic aggregate of what Aristotle described as corrective61 and distributive justice. However, because the distributive justice critique is, by definition, solely concerned with distributive justice, that component will be the sole focus of inquiry here. Aristotle defined distributive justice as that form of justice ‘concerned with the distribution of honors, or wealth’ (Aristotle Ethics, Bk. V), which ‘dictates that goods be disseminated to. . . claimants in equal shares’.62 ‘Equality’, However, Aristotle did not use ‘equality’ in its contemporary sense,63 but used it to mean ‘allocat[ion] according to a geometric proportion’ (Calnan, 1997, p. 85). With respect to the concept of ‘geometric proportion’, Aristotle (Ethics, Bk. V) wrote that distributive ‘[j]ustice implies four terms’, two of which represent the two parties to the distribution, and two of which represent the respective shares of those parties. For distributive justice to be ‘accomplished’, ‘the relation between the first pair’ must be ‘the same as the relation between the second pair’. That is, G1 : S1 :: G2 : S2 where G1 and G2 are the groups subject to the distribution and S1 and S2 are the distributively just shares of G1 and G2 respectively (Aristotle Ethics, Bk. V).64 The ‘relation between the first pair’, i.e. G1 : G2, is, itself, determined according to the ‘relative equality between the persons’, that is, according to

Of Distributive Justice and Economic Efficiency

147

‘some fair criterion’(Calnan, 1997, p. 85). Thus, Aristotle (Ethics, Bk. V) wrote that ‘[i]f the persons are not equal, they should not have equal shares’, rather, ‘justice in distribution must be according to some kind of merit’ although people may ‘not agree on the standard of merit’.65 Thus, distributive justice may be defined as the distribution of wealth between groups according to the respective merit of those groups, and justice, itself, as a concept which requires distributive justice. Having defined justice, it becomes possible to adequately define the distributive justice critique’s first premise, and, hence, to more precisely state the conclusion as to its validity. Specifically, it may be said that there is sufficient support for the proposition that judges seek to do justice, where justice itself requires distribution of wealth between groups according to the respective merit of those groups. The distributive justice critique’s second premise holds that traditional notions of Kaldor-Hicks efficiency either ignore or undercut distributive justice. With this proposition, there is little dispute. Indeed, the proponents of the traditional theory themselves acknowledge the validity of both branches of the second premise.66 This premise will be described and evaluated by first examining how traditional Kaldor-Hicks efficiency ignores distributive justice and second analyzing its capacity to actually undercut such justice. To understand the mechanism by which traditional Kaldor-Hicks efficiency ignores distributive justice, consider a change in the allocation of goods.67 Any such change will produce a welfare effect that may be divided into two parts: the first relating to production, and the second to distribution (Kaldor, 1939, p. 697; Zerbe, 1998, ‘Three Rules’, appendix).68 Economists, however, normally consider only the first part in measuring the efficiency of an allocative change, and, thus, ignore considerations of distributive justice (Zerbe, 1998, ‘Three Rules’, p. 426f).69 They do so to make their discipline a science. Because any science assumes the capacity to make accurate measurements of its subject phenomenon,70 phenomena not susceptible to measurement cannot be scientifically studied. In economics, utility71 cannot be measured, and, hence, interpersonal comparisons of utility cannot be scientifically made (Byrn’s & Stone, 1992, p. 121; Zerbe, 2001, Chapter 2). Because consideration of the distribution effects of an allocative change is thought to require interpersonal comparisons of utility, such consideration is also thought to lack a scientific foundation (Kaldor, 1939).72 Conversely, the production effect can be measured (Posner, 1992, Economic Analysis, pp. 14, 264–264). Therefore, although any allocative change has both production and distribution effects, only the first are considered in ‘scientific’ calculations of economic efficiency. Therefore,

148

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

traditional Kaldor-Hicks efficiency mistakenly believes that it is necessary to ignore distributive justice (Zerbe, 1998, ‘Three Rules’, p. 426f).73 To make scientific calcultions of economic efficiency, traditional KaldorHicks efficiency74 ‘takes the existing distribution of . . . wealth . . . as given’ and remains largely ‘uncritical of the change in that distribution brought about by efficient’ allocative changes (Posner, 1992, Economic Analysis, p. 264).75 Because distributive justice, however, requires a distribution of wealth between groups according to their respective merit, traditional Kaldor-Hicks efficiency may be, and, indeed, often has been, ‘criticized for ignoring distributive justice’ (Posner, 1992, Economic Analysis, p. 27).76 Indeed, the major criticism of the traditional Kaldor and Hicks compensation tests is that they ignore income distribution effects (Zerbe, 1998, ‘Three Rules’, p. 426, 2001, Chapter 3; Anderson, 1993; Fried, 1977),77 and, hence, distributive justice (Posner, 1992, Economic Analysis, pp. 14, 27).78 Therefore, the first portion of the distributive justice critique’s second premise seems firmly established. That is, it seems clear that traditional Kaldor-Hicks efficiency ignores distributive justice. The second portion of that premise holds that traditional economic efficiency may actually undercut distributive justice (Polinsky, 1989, p. 7; Posner, 1992, Economic Analysis, p. 263).79 Indeed, because traditional efficiency ignores distributive justice, it is possible that, in order to achieve a traditionally efficient result, one must sacrifice distributive justice (Polinsky, 1989, p. 8). Thus, Posner (1987) notes that ‘if the initial allocation [of goods] is thought to be unjust,80 [an efficient allocative] change, while increasing the wealth of society,81 may actually be carrying it further away from the [distributively] just allocation’. Moreover, Polinsky (1989, pp. 8–9) suggests that, where postdecision redistribution of income is impossible, and judicial decision-making is guided by traditional economic efficiency, this result cannot be corrected. Of course, whether the pursuit of efficiency actually undercuts distributive justice depends on both (1) the distributional consequences of the allocative change in question and (2) society’s notion of the just distribution of wealth (Polinsky, 1989, p. 9). Proponents of the traditional efficiency theory have attempted to defend against the normative82 implications of this result through two arguments. First, they argue that if income can be costlessly redistributed, it is impossible for the pursuit of efficiency to undercut distributive justice, because the total gain produced by an efficient allocative change can be distributed such that G1 and G2 always receive S1 and S2, respectively, without diminishing the magnitude of the total gain (Polinsky, 1989, p. 9). Second, as long as income redistribution is possible, even if it is costly,83 it is not only unnecessary, but actually inappropriate, for judges to consider distributive justice in deciding cases.84

Of Distributive Justice and Economic Efficiency

149

Rather, ‘efficiency should be the principal criterion for evaluating the legal system’ (Polinsky, 1989, p. 9). The reason for this result is said to rest in the comparative capacities of the branches of government.85 Specifically, it is said that ‘common law judges lack. . . adequate tools for redistributing wealth’, (Posner, 1992, Economic Analysis, pp. 262, 264) or, more precisely, that the political branches possess superior ones.86 The reason, it is argued, is two-fold. The first is jurisprudential: because conceptions of distributive justice are ‘highly controversial’ while efficiency has ‘broad support’ and the judiciary is inherently anti-majoritarian, judges should focus on the non-controversial efficiency issues to the exclusion of the controversial distributive justice concerns (Landes & Posner, 1987, p. 18).87 The second reason is more pragmatic: ‘effective redistributive policies require taxing and spending powers [and prowess] that judges lack’, but which the legistative and executive branches have cultivated (Landes & Posner, 1987, p. 18).88 Therefore, it is argued that rationality requires a division of labor by which courts decide cases according to traditional Kaldor-Hicks efficiency and the political branches ‘stand ready to correct any [resulting] maldistribution’ of goods (Landes & Posner, 1987, pp. 17–18; Posner, 1992, Economic Analysis, p. 264).89 This argument, however, does not refute the proposition that traditional Kaldor-Hicks efficiency can undercut distributive justice. It merely reccomends a method of post-hoc correction when such efficiency is used as the polestar of judicial decision.90 Therefore, the second portion of the distributive justice critique’s second premise survives unscathed. That is, traditional Kaldor-Hicks efficiency has the capacity to actually undercut distributive justice. Moreover, because both the first and second premises of the distributive justice critique are valid with respect to traditional Kaldor-Hicks efficiency,91 it follows, as a matter of logic, that the critique’s conclusion is valid with respect to traditional Kaldor-Hicks efficiency as well.92 Therefore, because (1) judges seek to do both corrective and distributive justice, and, (2) traditional KaldorHicks efficiency either ignores or undercuts distributive justice, judicial behavior, and hence the common law, cannot be completely described in terms of traditional Kaldor-Hicks efficiency. Nevertheless, establishing the validity of the distributive justice critique does not establish the invalidity of the traditional efficiency theory. Rather, as described above, because the traditional theory posits only that most commonlaw doctrines are traditionally Kaldor-Hicks efficient, it can only be invalidated by evidence (1) that most are traditionally Kaldor-Hicks inefficient or (2) that another theory explains more common-law doctrines than does the traditional theory (Landes & Posner, 1987, p. 24).

150

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

The distributive justice critique states that judicial behavior, and, hence, the common law itself, cannot be completely described in terms of traditional Kaldor-Hicks efficiency because judges also seek to do distributive justice. It follows from the above that this proposition would be sufficient to invalidate the traditional theory, if it could be empirically shown that the distributive justice concerns of common law judges led them to produce inefficient results in most cases. Yet the empirical evidence is clearly to the contrary (Landes & Posner, 1987, p. 24).93 Hence, the distributive justice critique will probably not lead to invalidation of the traditional theory through evidence that most common-law doctrines are traditionally Kaldor-Hicks inefficient. Moreover, because the distributive justice critique is not itself an alternative theory of the common law, it seems very unlikely that that critique could itself invalidate the traditional theory. Nevertheless, the critique does suggest how such an invalidation might occur, for in suggesting that the common law was shaped by both efficiency and distributive justice concerns, it suggests that an efficiency theory which considers both would provide a better description of the common law than does the traditional theory, which considers only one.94 d. Conclusion The traditional theory has thus encountered a barrage of criticism. Some of it, like that pertaining to the theory’s underlying assumptions, incomplete explanatory power, purported conservative bias, lack of causal explanation, and empirical procedure, is clearly unfounded. The distributive justice critique, which holds that the common law cannot be described solely in terms of traditional Kaldor-Hicks efficiency, because such efficiency either ignores or undercuts distributive justice, is valid, but it is insufficient to invalidate the traditional theory. Nevertheless, it does suggest that an alternative efficiency theory, which considers both efficiency and distributive justice, would provide a better description of the common law than does the traditional theory. C. KHZ: An Integrated Theory of the Common Law 1. Exposition This subsection proposes an alternative efficiency theory of the common law. That theory is based on an efficiency standard, which will be called ‘integrated efficiency’. Integrated efficiency incorporates a consideration of the distributive justice of an allocative change into the calculation of the efficiency of that change and is, itself, the derivative aggregate of both (1) an efficiency standard here called KHZ efficiency,95 and (2) the Aristotelian conception of distributive justice.96 The theory constructed upon such efficiency will be termed the

Of Distributive Justice and Economic Efficiency

151

‘integrated efficiency theory of the common law’. It will be described in two steps. First, its fundamental proposition will be enunciated. Second, the two components of that proposition, ‘integrated efficiency’ and ‘the common law’, will be explained and defined. The fundamental proposition of the integrated theory is similar in structure to that of the traditional Kaldor-Hicks efficiency theory. Quite simply, it posits that the common law is best explained as a system designed to promote integrated efficiency. While it is not argued that every common-law doctrine is integratively efficient, it will be argued that the common law is better explained by integrated than traditional Kaldor-Hicks efficiency.97 Integrated efficiency, itself, may be described as a KHZ efficiency standard in which Aristotelian distributive justice is considered. KHZ efficiency may, in turn, be defined as a Kaldor-Hicks efficiency standard that incorporates, among other things, the willingness to pay for, and the willingness to accept, a change in the distribution of wealth (Zerbe, 1998, ‘Integration’, p. 354).98 Therefore, to adequately describe integrated efficiency, it is necessary, first, to describe KHZ efficiency, and, second, to illustrate the incorporation of Aristotleian distributive justice into the calculation of such efficiency. KHZ efficiency will be described in three steps. First, the rationale for the consideration of distributional effects in the calculation of efficiency will be briefly explained.99 Second, the method of such incorporation will be noted. Third, the requisite conditions of KHZ efficiency will be stated. KHZ efficiency includes a consideration of distributional effects in its calculation of efficiency, because the traditional exclusion of such effects100 is based on a misunderstanding of, and is, in a sense, internally inconsistent with the traditional Kaldor-Hicks efficiency criteria’ (Zerbe, 1998,‘Three Rules’, p. 426f; Zerbe, 1998, ‘Integration’ p. 354; Zerbe, 2001, Chapter 2). That misunderstanding is, in turn, the result of two conceptual failures: first, the failure to adopt a sufficiently broad definition of the term ‘good’, and, second, the failure to properly consider the ‘regard for others’ (Zerbe, 1998, ‘Integration’, p. 354f). A ‘good’ is anything of value (Zerbe, 1998, ‘Three Rules’, p. 422),101 or, more precisely, ‘a ‘good’ in economic analysis is defined as what people care about’ (Zerbe, 1998, ‘Integration’, p. 353). The ‘regard of others’ is the concern of some for what they regard as fair outcomes for others, whether or not the regarding parties are themselves directly affected102 (Zerbe, 1998, ‘Integration’, p. 354).103 It follows from the definitions of these terms that a ‘good’ is often defined by the ‘regard of others’ (Zerbe, 1998, ‘Integration’, p. 354). Because any standard of economic efficiency compares states of the world as they differ

152

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

with respect to some good (Zerbe, 1998, ‘Three Rules’, pp. 422–423), calculations of efficiency must consider goods defined by the regard of others. The change in the distribution of wealth caused by an allocative change is often a good, defined by the regard of others. Indeed, as long as people care about ‘the income distribution’, that distribution must itself be a good (Zerbe, 1998, ‘Three Rules’, p. 426).104 Therefore, distribution must be incorporated into a calculation of the efficiency of the allocative change from which it arises (Zerbe, 1998, ‘Three Rules’, p. 426). Thus, the regard of others often requires, contrary to long-standing tradition, that distributional effects be a necessary part of the consideration of economic efficiency (Zerbe, 1998, ‘Three Rules’, p. 426; Zerbe, 1998, ‘Integration’, p. 354). To properly incorporate such effects into efficiency calculations, it is necessary to to be able to value a change in the distribution of wealth brought about by a change in the allocation of goods. Such a change in the distribution of wealth may be said to have a value equal to the difference between the aggregate willingness to pay for that change, WTP, and the aggregate willingness to accept that change, WTA (Zerbe, 1998, ‘Three Rules’, p. 426).105 Where KHZ efficiency is defined as a Kaldor-Hicks efficiency standard that incorporates the willingness to pay for and the willingness to accept a change in the distribution of wealth caused by a given allocative change in the calculation of the efficiency of that change, the conditions necessary to such efficiency may be stated as follows: an allocative change is KHZ efficient if the difference between the immediate gainers’ gain, G, and the immediate losers’ loss, L, plus the value of the change in income distribution, R, is positive. That is, (G-L) + R > 0 → KHZ efficiency, where R = WTP-TA, or substituting, (G-L) + (WTP-WTA) > 0 → KHZ efficiency. The variables G and L are intended to represent the gain and loss caused by the allocative change, independant of any consideration of the value of the change in the distribution of income. Thus, the G-L term is intended to represent the traditional Kaldor-Hicks efficiency calculation. The WTP-WTA term is intended to represent the additional distributional component of KHZ efficiency, that is, the valuation of the change in the distribution of wealth caused by the underlying allocative change. Given this definition of KHZ efficiency, the incorporation of Aristotleian distributive justice into the calculation of such efficiency may be described. To

Of Distributive Justice and Economic Efficiency

153

adequately incorporate a consideration of distributive justice into a KHZ efficiency calculation two things are necessary. First, the WTP must be the willingness to pay for a change in the distribution of wealth induced by a change in the allocation of goods. We will call this WTPDJ. Second, and similarily, the WTA must be the willingness to accept a change in the distribution of wealth caused by a change in the allocation of goods. We will call this WTADJ. Therefore, the conditions necessary to what is here termed integrated efficiency may be summarized as follows: (G-L) + (WTPDJ-WTADJ) > 0 → integrated efficiency. This leaves only the ‘common law’ component of the fundamental proposition undefined. To facilitate a comparison between the traditional and KHZ efficiency theories of the common law, the common law will be defined in the same manner that it is in the traditional theory.105 That is, ‘common law’ means ‘th[ose] fields of law that have been created largely by judges as a by-product of deciding cases’ (Posner, 1992, Economic Analysis, p. 31).106 Therefore, the KHZ theory of the common law holds that those fields of law which have been created largely by judges as a by-product of deciding cases are, in the aggregate, best explained as a system designed to promote both total and distributive efficiency, where a given allocative change is KHZ efficient if (G-L) + (WTPDJ-WTADJ) > 0. 2. Evaluation There is obviously no existing commentary on the integrated theory. Nevertheless, the literature extant may be reviewed for suggestions as to that theory’s validity. For example, as an initial matter, it seems clear that the integrated theory will be subject to many of the same ineffectual criticisms that have troubled the traditional theory.107 Nevertheless, it seems equally apparent that at least one criticism, the distributive justice critique, will be entirely inapplicable to the integrated theory. Indeed, as argued above, that ‘critique’ suggests the superior explanatory power, and hence the validity, of KHZ. To see this, recall the distributive justice critique of the traditional theory, and consider its corollary, the distributive justice argument for the integrated theory. The critique argues that because judges seek to do justice, where justice is both corrective and distributive, and traditional Kaldor-Hicks efficiency either ignores or undercuts distributive justice, judicial behavior, and hence the common law itself, cannot be completely described in terms of traditional Kaldor-Hicks efficiency. This suggests a limit on the traditional theory’s descriptive power.

154

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

It is a limit not applicable to KHZ. Indeed, according to what may be termed the distributive justice argument for the integrated theory, because the common law was shaped by both traditional efficiency and distributive justice concerns, it is likely to be better described in terms of a theory that considers both concerns as opposed to a theory which considers only the former. Therefore, the common law is likely to be better described by the integrated KHZ theory than by the traditional efficiency theory. Hence, an evaluation of the literature extant may be said to suggest the validity of the integrated theory and the invalidity of the traditional theory. D. Conclusion This result can, perhaps, be better understood in the light of recapitulation. Although both the traditional and integrated theories hold that the common law is best explained as a system designed to promote economic efficiency, they differ dramatically in their respective conceptions of efficiency. Specifically, although the traditional efficiency theory holds that the common law may best be described in terms of traditional Kaldor-Hicks efficiency, the integrated theory maintains that the common law is better described in terms of integrated efficiency. It is from this difference that larger distinctions arise. Because traditional Kaldor-Hicks efficiency erroneously mandates that the economic efficiency of an allocative change must be measured independently of its effect on the distribution of wealth, it can result in either the neglect or abrogation of distributive justice.108 Conversly, KHZ efficiency explicitly incorporates a consideration of the distributive justice of a given allocative change into its calculation of the efficiency of that change. Therefore, the fact that judges seek to do justice, where justice is, in part, distributive, suggests that the common law would be better described by KHZ than by traditional Kaldor-Hicks efficiency, and, hence, the validity of the integrated theory and the invalidity of the traditional theory. It remains only to test that suggestion.

III. A TEST OF THE THEORIES THROUGH A SURVEY OF THE COMMON LAW A. Introduction That suggestion will be tested through a two-step process. First, the concept of ‘descriptive capacity’, the most appropriate standard by which to evaluate

Of Distributive Justice and Economic Efficiency

155

theory in this case, will be more specifically developed.109 Second, the descriptive capacity of both theories will be compared through a three-part survey of common-law doctrine, which will broach the law of crimes, contracts, and torts.110 That survey will begin with criminal law, and particularly, the law of rape.111 Rape has been illegal in most cultures for most of their histories. Nevertheless, under the traditional theory, it will be shown that at least some rapes should be legal. Conversely, it will be demonstarted that under KHZ all rapes should be illegal. Therefore, rape law will be shown to be better described by KHZ than by its traditional counterpart. Next, contract law will be considered, and particularly, exculpatory clause jurisprudence. Although such jurisprudence generally holds that exculpatory clauses purporting to exclude liability for negligence are unenforceable, only the KHZ reaches this result. Therefore, KHZ will be shown to provide a better description of exculpatory clause jurisprudence than the traditional theory does. Last, the tort law movement toward contribution among negligent joint tortfeasors will be considered. This movement is inexplicable under the traditional theory, but predictable under KHZ. Therefore, it will be shown that contribution jurisprudence is also better explained by KHZ than by the traditional efficiency theory. Given that the law of rape, exculpatory clauses, and contribution are better described by KHZ than by its traditional counterpart, the former, ceteris paribus, provides a better description of the common law than the latter. Indeed, if further research supports this result, the traditional Kaldor-Hicks efficiency theory should be replaced by KHZ’s integrated theory of the common law. B. The Evaluation of Theory Although other tests are used to evaluate theory, the most common test, and, indeed, the one most often cited by economic analysts of law, is the ability of the theory in question to explain the reality of its subject phenomenon (Mansfield, 1994, p. 14).112 That is, to be valid, the theory ‘must correctly describe how the [particular aspect of] the world [it purports to explain] works’ (Byrns & Stone, 1992, p. 15).113 This test will be termed the test of ‘descriptive capacity’. It implies that the only way in which to invalidate an existing theory is by presenting evidence that an alternative theory better explains the subject phenomenon than does the existing theory.114

156

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

C. A Test of Descriptive Capacity Through a Survey of the Common Law 1. Criminal Law: The Case of Rape a. Introduction The most infamous result of traditional economic analysis115 is that some rapes are arguably efficient, and thus inexplicably illegal.116 Because the possibility of efficient, and therefore legal rape is contrary to existing laws,117 such laws are glaring and, potentially invalidating, anomalies of the traditional efficiency theory.118 No comparable anomaly arises under KHZ, however, and it will, therefore, be demonstrated that rape law is better explained by KHZ than by traditional KH efficiency. This demonstration will proceed in four steps. First, the law of rape will be briefly recapitulated. Second, the susceptibility of rape to market analysis will be considered. Third, the traditional Kaldor-Hicks efficiency theory of rape law will be reviewed and discussed. Finally, the integrated theory of rape law will be contrasted, and an appropriate conclusion drawn. b. Rape Jurisprudence Although the prohibition against rape is now defined statutorily, it is of common law origin.119 In early English law, as in ancient codes’, the legal treatment of rape depended on the victim’s relationship with a man’ (Dripps, 1992, pp. 1780–1781). Thus, the ‘rape of a virgin was punishable by castration and blinding’ (Dripps, 1992, p. 1782, citing Bracton, 1968, pp. 414–415), while that of non-virgins was a matter relegated to ‘the local feudal courts or private vengeance’ (Dripps, 1992, p. 1782). Ultimately, however, the common law came to protect all women against rape (Dripps, 1992, p. 1782),120 a crime it defined as ‘the carnal knowledge of a woman forcibly and against her will’ (Blackstone, 1900, p. 210).121 One of the most notorious elements of the common-law treatment of rape was the marital exemption, by which ‘a husband who forced his wife to engage in sexual intercourse with him was not guilty of rape’ (Dressler, 1987, p. 516, citing Hale 1664, p. 628).122 This exemption is indicative of the the common law’s more problematic justification for the rape prohibition itself: the conception that certain women are the property of certain men (Dressler, 1987, p. 520). Specifically, daughters were held to be the property of fathers, and wives that of husbands (Dressler, 1987, p. 520). Thus, at common law, rape was a ‘property offense’, by which the rapist stole ‘the property rights of the father or husband’, who were, consequently, considered the true victims of rape (Dressler, 1987, p. 520).123

Of Distributive Justice and Economic Efficiency

157

The repugnance of this conception to contemporary sensibilities124 erupted as the rape reform movement of the 1970s, and by the mid-1980s ‘nearly all states had enacted some form of rape reform legislation’ (Spohn & Horney, 1992, p. 20). Many of these states replaced the common-law based crime of rape with ‘a series of gender-neutral graded offenses’ (Spohn & Horney, 1992, p. 22).125 Despite such statutory renovation, however, the fact remains that, the law has historically and continues to punish at least those acts defined at common law as rape (Dripps, 1992, p. 1783).126 An efficiency theory should be able to explain the current state of rape law. c. The Susceptibility of Rape to Market Analysis Because any economic analysis of rape assumes the existence of a market for sexual services,127 the susceptibility of rape to such analysis has been the subject of considerable debate.128 However, it is important to note that there does not need to be a literal ‘sex market’ for economic analysis of rape to be possible; all that economics requires is that the metaphor of a market helps one describe and interpret the behavior of the actors (Zerbe, 2001).129 Furthermore, the market does not have to be defined as one for sexual services. For example, one can posit a market for sexual intimacy or romantic happiness. Moreover, speaking of sexuality in market terms does not imply that actors choose sexual partners in a calculated exchange of money for pleasure, or power for beauty (Zerbe, 2001). However, many writers who have conducted an economic analysis of rape are speaking of ‘sex markets’ in fairly literal senses (Posner, 1992, Sex). Most proponents of an economic analysis of rape have endorsed the ‘commodity theory’ of sexual relations.130 This theory holds that ‘sexual cooperation is a service much like any other’ (Dripps, 1992, p. 1786).131 Specifically, it assumes that each individual has a ‘property right. . .to his or her own body and [thus, to] his or her own sexual service’ (Dripps, 1992, p. 1786).132 As a result, such services can be ‘traded in social or intimate markets in exchange for some reciprocal bundle of goods. . .toward the personal and societal end of maximizing wealth’ (West, 1993, p. 2430). According to Posner, the commodity theory’s leading proponent,133 and contrary to the feminist prospective,134 ‘[r]ape is simply a substitute for consensual sex, engaged in by normal . . . heterosexuals for whom the cost of consensual heterosex is simply too high’ (West, 1993, p. 2421, citing Posner, 1992, Sex, pp. 106–107, 182–183, 384–385).135 This proposition ‘suggest[s] that a rational model of ‘normal’ human behavior can be used to analyze the behavior of rapists (Posner, 1992, Sex, p. 386),136 and implies a putative economic explanation for the illegality of rape’ (West, 1993, p. 2430).

158

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

According to that explanation, in the absence of market failure, the market assures efficient transactions (Cooter & Ulen, 1997, pp. 37–42), but efforts to bypass the market can result in inefficient transactions.137 As a result, ‘[a]ttempts to bypass the market [must be] discouraged by a legal sanction bent on promoting efficiency’ (Posner, 1985, p. 1195). Because ‘[r]ape bypasses the market in sexual relations’, it ‘should [therefore] be forbidden’ (Posner, 1992, Economic Analysis, p. 218).138 According to this conception, ‘the average forcible rapist is in effect a sex thief’ (Posner, 1992, Sex, p. 182), and thus, ‘rape is the theft of one’s sexuality’ (West, 1993, p. 2430). Opponents of economic analysis of rape argue that such ‘market rhetoric’ is improper because sexual services are incommensurable with market commodities (Radin, 1987, pp. 1921–1925). In other words, sexual services, on the one hand, and all other goods on the other, are not ‘comparable in terms of their value’ (Duxbury, 1995, pp. 657, 660).139 Therefore, no one measure can be used to value sexual services and all other goods, and insofar as economic analysis seeks to ‘evaluate [both types of] goods along a single metric’, it is inapplicable to the problem of rape (Duxbury, 1995, p. 674). Indeed, for those who accept the incommensurability of sexual services, the phantom of efficient rape vanishes.140 As will be explained further, efficient rape is conceivable only when it is deemed feasible to compare the value of sexual services taken by the rapist, with the loss incurred by the rape victim from having such services taken without his or her consent (Posner, 1992, Economic Analysis, p. 218).141 If sexual services are deemed incommensurable, however, it would be impossible to make such comparisons.142 Thus, proponents of incommensurability ‘would not presume collectively and objectively to value the cost of a rape to the victim against the benefit to the rapist, even if economic efficiency’ were the ‘sole motive’ (Calabresi & Malamed, 1972, p. 1125).143 Nevertheless, the real question is not what consequences follow from a conclusion of incommensurability, but whether that conclusion is, itself, proper (Duxbury, 1995, p. 678).144 There are at least three general theories of the propriety of the incommensurability of sexual services. First, Radin argues that ‘[t]hinking of rape in market rhetoric implicitly conceives of as fungible something we know to be. . .too personal even to be personal property’ (Radin, 1987, pp. 1921–1925).145 Thus it is argued that sexual services are so intimately interwoven with the person as to be inseparable therefrom, and thus incommensurable (Radin, 1987, pp. 1921–1925).146 The problem with this theory is that it amounts to a tautology by which anything called ‘personal’ is automatically deemed ‘incommensurable.147 The real question is why sexual services should be called personal at

Of Distributive Justice and Economic Efficiency

159

all.148 Until this question can be satisfactorily addressed, the theory itself must be considered inadequate.149 A second theory of incommensurability holds that, given the existing distribution of wealth, commodification of sexual services is morally problematic (West, 1993, pp. 2431–2432; Chamallas, 1988). For example, West (1993, pp. 2431–2432) argues that given the current ‘political and societal hierarchical structures’, a ‘woman’ may be induced ‘to consent to sex, that while not rape, might very well be unwanted, unenjoyed, invasive, painful, demeaning, and dehumanizing’.150 If such sex arises through a market transaction, it is, according to the commodity theory, nevertheless legal (West, 1993, p. 2431). Thus, it is argued, ‘the commodity theory of sex . . . legitimate[s] apparently consensual transactions, even in circumstances of grossly unequal distributions of sexual power’ thereby ‘perpetuat[ing] our collective blindness to the pervasive systems of sexual coercion that render all of our heterosexual practices, and not just rape, morally suspect (West, 1993, p. 2431, emphasis added).151 There are at least two problems with this theory. First, and foremost, its conclusion is irrelevant (Landes & Posner, 1987, p. 9).152 The law itself makes many immoral sexual acts legal and thus, arguably legitimate.153 The fact that a theory posited to explain that law reaches the same conclusion is not an indictment of that theory but an endorsement.154 Second, because there is no necessary correlation between law and morality, a theory which pronounces a given activity legal does not necessarily pronounce it moral or otherwise legitimate.155 Moreover, characterizing an activity as ‘efficient’ is not a moral justification either (Zerbe, 2001). It is possible, and even likely, for an immoral law to be efficient if the society’s regard for others is sufficiently misinformed or unenlightened (Zerbe, 2001). Put bluntly, a sexist law will usually be efficient in a sexist society. The solution is not to pretend that a sexist law is inefficient, but to attack – and thereby change – the social norms (Zerbe, 2001). A third theory of incommensurability holds that market valuations of sexual services degrade human sexuality itself.156 This is based on a misunderstanding of what it means to study sexuality from market perspective: a misunderstanding shared by both Posner (1992, Economic Analysis) and his critics.157 Contrary to Radin’s (1987, pp. 1921–1925) suggestion, speaking of sexuality in economic terms is not mere ‘rhetoric’ to justify the exploitation of women by men. Rather, an economic analysis of rape allows one to clarify (and quantify) what is gained and lost by the rapist, the victim, and society as a whole.158 An economic analysis of sexuality is only degrading if it ignores or undermines the importance of emotional, political, spiritual, or romantic concerns. It is

160

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

tempting (and customary) to speak of these things as noneconomic concerns. However, properly understood, an economic analysis of sex will incorporate all values that affect a person’s behavior in a sex market. Using economics to discuss people’s sexual behavior does not imply that they accept or reject potential partners in an emotionally detached or reductionist way. Since most people’s ‘willingness to accept’ a sexual partner depends on more than the other person’s beauty, money and power, an economic analysis of sex must incorporate a broader range of values. Furthermore, a properly conducted economic analysis of sex will incorporate the community’s regard for others (Zerbe, 2001). Therefore, an economic analysis of sex can only be degrading to the extent that the community’s norms are degrading, since an economic analysis of sex must consider every factor that the community views as important.159 d. The Traditional Kaldor-Hicks Efficiency Theory of Rape Law The traditional Kaldor-Hicks efficiency theory of rape law starts from the proposition that ‘the substantive doctrines of the criminal law. . . promote efficiency’ (Posner, 1985, p. 1194). Given low transactions costs, ‘the market is, virtually by definition, the most efficient method of allocating resources’ (Posner, 1985, p. 1195).160 Therefore, the traditional theory holds that ‘the major function of the criminal law. . . is to prevent people from bypassing the ‘market,’. . . in situations where transactions costs are low’ (Posner, 1985, p. 1195). In consequence, the criminal law forbids, among other things, inefficient transactions that occur outside of the relevant market, or ‘pure coercive transfers of wealth’ (Posner, 1992, Economic Analysis, pp. 217–220).161 Because rape ‘[b]ypasses the market in sexual relations’, it is often a pure coercive transfer, and should, therefore, be criminally forbidden (Posner, 1992, Economic Analysis, p. 218).162 Because rape has been a crime in most cultures throughout their histories, the traditional Kaldor-Hicks efficiency theory of rape law initially seems consistent with that law and thus, at least in this sense, valid.163 The traditional theory encounters difficulty, however, in explaining the prohibition against so-called ‘efficient rape’, that is, rape which is not a pure coercive transfer of wealth, but an impure coercive transfer with a positive social welfare effect.164 It is important to note that while Posner coined the phrase ‘efficient rape’, he is skeptical that efficient rapes exist, even in the narrow sense of the word (1985, p. 1199). Judge Posner, among others, attempts to justify the law’s prohibition of such rapes on three major grounds.165 First, Posner (1985, p. 1199) seems to argue that the opportunity cost of the activity commonly called efficient rape is high enough to make such

Of Distributive Justice and Economic Efficiency

161

rape inefficient. Second, Posner (1992, Sex, pp. 386–387) states that allowing so-called efficient rape ‘would not really be utility-maximizing, if only because of the fear it would engender in the community as a whole and the expense of the self-protective measures that this fear would incite’.166 Third, Landes & Posner (1987, p. 158) argue that the measurement problems in comparing the hypothetical offer price of the rapist with the hypothetical demand price of the victim ‘are overwhelmingly difficult and hardly worth undertaking to identify the rare rape that may in some sense be thought to increase social wealth’. Posner’s (1985 , p. 1199) first argument has two problems. First, since utility is not measurable – as Posner (1985, p. 1197) himself recognizes – it is impossible to compare the ‘opportunity cost’ of rape with the utility to the rapist or its disutility to the victim. Since Posner provides no means of measuring the gains or losses associated with rape, it is impossible to determine whether the opportunity cost of rape is ‘too high’. More to the point, Posner’s (1985 , p. 1199) opportunity cost argument depends on the existence of a sufficient market substitute for rape. Posner (1992, Economic Analysis, p. 142) argues that there are a variety of markets which furnish low cost substitites for rape, such as prostitution and dating markets. To those who argue that there is no market substitute for rape, since an essential element of pleasure for a rapist is violent subjugation, Posner (1985, p. 1199) makes two replies. First, Posner argues that many potential rapists, who are actually deterred by the rape laws, do not derive the bulk of their pleasure from violence, and therefore can find a market substitute for rape (indeed, if the rape law deterred them and they aren’t currently celibate, they did find a market substitute). Second, Posner (1985, p. 1198) argues that even rapists who prefer nonconsensual sex to consensual sex probably receive some pleasure from consensual sex. Therefore, dating and prostitution are market substitutes for rape even if they aren’t perfect substitutes (Posner, 1985, p. 1198). Posner’s (1985F, p. 1198) first defense of his statement that there are market substitutes for rape is highly speculative, since it is impossible to know the psychological traits of people who would commit rape if it wasn’t illegal. In any event, there is a wealth of psychological evidence that an essential element of the pleasure of rape to people who actually commit rape is the forcible subjugation of the victim (Schwartz, 1979, p. 806). Posner’s (1985 , p. 1198) second argument for the existence of market substitutes strains the concept of market substitutes past its breaking point. At some point a substitute is so poor that forcing somebody to rely on it deprives him or her of almost as much wealth as depriving them of the good altogether. Although there may be poor substitutes for it, rape as unwanted sexual violence has no good market substitute (Schwartz, 1979, p.

162

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

806), as Posner (1992, Economic Analysis, p. 218) himself acknowledged in an earlier work. Posner’s (1992, Sex, p. 386–387) second argument, that even efficient rape is ‘not really utility maximizing’, is unconvincing, because Posner does not provide any criteria to help us test or evaluate the claim. Just as the utility of a rapist and disutility of the victim cannot be compared, the disutility of rape for the community as a whole cannot be measured. While it is true that ‘the crime of rape can be supported by the idea that its deterrent effect spares [potential victims] the discomfort of worrying about being subjected to rape at some later time’ the law also ‘deprives potential rapists of the pleasures of anticipation’ (Schwartz, 1979, p. 807, note 33).167 Frankly, ‘there is insufficient economic reason for assuming that the reduced apprehension is greater than the deprived anticipation’ (Schwartz, 1979, p. 807, note 33). The source of the problem is that Posner (1992, Sex, p. 386) doesn’t view this argument as ‘economic’ at all, but as a ‘utilitarian’ argument. In fact, the effect of rape on the community as a whole can be studied economically under KHZ, through the concept of the regard for others (Zerbe, 2001). Once the effect on the community is recognized as an ‘economic’ concern, it can be tested and evaluated in a meaningful way. Posner’s third argument, that efficient rapes should be illegal because the transaction costs for identifying efficient rapes would be too high, cannot be dismissed lightly (Landes & Posner, 1987, p. 158). It is likely true that the cost of adjudicating claims of efficient rape would be high, precisely because it is problematic to compare the rapist’s utility with the victim’s disutility. However, a court might develop a low cost ‘rule of thumb’ to identify rapes that are likely to be efficient. For example, the ancient common law rule allowing husbands to rape their wives may have been such a low cost rule of thumb for determining efficiency, as we discuss below (Dressler, 1987, p. 516). More importantly, it is unsatisfying to say that the primary reason for outlawing efficient rapes is that they are costly to identify. Intuitively, it simply does not feel right to say that the reason the law condemns efficient rapes is because efficient rapes are too hard to identify. By using KHZ efficiency analysis, we can explain this intuitive sense of inadequacy in economic terms. Like Posner’s other explanations for rape law, it leaves open the possibility of the act of rape itself being efficient. Under KHZ, rape is never efficient, at least in modern American society, so the problem of efficient rape disappears. e. KHZ Theory of Rape Law A KHZ analysis of rape shows that the act of rape itself will be almost invariably inefficient given current social attitudes, even in the rare (and

Of Distributive Justice and Economic Efficiency

163

arguably nonexistent) case where the rapist’s pleasure exceeds the victim’s pain, and not just because of transaction costs. Posner (1992, Economic Analysis, p. 1197) compares the ‘utility’ of rape to the rapist with the ‘disutility’ of rape to the victim, which is problematic because ‘utility’ is not empirically testable. KHZ’s analysis of rape differs from Posner’s (1992, Sex) in two regards. First, KHZ compares the rapist’s pleasure and the victim’s pain by comparing the WTP of the rapist (WTPr) with the WTA of the victim (WTAv). No one has attempted to study WTPr and WTAv, but unlike ‘utility’ these two criteria could be measured.168 Second, and more importantly, KHZ considers the economic effect of rape on society as a whole (through the regard for others: R) as well as its effect on the immediate actors. For the act of rape to be efficient, WTPr must be higher than the sum of WTAv and R. In other words, WTPr > WTAv + R → Rape is efficient. Therefore, even in the rare (and arguably nonexistent) case in which WTPr was higher than WTAv, rape would be inefficient if R was greater than the difference between WTPr and WTAv. There is a wealth of anecdotal evidence concerning R, so R may be reasonably estimated.169 Rape is emphatically wrong by all popularly held moral standards in America.170 Indeed, American society’s disgust with rape is so intense that some segments would like to see some convicted rapists put to death.171 Posner (1992, Sex, pp. 386–387) himself does not question the validity of the social conviction that rape is morally reprehensible, he simply regards this as economically irrelevant.172 Given how widely and roundly rape is condemned, R may safely be presumed to be a very large number, probably approaching infinity.173 Therefore, for rape to be efficient, WTPr must be higher than WTAv by an almost infinite amount. In other words, for rape to be efficient, the victim would have to feel only the slightest discomfort, while the rapist would have to feel an epiphany of perfect delight. As a practical matter, given current norms, rape is KHZ inefficient even when WTPr was higher than WTAv, so the failure to make an exception for these ‘efficient’ rapes is also efficient, even without considering transaction costs. Conversely, some rapes could be efficient in a society which tolerates rape, since the results under KHZ depend on the sentiments of a community. For example, consider the ancient common law rule’s exemption for husbands (Dressler, 1987, p. 516). Under the norms prevailing during the Middle Ages, a woman’s sexuality was regarded as the property of her husband or father (Dressler, 1987, p. 520).174 A woman who shared the norms of her age might

164

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

not feel a psychological entitlement to her sexual autonomy, and thus would experience rape in WTP, not WTA terms. On the other hand, the husband might feel ‘entitled’ to rape his wife, and might experience rape as a change in wealth in WTA, not WTP terms. Most communities during the Middle Ages seemingly had little or no regard for women’s sexual autonomies, and did not experience a ‘loss’ if their autonomies were violated by their husbands, so R would be nearly zero (Dressler, 1987, p. 520). In such a sexist society, rape might be efficient whenever WTAr > WTPv. Since WTA is frequently higher than WTP, rape might be efficient on a fairly frequent basis. Therefore, the ancient common law exemption may have been efficient (Dressler, 1987, p. 520). Of course, the ancient common law rule may simply reflect men’s power, as well as their indifference towards or ignorance of the interests of women. The fact that the rape law of the Middle Ages may have been KHZ efficient shows that KHZ does not always show us what we should do (Zerbe, 2001). It simply allows us to evaluate the efficiency of a rule, given current social attitudes (Zerbe, 2001). KHZ can lead to horrifying results when a community has horrifyingly ignorant attitudes (Zerbe, 2001). Furthermore, it is not unreasonable to argue that regardless of law and custom, women have always felt ownership of their bodies and of their sexual autonomies. If so, they would experience the loss associated with rape in a WTA sense. This would lead to the conclusion that the ancient custom was inefficient, since its prediction that a husband’s gain was lower than his wife’s loss would frequently be incorrect. The eventual abandonment of the exemption for husbands can be explained by a change in the regard for others. When social attitudes about a woman’s right to her sexual autonomy changed, communities did experience a loss when a woman was raped by her husband (Spohn & Horney, 1992, p. 20). When the regard for others changed, the exemption was no longer efficient (or it was even more glaringly inefficient), and so the exemption was largely abandoned (Spohn & Horney, 1992, p. 20), just as KHZ would predict.175 Thus, efficient rapes have not been invariably illegal. When there were communities where efficient rape was possible (or when male law makers thought efficient rape was possible), an exemption from rape liability was recognized. The common law exemption also had a low transaction cost, since it would be simple to determine if a rapist was a woman’s husband. This shows that, contrary to Landes & Posner’s (1987) suggestion, high transaction costs will not necessarily prevent a society from making an exemption for so-called ‘efficient rapes’, since a society may use easily administered ‘rules of thumb’ as proxies for efficiency.

Of Distributive Justice and Economic Efficiency

165

Posner (1992, Economic Analysis, pp. 218–219) discusses the ancient common law’s exemption for husbands, but he apparently does not view it as an ‘efficient rape’. Rather, Posner (1992, Economic Analysis, pp. 218–219) suggests that marital rape wasn’t really rape at all in traditional communities: marital rape was at most a ‘dilute[d]’ form of rape. In essence, marriage was a market in which men offered their services as protectors and providers, and women offered their services as sexual partners (Posner, 1992, Economic Analysis, pp. 218–219). While women did not consent to be raped, they consented to a social order in which fairly frequent marital rapes were inevitable, and where marital rape was unpunished (Posner, 1992, Economic Analysis, p. 219).176 While Posner (1992, Economic Analysis, p. 218) acknowledges that there is a difference between having the right to ‘demand something’ and having ‘the right to take it by force’, he urges that the sense of impropriety is ‘dilute[d]’ in such cases.177 For Posner, wealth maximization apparently refers both to Kaldor-Hicks efficiency and to increasing productivity; indeed, in some points in his writing he incorrectly views Kaldor-Hicks efficiency and productivity as synonymous.178 Therefore, Posner (1992, Economic Analysis, pp. 218–219) believes that rape law, and, indeed, all law, is most efficient when it maximizes productivity. In traditional communities, the most prized goods that women produced were sexual and procreative services (Posner, 1992, Economic Analysis, p. 218). A law which prevented men from raping their wives would reduce the productivity of women in that community (Posner, 1992, Economic Analysis, p. 218). The law of rape changed when women’s productivity changed (Posner, 1992, Economic Analysis, p. 219). When increased numbers of women began working, their productivity increased and expanded beyond providing sexual and procreative services (Posner, 1992, Economic Analysis, p. 219). When women’s productivity increased they became wealthier as a class, and therefore women were less willing to sacrifice sexual autonomy in return for protection (Posner, 1992, Economic Analysis, p. 219). KHZ would similarly expect women’s WTP for marital rape to increase as women became wealthier, since WTP, to a greater extent than WTA, is dependent on the actor’s wealth. The problem with Posner’s (1992, Economic Analysis, pp. 218–219) discussion of the marital exemption is that it is incomplete, because it ignores the roles of psychological reference points and the regard for others. Productivity is not synonymous with KH efficiency (let alone KHZ), since it does not recognize the psychological difference between gains and losses. The fact that women became more productive (at least in a materialistic sense) when they entered the workforce in larger numbers may explain why more

166

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

women were ‘willing to pay’ for less abusive husbands, but it does not explain why the legal rules were changed. In fact, even if the rules for rape hadn’t been changed, we would expect a higher percentage of working women to avoid marital rape by ‘buying’ nonabusive husbands. Posner (1992, Economic Analysis, pp. 218–219) may mean that women were not only individually willing to pay for nonabusive husbands, but collectively willing to pay for a change in the social order, but it is not clear that Posner is saying this. Assuming Posner (1992, Economic Analysis, pp. 218–219) is saying this, his argument is incomplete: the same norm that legitimated marital rape would have excluded women from the workforce. For women’s productivity to change, the norms regarding women’s roles as producers would have to change as well. In contrast, KHZ sees the change in rape law as reflecting a change in norms, as well as a change in women’s bargaining position. KHZ does not quarrel with Posner’s (1992, Economic Analysis, pp. 218–219) suggestion that women’s willingness to tolerate abuse – individually and collectively – changed when they became less dependent on men for financial support. KHZ simply points out that the change in norms was an important factor in changing the law. Indeed, there was most likely a complex relationship between women’s increased presence in labor and political markets and the collapse of the norm which excluded women from the workforce and which tolerated marital rape. When the norms changed, rape was no longer seen as something that a woman had to ‘bargain’ to avoid. Marital rape was seen as a ‘bad bargain’ for the entire community, because the entire community suffered a loss when a woman was raped by her husband. Therefore, the community made marital rape illegal without inquiring into the husband’s pleasure or the wife’s pain, because no amount of pleasure on the husband’s part would overcome the community’s sense of loss. When the community’s norms were different, marital rape may not have been a bad bargain for the entire community, and it may have been KHZ efficient to leave marital rape up to the market: in other words, to place the burden on women to avoid marital rape by ‘purchasing’ non-abusive husbands. Thus, KHZ efficiency theory is better able to explain rape law than its traditional counterpart. In modern American society, all rapes are inefficient under KHZ, while some may be efficient under KH. Furthermore, KHZ offers a better explanation than Posner does as to why the law used to allow exemptions from rape liability, but does not do so (to the same extent) today. The result is that, ceteris paribus, the criminal law is better explained by KHZ than by traditional Kaldor-Hicks efficiency (Schwartz, 1979, pp. 812–813).

Of Distributive Justice and Economic Efficiency

167

2. Contract Law: The Case of Exculpatory Clauses a. Introduction An exculpatory clause is ‘a contract clause that [purports to] release one of the parties [to the contract] from liability for his or her wrongful acts’ (Black’s Law Dictionary 1990).179 Exculpatory clauses, which putatively relieve a party from liability for that party’s own negligence, form another body of jurisprudence better explained by KHZ than by traditional Kaldor-Hicks efficiency. This will be shown in three stages. First, the relevant law will be reviewed. Second, the traditional Kaldor-Hicks efficiency theory of that law will be explored. Finally, the integrated KHZ theory of exculpatory clause jurisprudence will be compared. b. Exculpatory Clause Jurisprudence In general contract law favors enforcement of agreements without judicial review of either the overall fairness or individual terms of the exchange’ (Schell, 1993, p. 433).180 Nevertheless, there are times when ‘a court will decide that the public interest in freedom of contract is outweighed by some other public policy, and will refuse to enforce the agreement or some part of it on that ground’ (Farnsworth, 1982, pp. 52–53).181 One of the most important public policies under which contracts have been held unenforceable is that against the commission or inducement of torts (See Restatement of the Law, Second - Contracts. 2d §§192 and 194; Farnsworth, 1990, pp. 352–3, 1982, p. 332). It is under this policy that courts have held certain exculpatory clauses unenforceable.182 While it is true that the enforceability of exculpatory clauses varies with the nature of the tort liability such clauses purport to excuse, only clauses purporting to excuse a party from liability for personal injuries resulting from its own negligence will be considered here.183 Such clauses are generally held unenforceable at least if [1] the parties did not have equal bargaining power or [2] the public interest is involved’ (Eisenberg, 1993, p. 110; Talbott, 1988, p. 30, citing Schlobohm v. Spa Petite, 326 N.W.2d 920, 925–26 (Minn, 1982)).184 The public interest has been held to be involved in services provided by ‘[c]ommon carriers, hospitals and doctors, public utilities, innkeepers, public warehousemen, employees, and [in] services involving extra-hazardous activities’ (Talbott, 1988, p. 30).185 In other recreational situations such as ‘auto racing, gymnasium[s] and health club[s], spa[s] and gym[s], sky diving, and horse and saddle rental’, exculpatory clauses excusing the party from liability for personal injury caused by that party’s negligence have been upheld (Talbott, 1988, p. 30).186

168

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

c. The Traditional Kaldor-Hicks Efficiency Theory of Exculpatory Clause Jurisprudence There is a paucity of literature regarding the traditional Kaldor-Hicks efficiency of clauses seeking to excuse liability for personal injuries caused by negligence.187 However, it seems clear that, in the absence of third-party effects or other market failures, the enforcement of a contract ‘makes [the parties to that contract] better off, as measured by their own desires, without making anyone worse off’ and is therefore, traditionally Kaldor-Hicks efficient.188 Consequently, the nonenforcement of certain exculpatory clauses can only be explained under the traditional theory as a response to market failure and indeed, the rule has been rationalized as necessary to protect consumers against overreaching monopolists.189 According to this rationalization, a promisor individual is unable to effectively bargain with a promisee institution, and, as a result, must either agree to exculpate the promisee for its negligence in the performance of the underlying service, or go without that service altogether (Posner, 1992, Economic Analysis, p. 114).190 Therefore, it is argued, society is forced to endure a super-competitive price for the underlying service, a price equal to the explicit monetary price for that service plus the value of the exculpatory clause, and this, by definition, is traditionally Kaldor-Hicks inefficient.191 As a result, nonenforcement of exculpatory clauses might seem efficient where such clauses appear in contracts for the sale of goods and services in monopolized industries. Indeed, this result would be consistent with the current legal standard, which makes inequality of bargaining power a consideration in enforceability.192 Thus, the traditional efficiency theory seems to explain why exculpatory clauses excusing a party from liability for personal injuries resulting from its own negligence are held unenforceable in at least some cases.193 The problem with this conclusion is that the assumption upon which it relies is too often false. Although many exculpatory clauses are formed between an individual promisor and an institutional promisee, it does not follow that the institutional promisee is always a monopolist, or even that it exercises significant market power with respect to the terms of the sales contract. Indeed, if there is any competition among institutional promisees in their respective product markets, as there undoubtedly is between most common carriers, hospitals, doctors, inkeepers, public warehousemen, and perhaps even public utilities, such institutional promisees will have an incentive to engage in nonprice competition.194 Thus, ‘[i]f one [promisee] offers unattractive [contract] terms, a competing [promisee], wanting sales for [her- or] himself, will offer more attractive terms’, and this process ‘will continue until the terms are optimal’, i.e. until the competitive price is reached (Posner Economic Anlaysis,

Of Distributive Justice and Economic Efficiency

169

p. 114).195 Thus, because there is competition among the institutional promisees in their respective product markets, such products will, ceteris paribus, be sold at competitive prices and the use of exculpatory clauses will merely be a component of such prices. Therefore, exulpatory clauses are, at least, theoretically the result of competition rather than monopoly, and because they are a part of a contract that ‘makes [its parties] better off, as measured by their own desires, without making anyone worse off’ they are traditionally KaldorHicks efficient (Cooter & Ulen, 1997, p. 167).196 Thus, the nonenforcement of exculpatory clauses cannot be explained economically as a response to market failure. Indeed, such nonenforcement is entirely inexplicable under the traditional Kaldor-Hicks efficiency theory.197 d. An Integrated Theory of Exculpatory Clause Jurisprudence Such nonenforcement may, however, be explained under an integrated approach. Recall that, broadly speaking, the integrated efficiency standard, unlike its traditional counterpart, incorporates the change in the distribution of income caused by an allocative change into the calculation of the efficiency of that change.198 Because KHZ efficiency is here being used as a proxy for true integrated efficiency, the relevant change in the distribution of income has a value equal to the difference between the aggregate willingness to pay for that change, WTP, and the aggregate willingness to accept it, WTA (Zerbe & Dively, 1994, p. 80). Therefore, an allocative change is, at least, KHZ efficient, if the difference between the immediate gainers’ gain, G, and the immediate losers’ loss, L, plus the value of the change in income distribution, R, is positive.199 That is, (G-L) + R > 0 → KHZ efficiency, where R = WTP-WTA, or substituting, (G-L) + (WTP-WTA) > 0 → KHZ efficiency. To determine if a rule against the enforcement of the relevant exculpatory clauses is KHZ efficient, ‘we must decide on the starting point, on the status quo position, in order to determine for whom the WTA applies and to whom the WTP applies’ (Zerbe, 1998, ‘Three Rules’, p. 432f). Because a rule against the enforcement of the relevant exculpatory clauses was determined to be traditionally Kaldor-Hicks inefficient, the KHZ analysis presented here will start with a presumption of enforceability, and consider the KHZ efficiency of a change to unenforceability, that is, to the current rule. From the above discussion of the traditional theory, it is clear that, when the relevant product market is competitive, the quantity (G-L) must be positive.200

170

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

Hence, assuming competition, a prohibition against the relevant exculpatory clauses will only be KHZ inefficient if R, the value of the change in income distribution, is a negative number of greater magnitude than the quantity (G-L). There is, however, good reason to believe that R is actually positive. That is, there is good reason to believe that there is large willingness to pay for the change in income distribution that results from a change from the enforcement to the nonenforcement of the relevant exculpatory clauses. If such clauses were enforced, they would create what Talbot (1988, p. 40) describes as a ‘benefit spreading effect’, that is, ‘they would have the effect of replacing a relatively small number of relatively large benefits (payments of compensation to the injured) with a relatively large number of relatively smaller benefits (to all [promisors] who execute the waiver)’.201 Because ‘there is a strong sense that something is wrong with [an] outcome in which the uninjured benefit, while the injured are left substantially uncompensated’, the vast majority of society is probably willing to pay at least a small amount to insure that the injured receive just compensation (Talbott, 1988, p. 35).202 This the majority can do by prohibiting the enforcement of exculpatory clauses. For, as Talbott (1988, p. 32) stated, such a prohibition spreads costs by ‘replac[ing] relatively large costs to a relatively small number of people (those who are actually injured) with relatively small costs to a relatively large number of people (in this case, all [promisors]’. Hence, R is probably positive, and the prohibition against exculpatory clauses is KHZ efficient. Because the nonenforcement of exculpatory clauses which purport to relieve a party of liability for that party’s own negligence is KHZ efficient, but traditionally Kaldor-Hicks inefficient, the jurisprudence of such exculpatory clauses is better explained by integrated than traditional Kaldor-Hicks efficiency. The ultimate result is that, ceteris paribus, contract law is better explained by integrated than traditional efficiency. 3. Tort Law: The Case of the Movement Toward Contribution Among Negligent Joint Tortfeasors a. Introduction ‘[T]he modern movement[] to substitute. . . contribution for no contribution among [negligent] joint tortfeasors’ is among ‘[t]he most important counterexamples. . . to the [traditional Kaldor-Hicks] efficiency theory of the common law’, for although that theory posits that no-contribution is more efficient than contribution, the movement toward contribution has continued unabated (Posner, 1992, Economic Analysis, p. 255).203 Thus the contribution rule movement is a troubling paradox for the traditional efficiency theory, and

Of Distributive Justice and Economic Efficiency

171

makes the traditional theory vulnerable to invalidation by a theory with better explanatory power.204 However, under KHZ the paradox disappears. Indeed, the movement toward contribution seems KHZ efficient. It will therefore be argued that the movement from no-contribution to contribution, and hence, contribution jurisprudence itself, is better explained by integrated than by traditional efficiency. That argument will proceed in three steps. First, contribution jurisprudence will be summarized. Second, the traditional Kaldor-Hicks efficiency theory of that jurisprudence will be reviewed. Last, KHZ’s view of contribution will be presented. b. Contribution Jurisprudence The law of contribution exists because injuries are often caused by the negligence of more than one actor (Kornhauser & Revesz, 1989).205 In such circumstances, the common law sometimes rendered the actors jointly and severally liable, and, accordingly, dubbed them ‘joint tortfeasors’.206 When negligent actors are held jointly and severally liable for an injury, the injured plaintiff may proceed jointly against all the injurers (Stanley, 1994), and, if successful in obtaining a judgment, satisfy it in whatever proportions he or she chooses. Alternatively, the plaintiff may recover all damages from only some or one of the injurers (Landes & Posner, 1980).207 If the plaintiff chooses the latter approach, the question of whether to allow contribution arises (Cooter & Ulen, 1997, p. 301). This subsection will summarize contribution jurisprudence in three steps. First, what has been termed the general common law rule against contribution will be discussed (Leflar, 1932). Second, the movement toward contribution among negligent joint tortfeasors noted by Posner (1992, Economic Analysis, p. 255) will be described. Finally, the effect of contribution on pre-trial settlement will be considered.208 Contribution must be distinguished from indemnity.209 Keeton et al. (1984, p. 341) define contribution as ‘an order distributing loss among tortfeasors by requiring. . . each [non-paying tortfeasor] to pay a proportionate share to one who has discharged their ‘joint’ liability’.210 While it is often flatly said that the common law allowed no right to contribution among negligent joint tortfeasors, it seems that, at least prior to 1799, and probably as late as 1894,211 the common law did allow contribution (Rose, 1979). Unfortunately, that rule was founded in perspicacity (Keeton et al. 1984, p. 337),212 and was ultimately lost to confusion (Cavanagh, 1987; Landes & Posner, 1987, p. 204).213 The source of that confusion, and the origin of what is commonly termed the ‘general rule’ at common law (Higgenbotham & Wiggens, 1990, pp. 700–701), was the English case of Merryweather v. Nixan (Reath 1898).214 Merryweather, itself, which involved an intentional tort,215 held only that contribution is not

172

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

permitted in cases of intentional misconduct (Cavanaugh, 1987, p. 1285).216 It thus stated an exception to what was actually the general rule that contribution was allowed among negligent joint tortfeasors.217 For a time, later English cases continued to recognize the distinction between intentional and negligent torts by ‘permitt[ing] contribution among tortfeasors as long as the [underlying] act was unintentional and not malicious’ (Schwartz, Simpson & Arnold, 1979, p. 782).218 Similarly, the early American cases applied a rule against contribution to cases involving intentional torts, but allowed contribution in those involving negligence (Keeton et al. 1984, p. 337).219 Nevertheless, whether through an erroneous construction of Merrywhether or a misguided expansion of its holding (Cavanaugh, 1987, p. 1285), the distinction between negligent and intentional was gradually lost, and a rule barring contribution in all tort cases was erected in its place (Higgenbotham & Wiggins, 1990, p. 701; Schwartz, Simpson & Arnold, 1979, p. 782; Stanley, 1994, pp. 3–4).220 The majority of American courts proceeded to apply this rule, refusing to allow contribution even between negligent joint tortfeasors (Keeton et al. 1984, p. 337; Easterbrook, Landes & Posner, 1980).221 Indeed, the rule became so firmly entrenched that, until the 1970s, for a period of more than a century, only nine American jurisdictions allowed contribution among negligent tortfeasors (Keeton et al. 1984, pp. 337; Schwartz, Simpson & Arnold, 1979, p. 782).222 Despite its prevalence, however, the rule against contribution was vigorously attacked for its apparent inequity.223 As Fleming (1983) observed, given the link forged by notions of natural justice between tort liability and fault, ‘it was difficult to resist the demand for permitting contribution between tortfeasors’ as a means of ensuring ‘a ‘fair’ distribution of the loss in accordance with. . . responsibility’.224 In consequence, many courts, although constrained by stare decisis, strained to avoid inequity while retaining the underlying rule.225 Gradually, however, the rule began to be discarded (Schwartz, Simpson & Arnold, 1979, p. 784). First courts, (Kornhauser & Revesz, 1989, p. 841, note 48)226 and then legislatures, began to permit contribution between negligent tortfeasors.227 Indeed, England itself allowed for contribution in 1935,228 and ‘[i]n recent years the trend, both of the legislation and of decisions in absence of it, has been toward recognition of the right of contribution’ (Restatement of the Law, Second, of Torts. 2d §886A, comment a).229 This trend, called ‘the movement toward contribution’ by Landes & Posner (1987, p. 219), is illustrated in the graph below.230 In an effort to approach to the trend to abandon the common-law rule against contribution consistently (Higgenbotham & Wiggens, 1990, p. 702; Schwartz, Simpson & Arnold, 1979, p. 784), several model statements of the law have been drafted (Kornhauser & Revesz, 1989, p. 841, note 48). Among these are

Of Distributive Justice and Economic Efficiency

173

the Uniform Contribution Among Tortfeasors Act,231 the Uniform Comparative Fault Act,232 the Restatement of the Law, Second, of Torts233 and the Restatement of Restitution.234 Although variations of each appear among state statutes, the ‘approaches that the states have taken to [contribution] are remarkably fragmented’ (Eggen, 1995, pp. 1701–1702).235 In general, however, the existing ‘judicial and statutory rules are similar in that they all provide for contribution if three elements are present’: (1) a common liability on the part of the parties to the action, (2) a legally mandated discharge of that liability, and (3) the placement of an unequal portion of that liability upon the party seeking contribution (Schwartz, Simpson & Arnold, 1979, p. 786).236 One of the most difficult issues in contribution jurisprudence is that of the effect of contribution on the pre-trial settlement of litigation (Keeton et al. 1984, p. 340).237 While a settling defendant is, by virtue of covenant with the original plaintiff, usually protected from efforts by that plaintiff to collect any part of a related subsequent judgment (Easterbrook, Landes & Posner, 1980, p. 333), that defendant may remain subject to a claim for contribution by joint tortfeasors bound by the judgment (Fleming, 1983, p. 235).238 The dilemma is that by allowing contribution, settlement may be discouraged,239 but by prohibiting it, equity is impaired.240 Thus, a hydra thrives at the heart of the settlement-contribution nexus, meeting any gain in judicial economy with an equivalent sacrifice of distributive justice.241 The common law seems to have favored equity, for the majority rule was that a settling defendant is not released from contribution (Keeton et al. 1984, p. 340), and, indeed, this remains the rule in England today (Fleming, 1983, p. 235, citing Dutton v. Bognor Regis, Q. B. (1972, pp. 373, 399). However, the statutes are divided, ‘without any semblance of consensus’ (Restatement of the Law, Second, of Torts.2d comment m). The 1935 Uniform Contribution Among Tortfeasors Act allowed contribution, but the 1955 revision took the opposite approach (Fleming, 1983, p. 235).242 Restatement of the Law, Second, of Torts refuses to take any position.243 Yet, in part because of the troubling conundrum of settlement effects, the law of contribution remains dynamic and controversial.244 Under traditional efficiency theory, it remains something of a mystery. c. The Traditional Kaldor-Hicks Efficiency Theory of Contribution Jurisprudence That mystery stems from paradox, for while there is an undeniable trend toward contribution in Anglo-American law, the traditional efficiency theory would have predicted the opposite result (Landes & Posner, 1987, pp. 27–28; Posner, 1992, Economic Analysis, p. 255).245 Indeed, according to that theory,

174

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

contribution among negligent tortfeasors246 is patently inefficient.247 It is thought to be inefficient, even though both contribution and no-contribution can, under certain conditions, ‘provide incentives for efficient accident avoidance (Landes & Posner, 1987, p. 201),248 because contribution is both ‘more costly to administer’ (Posner, 1992, Economic Analysis, p. 189), and, at least when tailored to maximize its equitable effect, a significant disincentive to pre-trial settlement (Landes & Posner, 1987, p. 202; Stanley, 1994, p. 6). Thus, a movement from no-contribution to contribution would increase administrative cost, but have no effect on the level of accident deterrence, and would, accordingly, be traditionally Kaldor-Hicks inefficient.249 This subsection will explain that result in two steps. First, the deterrent effects of no-contribution and contribution will be considered. Second, the administrative costs of each will be compared. In considering the deterrent effect of the rules of no-contribution and contribution, several important assumptions are traditionally made.250 With respect to the actors themselves, it is generally assumed that (1) all actors have perfect information,251 (2) the defendants are solvent,252 (3) the actors are risk neutral,253 and (4) the actors are rational.254 With respect to legal background principles, it is similarly assumed that (1) a negligence regime prevails255 (2) ‘negligence and due care are correctly defined from an economic standpoint’ (Landes & Posner, 1987, p. 197),256 (3) liability is joint and several,257 (4) a rule of contributory negligence is employed (Landes & Posner, 1987, p. 194),258 (5) where liable, an actor is responsible for the full loss caused by its actions (Di Cola, 1992, p. 1556),259 (6) liable actors pay for damages that are attributable to non-negligent actors (Di Cola, 1992, pp. 1556–57; Kornhauser & Revesz, 1989, pp. 841–843),260 and (7) there is no legal error in deciding cases.261 A further assumption pertains to the type of activity giving rise to the underlying tort. All activity which may result in a tort may be classified as either a ‘joint care’ (Landes & Posner, 1987, pp. 60, 190–91)/‘bilateral precaution’ (Cooter & Ulen, 1997, p. 275) case or an ‘alternative care’ (Landes & Posner, 1987, pp. 60–61, 191, 198)/‘unilateral precaution’ case (Cooter & Ulen, 1997, p. 275, 302).262 In the former, efficiency requires both the victim and the potential injurers to take precaution against an accident.263 In the latter, alternative care/unilateral precaution case, ‘optimal accident avoidance requires that only one potential injurer take care’.264 Since the question of employing contribution or no-contribution only arises in the joint care/bilateral precaution case,265 the efficiency of these rules will be considered under an assumption of bilateral precaution.266 Given these assumptions, both qualitative and quantitative analyses become possible (Balkin, 1987). In the joint care/bilateral precaution case, under a

Of Distributive Justice and Economic Efficiency

175

negligence regime,267 ‘the no-contribution rule, by imposing residual liability on each [negligent injurer], creates the optimal incentive for each. . . to take precaution’ (Cooter & Ulen, 1997, p. 302).268 To see why, qualitatively consider the case of two potential injurers.269 If either fails to satisfy the legal standard of care, while the other does not,270 it will be responsible for the entire expected social cost of any resulting accident (Cooter & Ulen, 1997, pp. 302–303).271 Since the cost of compliance with the legal standard is, by definition, less than the expected social cost (Landes & Posner, 1987, p. 196; Kornhauser & Revesz, 1989, pp. 847–848),272 however, at least one of the potential injurers will always have an incentive to comply, and if one has an incentive to comply, it follows that the other must as well (Landes & Posner, 1987, p. 196). Thus, ‘when the negligence standard is combined with the no-contribution rule in a case involving joint care, all parties have incentives to select levels of care that minimize the total cost of an accident’.273 This proposition may also be proven more formally through consideration of the expected social cost function. The expected social cost of an accident has been defined as the sum of the cost of precaution plus the cost of expected harm from an accident (Cooter & Ulen, 1997, p. 271), and, hence, the expected social cost function expressed as follows: Cs(x, y, z) = p(x, y, z)D + A(x) + B(y) + C(z) where Cs = expected social cost of an accident as a function of the parties’ levels of care; x = potential injurer A’s level of care; y = potential injurer B’s level of care; z = victim C’s level of care; p(x, y, z) = probability of an accident as a function of the parties’ levels of care; D = victim C’s damages from an accident; A(x) = A’s cost of care; B(y) = B’s cost of care; C(z) = C’s cost of care; x* = level of care of A that minimizes Cs; y* = level of care of B that minimizes Cs; and z* = level of care of C that minimizes Cs (Landes & Posner, 1987, pp. 193–194). Because the two274 potential injurers are assumed to be rational, each will choose and employ a level of precaution which minimizes its total expected loss,275 that is, the sum of its expected losses from liability plus the cost of precaution (Landes & Posner, 1987, p. 195; Cooter & Ulen, 1997, p. 273). In a joint tort, under a rule of no-contribution, the victim plaintiff determines, by choice of suit, the allocation of damages, D, between A and B, ex post (Easterbrook, Landes & Posner, 1980, p. 347), and hence, that allocation is ex ante uncertain (Landes & Posner, 1987, pp. 194–195). Therefore, A’s total expected loss may be expressed as sAp(x, y, z*)D + A(x)

176

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

and B’s total expected loss expressed as sBp(x, y, z*)D + B(y), where sA = the product of the probability that C sues both A and B and the expected fraction of the subsequent damages imposed on A given that both are sued; sB = the product of the probability that C sues both A and B and the expected fraction of the subsequent damages imposed on B, given that both are sued (Landes & Posner, 1987, p.194; Easterbrook, Landes & Posner, 1980, pp. 345–346);276 z*’.x < x *; and y < y* (Landes & Posner, 1987, p. 195). Given that all actors are assumed to have perfect information, sA + sB = 1.277 Assuming that sA and sB are positive, each potential injurer’s total expected loss will vary with the level of precaution chosen by the other injurer.278 Thus, with respect to potential injurer A, if B is expected to and does employ precaution y*, A’s total expected loss will be either A(x*) if A is non-negligent, or p(x, y*,z*)D + A(x) if A is negligent (Landes & Posner, 1987, p. 195). By definition, however, x* is the level of care of A that minimizes Cs (Landes & Posner, 1987, p. 195). Thus, p(x, y*, z*)D + A(x) > p(x*,y*,z*)D + A(x*) > A(x*), and given that A will choose that level of precaution which minimizes its total expected loss (Easterbrook, Landes & Posner, 1980, p. 345; Kornhauser & Revesz, 1989, p. 836). If B is non-negligent, A will choose to be non-negligent and employ precaution x* (Landes & Posner, 1987, p. 195).279 If, however, B is expected to employ precaution y,280 A’s total expected loss will be either A(x*), if A is non-negligent, or sAp(x, y, z*)D + A(x), if A is negligent (Landes & Posner, 1987, p. 195). Thus, given that A will choose that level of precaution which minimizes its total expected loss, A will only choose the negligent level of precaution x if sAp(x, y, z*)D + A(x) < A(x*) (Landes & Posner, 1987, p. 196). B, however, will only select the negligent level of precaution y if sBp(x, y, z*)D + B(y) < B(y*) (Landes & Posner, 1987, p. 196). If (1) sAp(x, y, z*)D + A(x) < A(x*) and (2) sBp(x, y, z*)D + B(y) < B(y*), then (sA + sB)p(x, y, z*)D + A(x) + B(y) < A(x*) + B(y*) (Landes & Posner, 1987, p. 196). Nevertheless, because x* and y* are, by definition, the levels of precaution which minimize Cs, (sA + sB)p(x,y,z*)D + A(x) + B(y) > p(x*,y*,z*)D + A(x*) + B(y*) > A(x*) + B(y*)

Of Distributive Justice and Economic Efficiency

177

(Landes & Posner, 1987, p. 196). Therefore, inequalities (1) and (2) cannot simultaneously hold, and, as a result, even if sAp(x, y, z*)D + A(x) < A(x*),281 sBp(x, y, z*)D + B(y) > B(y*), and, it follows (1) that B will be expected to select the non-negligent level of precaution y*282 and (2) that, as a result, A will also choose the non-negligent level of precaution x* (Landes & Posner, 1987, p. 196). Conversely, if sAp(x, y, z*)D + A(x) > A(x*), given that A will choose that level of precaution which minimizes its total expected loss, A will still choose the non-negligent level of precaution x* (Landes & Posner, 1987, p. 196). The same analysis is applicable to potential injurer B. Therefore, given the assumptions discussed above, under a rule of nocontribution, both potential injurers will always have an incentive to adopt the nonnegligent level of precaution and, hence, a rule of no-contribution will provide incentives for efficient accident avoidance (Posner, 1992, Economic Analysis, p. 188).283 A rule of contribution, however, will provide the same incentives.284 Indeed, it follows from the above analysis of the no-contribution rule that ‘any allocation rule. . . under which the sum of the tortfeasor’s expected shares [sA + sB] in joint care cases sum to one’ will ‘provide incentives for efficient accident avoidance’ (Landes & Posner, 1987, p. 201).285 To see this, consider pro rata contribution, the rule of contribution in some jurisdictions, which requires each joint tortfeasor ‘to pay a pro rata share, arrived at by dividing the [total] damages by the number of tortfeasors’ (Keeton et al. 1984, pp. 340–341).286 Under such a rule, potential injurer A’s total expected loss may be expressed as (1/n)p(x, y, z*)D + A(x) and potential injurer B’s total expected loss expressed as (1/n)p(x, y, z*)D + B(y), where n = the number of potential injurers (Landes & Posner, 1987, p. 201). Recall (1) that because the potential injurers are assumed to be rational, each will choose and employ a level of precaution which minimizes its total expected loss and (2) that each potential injurer’s total expected loss will vary with the level of precaution chosen by the other injurer. With respect to potential injurer A, if B is expected to and does employ non-negligent precaution y*, it follows from the above analysis that A will also employ nonnegligent precaution x* (Landes & Posner, 1987, p. 197). If B is expected to select precaution y, A’s total expected loss will be either A(x*), if A is nonnegligent, or (1/n)p(x, y, z*)D + A(x), if A is negligent. Given that both

178

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

potential injurers are assumed to employ a level of precaution which minimizes their respective total expected losses, A will select precaution x only if (1/n)p(x, y, z*)D + A(x) < A(x*) and B will only select precaution y only if (1/n)p(x, y, z*)D + B(y) < B(y*). Nevertheless, it follows from the above analysis that even if (1/n)p(x, y, z*)D + A(x) < A(x*), (1/n)p(x, y, z*)D+ B(y) > B(y*), and, therefore that (1) B will be expected to select the non-negligent level of precaution y*287 and, as a result, (2) A will also choose the non-negligent level of precaution x* (Landes & Posner, 1987, p. 197; Easterbrook, Landes & Posner, 1980, p. 349). Conversely, if (1/n)p(x, y, z*) D + A(x) > A(x*), given that A will choose that level of precaution which minimizes its total expected loss, A will still choose the non-negligent level of precaution x* (Landes & Posner, 1987, p. 197; Easterbrook, Landes & Posner, 1980, p. 349). A similar analysis is applicable to potential injurer B. Therefore, given the aforementioned assumptions, a rule of contribution will also induce both potential injurers to adopt the non-negligent level of precaution, and hence, the deterrent effects of no-contribution and contribution are identical.288 However, their respective administrative costs are not identical (Landes & Posner, 1987, pp. 201–202; Easterbrook, Landes & Posner, 1980, pp. 349–350). The administrative costs of either allocation rule are determined by both (1) the rate of pre-trial settlement of cases and (2) the costs of those trials which actually occur (Easterbrook, Landes & Posner, 1980, p. 354). Each determinant will be considered first, with respect to no-contribution, and then, with respect to contribution. A no-contribution rule is said to produce incentives to settle (Di Cola, 1992, p. 1555; Stanley, 1994, pp. 6, 59–69). To prove this result, several assumptions are usually made: first, that there are multiple potential defendants, second, that all actors are rational, and third, that all are risk-neutral (Easterbrook, Landes & Posner, 1980, p. 356). Given these assumptions, it may be shown that the victim plaintiff will always have an incentive to settle with at least n ⫺ 1 defendants, where n = the total number of prospective defendants, and, conversely, that n ⫺ 1 defendants will always have an incentive to settle with the plaintiff (Easterbrook, Landes & Posner, 1980, p. 357). Qualitatively, given that the plaintiff is assumed to be risk-neutral, it will prefer to settle with at least one defendant because any positive settlement represents a certain amount which is, by definition, of greater value than an expected equivalent amount (Easterbrook, Landes & Posner, 1980, p. 356). For the same reason, the plaintiff will continue to settle with additional defendants at least until it has settled with all but one (Easterbrook, Landes & Posner, 1980, p. 356). More

Of Distributive Justice and Economic Efficiency

179

formally, in the absence of any settlement, the value of plainitff c’s expected claim, VECc is VECc = pcD, where (1) pc < 1 and (2) p is constant (Easterbrook, Landes & Posner, 1980, p. 357, note 59). Settlement with the first defendant, defendant 1, for S1 > 0 results in VEC1c = S1 + pc(D ⫺ S1), and, given that pc < 1, VEC1c = S1 + pc(D ⫺ S1) > VECc (Easterbrook, Landes & Posner, 1980, p. 357, note 59). As a result, because c is assumed to be rational, c will settle with defendant 1, and after settlement with j–1 defendants, where 1 < j < n, c will have an incentive to settle with j because VECjc ⫺ vECj ⫺ 1c = Sj(1 ⫺ pc) > 0, where Sj > 0 and pc < 1 (Easterbrook, Landes & Posner, 1980, p. 357, note 59). Because c has an incentive to settle with defendants one through j, where j includes all defendants up to n ⫺ 1, c has an incentive to settle with at least n ⫺ 1 defendants (Easterbrook, Landes & Posner, 1980, p. 357, note 59). Therefore, given the aforementioned assumptions, under a rule of nocontribution, the victim plaintiff will always have an incentive to settle with at least n ⫺ 1 defendants, where n = the total number of prospective defendants (Easterbrook, Landes & Posner, 1980, p. 357, note 59). Similarly, given the above assumptions, under a rule of no-contribution, at least n ⫺ 1 defendants have an incentive to settle with c (Easterbrook, Landes & Posner, 1980, p. 357, note 59).289 To see this, assume defendant A’s expected liability upon trial is VELc = 1/(n ⫺ ns)[pc(D ⫺ nsS)], where ns = the number of defendants who have already settled.290 S = the average settlement per defendant. PA = defendant A’s estimate of the probability that c will win at trial (Easterbrook, Landes & Posner, 1980, p. 357, note 60). Because defendant A is, by assumption, rational, an offer to settle for less than VELA, defendant A’s expected liability upon trial, would induce A to settle with c (Easterbrook, Landes & Posner, 1980, p. 357, note 60). The above analysis of plaintiff c’s incentives indicates that c will accept any positive settlement from n ⫺ 1 defendants and since VELA is positive, there must be a set of settlements at which defendants n ⫺ 1 will be induced to settle with c (Easterbrook, Landes & Posner, 1980, p. 357, note 60). Therefore, under a rule of no-contribution, the plaintiff and n ⫺ 1 defendants will have incentives to

180

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

settle rather than proceed to trial.291 Indeed, under a rule of no-contribution, defendants compete to settle because the expected liability of the non-settling defendants increases as each defendant settles.292 The settlement effects of contribution are significantly less sanguine.293 Although many varieties of contribution could be devised,294 ‘two broad classes. . . appear to capture the essential features of contribution’ and thus, provide the subject of this analysis (Easterbrook, Landes & Posner, 1980, pp. 360–361).295 First, a ‘traditional’ contribution rule allows joint tortfeasors who lose at trial to obtain contribution from settling defendants (Easterbrook, Landes & Posner, 1980, pp. 360–361).296 Second, what is known as a ‘settlement bar’ rule ‘forbids nonsettling defendants to obtain any contribution from settling ones’ (Landes & Posner, 1987, p. 203).297 The first rule may be modeled both for the case where contribution is certain and for that in which it is uncertain (Easterbrook, Landes & Posner, pp. 361–363). Where contribution is certain, defendants have no incentive to settle (Easterbrook, Landes & Posner, p. 362; Landes & Posner, 1987, p. 202). If the defendant settles, then its total loss is 1/n.298 If, however, the defendant goes to trial, its total loss will vary with the outcome: if the defendant is successful, its total loss is zero; if, however, the defendant loses, given that ‘pro rata contribution is assured’, its total loss will only be 1/n (Easterbrook, Landes & Posner, 1980, p. 362). Therefore, under a traditional rule of contribution where pro rata contribution is certain, the defendant ‘has nothing to lose and everthing to gain from refusing to settle and going to trial’ (Easterbrook, Landes & Posner, 1980, p. 362). Where pro rata contribution is uncertain, however, the result is more similar to, but nevertheless more administratively costly than, that under the nocontribution rule. Specifically, if contribution is uncertain, then, just as was the case under no-contribution, n ⫺ 1 settlements will be induced (Easterbrook, Landes & Posner, 1980, p. 362). Conversely, defendant n will be less likely to settle under a rule of uncertain traditional contribution than under a rule of no contribution (Easterbrook, Landes & Posner, 1980, p. 362, note 69). Therefore, a traditional rule of contribution is overall less likely to induce settlement than a rule of no contribution and, is consequently, more administratively costly.299 A settlement bar rule need not be (Di Cola, 1992, p. 1556). Indeed, ‘as a first approximation, this rule is equivalent in its effect on settlement to a rule of no contribution’ (Easterbrook, Landes & Posner, 1980, p. 363).300 Nevertheless, some commentators note that a settlement bar rule could quite possibly abrogate the fairness objectives underlying a rule of contribution.301 Therefore, courts which choose to apply such a rule must ‘hold a fairness hearing before approving the settlement’ which would make the settlement bar rule more

Of Distributive Justice and Economic Efficiency

181

administratively costly than the no-contribution rule (Landes & Posner, 1987, p. 203). Because both the traditional and settlement bar versions of the contribution rule either (1) induce a lower rate of pre-trial settlement than a nocontribution rule or (2) require greater administrative expense to produce the same rate, the no-contribution rule is generally considered to be, ceteris paribus, less administratively costly.302 Indeed, unless the cost of trial under a rule of contribution is sufficiently less than that under a rule of no-contribution, the latter may conclusively be said to be less administratively costly than the former (Easterbrook, Landes & Posner, 1980, pp. 353–354). Nevertheless, the opposite is true, because contribution ‘requires the courts to decide another issue and supervise another set of transfer payments’ (Posner, 1992, Economic Analysis, p. 189). Specifically, a rule of contribution would increase trial costs by either bringing additional defendants into the original suit, or by requiring subsequent litigation by which the original defendants may force non-party joint tortfeasors to contribute (Easterbrook, Landes & Posner, 1980, pp. 349–350; Landes & Posner, 1987, pp. 201–202).303 Although it may be argued that a mechanical contribution rule, such as the pro rata rule discussed above, would result in only an insignificant increase in trial costs, such a rule (1) may not actually reduce trial expense and (2) even if properly corrected to do so, would probably raise fairness objections (Landes & Posner, 1987, p. 202). Therefore, it seems that the trial costs are greater under a rule of contribution than under a rule of no-contribution, and, thus, given the foregoing settlement analysis, the latter may conclusively be said to be less administratively costly than the former (Landes & Posner, 1987, p. 201).304 To summarize, although the rules of contribution and no-contribution both produce incentives for efficient accident avoidance, contribution is more costly to administer.305 Thus, according to the traditional Kaldor-Hicks efficiency theory, ‘contribution seems to be a less efficient rule than no contribution’, which raises the question ‘why so many states have abandoned the common law [sic] approach’ of no-contribution (Landes & Posner, 1987, p. 219). It is a question which remains unanswered under the traditional theory; indeed, under this theory it is a paradox without solution.306 d. An Integrated Theory of Contribution Jurisprudence That paradox finds a solution in KHZ’s integrated theory of the common law. Recall that, broadly speaking, the integrated efficiency standard, unlike its traditional counterpart, incorporates the change in the distribution of income caused by an allocative change into the calculation of the efficiency of that change.307 The relevant change in the distribution of income has a value equal

182

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

to the difference between the aggregate willingness to pay for that change, WTP, and the aggregate willingness to accept it, WTA (Zerbe & Dively, 1994, p. 81). Therefore, an allocative change is KHZ efficient, if the difference between the immediate gainers’ gain, G, and the immediate losers’ loss, L, plus the value of the change in income distribution, R, is positive.308 That is, (G-L) + R > 0 → KHZ efficiency, where R = WTP-WTA, or substituting, (G-L) + (WTP-WTA) > 0 → KHZ efficiency.309 To determine if a rule of contribution is KHZ efficient, we must decide on the starting point, on the status quo position, in order to determine for whom the WTA applies and to whom the WTP applies (Zerbe, 1998, ‘Three Rules’, p.432f). Because the focus of concern is the movement toward a rule of contribution from a rule of no contribution among negligent joint tortfeasors,310 that starting point may most appropriately be defined as one in which the rule of allocation is no-contribution. In a movement from a rule of no-contribution to one of contribution, the immediate gainers are those defendants who, unsuccessful at trial, can now obtain contribution from other joint tortfeasors who were not parties to the original suit. Among the immediate losers are those previously non-party joint tortfeasors from whom the unsuccessful defendants may now obtain contribution. Nevertheless, because the unsuccessful defendants gain the same amount as the non-party joint tortfeasors lose, this is merely a transfer of wealth, and, thus, of no consequence to the G-L calculation (Easterbrook, Landes & Posner, 1980, pp. 349–350).311 What is of consequence to that calculation is the loss in judicial economy caused by the increase in administrative cost in moving from a rule of no contribution to one of contribution which may be modeled as L > 0 (Keeton et al. 1984, p. 338).312 Because this loss is not offset by a compensating gain in accident deterrence, G = 0. Therefore, G-L < 0 (Keeton et al. 1984, p. 338). It follows that, for the movement from no-contribution to contribution to be KHZ efficient, R must be a positive number of greater magnitude than G-L. Here, R may be defined specifically as the value of the change in the distribution of wealth brought about by the movement from no-contribution to contribution. That change is a transfer between two parties, both equally responsible for a loss, of one party’s fair share of that loss to another party who has already discharged the entire loss (Keeton et al. 1984, pp. 337–338). The value of that change is, as described above, equal to the aggregate willingness to pay for the change minus the aggregate willingness to accept it. Although, as in the case of rape, there are no empirical studies of either WTP or WTA for

Of Distributive Justice and Economic Efficiency

183

the change in the distribution of wealth induced by a change from nocontribution to contribution, there is a vast amount of anecdotal support for the proposition that, in this case, WTP far exceeds WTA.313 The aggregate willingness to pay for the movement from no-contribution to contribution among negligent joint tortfeasors is almost undoubtedly a very large positive number, perhaps even one approaching infinity. The support for this proposition is ubiquitous.314 For example, Keeton et al. (1984, p. 338) note that the primary impetus for the change from no-contribution to contribution was ‘half a century of vigorous attack upon the original rule’. Prosser (Keeton et al. 1984, pp. 337–338) said this of the no contribution rule: There is an obvious lack of sense and justice in a rule which permits the entire burden of a loss, for which two defendants were equally liable to be shouldered onto one alone, according to the accident of a successful levy of execution, the existence of liability insurance, the plaintiff’s whim or spite, or the plaintiff’s collusion with the other wrongdoer, while the latter goes scot free.

These remarks indicate that the movement from no-contribution to contribution among negligent joint tortfeasors was caused by a popular dissatisfaction with the inequity entailed by a rule of no-contribution.315 This, in turn, is indicative of a large aggregate willingness to pay for that movement, even if the willingness to pay of each individual is relatively slight for, as Zerbe (1998, ‘Integration’, p. 354; 1998, ‘Three Rules’, p. 437f) notes, the concern or regard for others, which underlies the willingness to pay, is a concern for what people regard as fair outcomes for others.316 Thus, the aggregate WTP is probably a very large positive number. The aggregate willingness to accept the movement from no-contribution to contribution is, however, probably a very small positive number. The rule of nocontribution among negligent joint tortfeasors seems virtually bereft of supporters.317 Assuming that what supporters there are are not substantially more wealthy and more adamant than those who oppose a rule of nocontribution, the aggregate WTA is probably very small compared to the aggregate WTP. As a result, (WTP-WTA) is probably a very large positive number and hence, probably a number in excess of G-L. Thus, the movement from no-contribution to contribution, although inefficient under the traditional theory is KHZ efficient, and therefore, KHZ’s integrated efficiency theory is better able to explain contribution jurisprudence than its traditional counterpart. The ultimate result is that, ceteris paribus, tort law is better explained by integrated than traditional efficiency.

184

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

D. Conclusion This section sought to test the distributive justice critiques’s suggestion that the common law is better explained by the integrated efficiency theory than by its traditional counterpart. That test proceeded in two steps. First, ‘descriptive capacity’, the most appropriate standard by which to evaluate theory in this case was explicitly considered. Second, the descriptive capacity of both theories was compared through a three-part survey of common-law doctrine. That survey began with the criminal law of rape, and found that, although under the traditional theory, some rapes should be legal, under the integrated efficiency theory, all rapes should be illegal. Therefore, rape law was shown to be better described by the integrated theory than by its traditional counterpart. Next, exculpatory clause jurisprudence was considered. Although such jurisprudence generally holds that exculpatory clauses purporting to exclude liability for negligence are unenforceable, only the integrated theory reached this result. Therefore, the integrated theory was shown to provide a better description of exculpatory clause jurisprudence than the traditional theory. Last, the tort law movement toward contribution among negligent joint tortfeasors was considered. This movement is inexplicable under the traditional theory, but predictable under the integrated theory. Therefore, it was shown that contribution jurisprudence is also better explained by the integrated than the traditional efficiency theory. Given that the law of rape, exculpatory clauses, and contribution is better described by the integrated theory than by its traditional counterpart, the former, ceteris paribus, provides a better description of the common law than does the latter. Nevertheless, because there is a distinct lack of empirical evidence in this regard, the traditional Kaldor-Hicks efficiency theory should be replaced by the integrated theory only if a sufficient quantum of future research confirms or expands this result.

IV. GENERAL CONCLUSION The purpose of this paper was two-fold. First, it sought to propose an alternative efficiency theory of the common law, deemed the integrated theory. Second, it sought to test the integrated theory’s descriptive capacity against that of its traditional counterpart. Thus, Section II described and theoretically evaluated both the traditional and integrated efficiency theories and suggested that, because only the latter considers distributive justice, it is likely to produce a better description of the common law than does the former.

Of Distributive Justice and Economic Efficiency

185

Section III attempted to test this suggestion in two steps. First, it set forth ‘descriptive capacity’ as the basis for the evaluation of theory. Second, it attempted to compare the descriptive capacity of both theories through a threepart survey of common-law doctrine, which broached the fields of crimes, contracts, and torts. That survey revealed that the law of rape, exculpatory clauses, and contribution is better described by the integrated than by the traditional efficiency theory. It was, therefore, concluded that, ceteris paribus, the former provides a better description of the common law than does the latter. Nevertheless, a distinct lack of empirical evidence was noted. Consequently, only if a sufficient quantum of future research should confirm or expand this result, should the traditional theory be supplanted by its integrated rival, KHZ.318 This task, while perhaps daunting in its practical complexity, would, if successful, provide powerful support for an efficiency theory of the common law.319

NOTES 1. For sources on microeconomics in general, see Katz & Rosen (1998); Kreps (1990); and Boadway & Bruce (1984). With respect to economics in general, see Eatwell et al. (1991). 2. The term ‘common’ law will generally be used to mean that ‘body of law that develops and derives from judicial decisions’. Black’s Law Dictionary 276 (6th ed. 1991) 3. Although the terms ‘model’ and ‘theory’ are ’synonyms for many purposes’, the former is considered to be a ‘representation[] of theor[y]’, and hence, narrower in scope (Byrns & Stone, 1992, p. 16). 4. Yet, in so sweeping a statement resounds an Einsteinian echo: ‘[e]verything should be made as simple as possible, but not more so’. In this regard, Mansfield (1994, p. 14) notes that ‘[t]he trick is to construct a model in such a way that irrelevant and unimportant considerations and variables are neglected, but the important factors –those that have an important impact on the phenomena the model is designed to predict – are included’. One factor of particular importance to a model of common law decisionmaking is the judge’s regard for distributive justice. See part (II)(B)(2)(c). 5. The theory will also, at times, be called simply ‘the traditional theory’. For a general exposition of that theory see, for example, Posner (1992, Economic Analysis). It is apparently not the only theory. Parson (1983) criticizes Posner’s traditional efficiency theory and attempts to offer some solutions. Moreover, there are wellrespected non-economic theories, such as those of Dworkin (1986) and Rawls (1971). 6. This traditional notion has recently been disputed (Zerbe, 2001; Zerbe, 1998, ‘Integration’). We critique the traditional notion in part (II). 7. Posner (1992, Economic Analysis, p. 26) notes that ‘the [traditional Kaldor-Hicks] economic [efficiency] theory of law is the most promising theory of law extant’.

186

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

8. Byrns & Stone (1992, pp. 15–16) note that any theory can be supplanted by one with superior explanatory power. 9. Flatt (1994, p. 95) notes that ‘if society is willing to spend money that has no perceivable effect beyond an alteration of the allocation of a risk, the allocation of that risk must be a value’. 10. Posner (1987 p. 23; 1992, Economic Analysis, p. 264) neglects these concerns. 11. See our discussion in part (II)(B)(2)(c). 12. The term ‘logically’ is used here, and elsewhere, to indicate that the analysis of this section is ‘theoretical’ rather than ‘empirical’. 13. That is, if subsequent research should indicate either (1) that the remaining common-law doctrine is as well explained by the integrated theory as by the traditional theory, or (2) that it is better explained by the former than the latter, the traditional theory must be displaced by the integrated theory. 14. For example, Cooter & Ulen (1997, p. 1) state that ‘[e]conomics has already had a profound impact on legal scholarship’. Cunningham (1992, p. 141) notes that ‘[d]uring its relatively short life span, economic analysis of law has become a dominant player in the intellectual games of legal academia’. For another view, see Dworkin (1986, p. 276). 15. Kuperberg & Beitz (1983, p. 3) note that ‘[t]he economic approach to law is much more than a methodology for analyzing the consequences of judicial decisions. In addition to its methodological elements, the economic approach includes both positive and normative theses’. Posner (1992, Economic Analysis, pp. 21–22) states that ‘the hallmark of the ‘new’ law and economics . . . is the application of economics to the legal system across the board’. Cooter & Ulen (1997, p. 2) have a similar view as to the importance of economics in studying law, but Dworkin (1986, p. 276) disagrees. 16. This term, which will be used interchangeably with ‘the traditional efficiency theory’, is not found in the legal or economic literature. Indeed, Posner & Landes (1987, p. 1) (emphasis in the original) refer to their theory of tort and, by implication, common law as ‘the positive economic theory of tort law because no rival positive economic theory has been proposed’. Because this paper proposes a rival theory, it is necessary to be somewhat more precise in nomenclature; hence, the term ‘the traditional KaldorHicks efficiency theory’. 17. See, for example, Posner (1981, p.775; 1992, Economic Analysis, p. 23); and Kuperberg & Beitz (1983, p. 3). Interestingly, Posner (1992, Economic Analysis, p. 25) argues that the traditional theory is itself merely a subset of a larger ‘economic theory of law’. Posner, 1992, (Economic Analysis, p. 25) notes that the ‘economic theory of law. . . tries to explain as many legal phenomena as possible through the use of economics [while] the efficiency theory of the common law [which is included in that theory] hypothesizes a specific economic goal for a limited subset of legal rules, institutions and so forth’. This paper is concerned only with what Posner calls the efficiency theory of the common law as compared to the integrated theory it seeks to advance. 18. Posner (1992, Economic Analysis, p. 23) concludes that ‘the great common law [sic] fields of property, torts, crimes, and contracts, bear the stamp of economic reasoning’. See also Landes & Posner (1987, p. 1) and Kuperberg & Beitz (1983, p. 775).

Of Distributive Justice and Economic Efficiency

187

19. Posner (1992, Economic Analysis, p. 23) notes that normatively, ‘the economist . . . can . . . clarify a value conflict by showing how much of one value – efficiency – must be sacrificed to achieve another’. 20. Kuperberg & Beitz (1983, pp. 3–4) argue that ‘the normative thesis . . . is more directly related to the future evolution of the law and therefore more important for practicing lawyers and judges’. 21. This focus should not raise methodological objections because ‘the positive and normative theories of the common law. . . are independent of each other’ (Posner, 1981, p. 776). Or as Kuperberg & Beitz (1983, p. 3) say, ‘Neither thesis implies the other; each must stand or fall on its own merits’. 22. This statement is, however, somewhat ambiguous as well, given that the positive theory itself ‘has two aspects’: one ‘about the content of the common law [sic] rules and [another] about the effect of those rules on the people subject to them’; that is, whether such rules actually promote efficient behavior (Posner, 1981, p. 776). This paper will focus on the former although, in so doing, the latter is necessarily implicated. Posner (1981, p. 776) offers evidence that the ‘common law does promote efficient behavior’. 23. See also Posner (1981, p. 777) who notes that ‘[n]o one contends that every rule of the common law is efficient’ and Dworkin (1986, pp. 283–85) who notes that, although ‘no one would claim a perfect fit’ between actual common-law doctrine and traditional Kaldor-Hicks efficiency, ‘[m]any prominent academic lawyers do claim a very substantial fit’ although he cautions that ‘that claim is hotly disputed’. 24. Landes & Posner (1987, p. 1) note that ‘the positive economic theory of tort law’ is ‘that the common law of torts is best explained as if the judges who created the law . . . were trying to promote efficient resource allocation’. Englard (1993, p. 7) notes that, ‘according to the positive efficiency theory of tort law, the ‘traditional rules of tort liability can be explained, notwithstanding different judicial arguments and reasonings, as an approximation to the overriding principle of wealth maximization’. Posner (1975, pp. 757, 763–64) states that ‘the logic of the law is really economics’. 25. Posner (1981, pp. 776–77) notes that ‘several such explanations have been offered’. For some of these explanations, see Posner (1992, Economic Analysis, pp. 254–55, 519–94. However, Zerbe (2001, Chapter 8) argues that the explanations offered as to why the common law should generally be efficient are not compelling because the relationship between norms and efficiency has not been understood and provides an explanation for common-law efficiency that exploits the necessary efficiency of uncontentious norms. 26. Black’s Law Dictionary (1990) defines ‘common law’ as ‘that body of law that develops and derives through judicial as distinguished from legislative enactments’. 27. Zerbe(2001, Chapter 1) notes that the Kaldor-Hicks compensation tests represent what economists and lawyers generally mean when they speak of economic efficiency. Landes & Posner (1987, p.16) state that ‘[w]e use efficiency throughout this book in the Kaldor-Hicks (or potential Pareto superiority) sense’. Posner (1992, Economic Analysis, pp. 13–14) notes that ‘[w]hen an economist says that . . . [some] policy or state of the world is efficient, nine times out of ten, he [or she] means . . . Kaldor-Hicks efficiency’. 28. The Kaldor-Hicks standard is usually compared to Pareto efficiency, under which ‘only those changes are [efficient] in which at least one person is made better off and no one is made worse off’ (Cooter & Ulen, 1997, p. 41).

188

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

29. What is here referred to as the Kaldor-Hicks standard is, technically, two tests not one (Zerbe & Dively, 1994, p. 97). 30. Cooter & Ulen (1997, p. 41) state that ‘a potential Pareto improvement allows changes in which there are both gainers and losers but requires that the gainers gain more than the losers lose’. Landes & Posner (1987, p. 16) note that under the KaldorHicks standard, ‘a policy change is said to be efficient if the winners gain more from the change than the losers lose’. Kuperberg & Beitz (1983, pp. 7–8) hold a view similar to Landes & Posner, but Polinsky (1989) holds a somewhat contrary view. 31. Cooter & Ulen (1997, pp. 41–42) state that ‘[f]or a potential Pareto improvement, compensation does not actually have to be made, but it must be possible in principle’. Kuperberg & Beitz (1983, pp. 7–8) agree. This is, according to Cooter & Ulen (1997, p. 41), in sharp contrast to Pareto efficiency, which ‘requires that gainers explicitly compensate losers in any change’. Indeed, it is because Kaldor-Hicks efficiency only requires that compensation be possible that it is sometimes referred to as potential Pareto superiority (Posner, 1992, Economic Analysis, p. 14). Coleman (1982) criticizes the use of the term ‘potential Pareto superiority’. 32. But, as a matter of logic, this need not necessarily be the case, for it is perfectly possible to have a traditionally Kaldor-Hicks efficient change in which there are no losers. 33. Landes & Posner (1987, p. 16) note that ‘[a] change is wealth maximizing if the dollar value of the gains to the winners is greater than the dollar cost of the losses to the losers’. Posner (1992, Economic Analysis, p. 13), argues that ‘wealth maximization is purportedly identical to Kaldor-Hicks efficiency’. 34. Although the literature is immense, with respect to the normative value of economic efficiency, see, for example, Dworkin (1980, pp. 276–312) or Gadre (unpublished). With respect to the positive efficiency theory, see, for example, Parsons (1983, p. 862), Kornhauser (1980, p. 591) and Tullock (1980, p. 659). 35. Posner (1992, Economic Analysis, pp. 25–27) and Landes & Posner (1987, pp. 9–24) refute many of these criticisms. Indeed, the arguments of Posner & Landes echo throughout the first part of this subsection. 36. This critique will be termed the ‘distributive justice critique’ throughout this paper. 37. This critique is consistent with the traditional understanding of Kaldor-Hicks efficiency, but inconsistent with KHZ, which holds that the traditional understanding of Kaldor-Hicks efficiency is itself flawed. See our discussion in part (II)(C). 38. For example, see Posner (1992, Economic Analysis, pp. 263–64). 39. Shavell (1981, p. 414), Posner, (1992, Economic Analysis, p. 27), Okun (1975) and Polinsky (1983, pp. 110–13) all undercut distributive justice. 40. This list is not exhaustive. 41. Mansfield (1994, p. 13) notes that ‘the trick is to construct a model in such a way that irrelevant and unimportant considerations and variables are neglected, but the important factors – those that have an important impact on the phenomena the model is designed to predict – are included. 42. Mansfield (1994, pp. 15–16) concludes that ‘the choice is not between a model and no model’ but ‘between one type of model and another’.). 43. This paper hopes to suggest that the traditional efficiency theory is no longer the model that predicts best, and hence, that it should be abandoned in favor of an integrated theory. See our discussion in part (II).

Of Distributive Justice and Economic Efficiency

189

44. Landes & Posner (1987, p. 19) state that ‘[w]e feel on fairly safe ground in assuming that judges are good enough agents of society’s dominant groups that if an efficient system of tort law is demanded judges will supply it’. 45. The broader criticism, that by ignoring the other language in opinions, one ignores something ‘essential for determining the actual meaning of the rule that can be extracted from the judicial opinion’, is, of course, more damaging (Landes & Posner, 1987, p. 22). 46. Posner (1992, Economic Analysis p. 23) states that ‘[i]t would not be surprising to find that many legal doctrines rest on inarticulate gropings toward efficiency’. Posner (1992, Analysis, p. 255) also notes that ‘[t]he language of economics is a language for scholars and students, not for the people whose behavior the economist studies’. Cooter & Ulen (1997, p. 379) argue that although ‘judges often prefer more efficient rules. . . their own descriptions employ terms other than ‘efficiency’ for ‘[t]he law embeds efficiency principles under other names’. Landes & Posner (1987, pp. 22–23 note that ‘[p]eople can apply economic principles intuitively – and thus ‘do’ economics without knowing they are doing it’. With respect to the ‘inarticulate gropings’ to which Judge Posner refers, compare Blyth v. Birmingham Waterworks Co., 156 Eng. Rep. 1047 (1856) with U. S. v. Carroll Towing Co., 159 F.2d 169 (1947). Cooter & Ulen (1997, pp. 281–283) note a historical progression toward a negligence standard set at the ‘efficient level of precaution’. 47. Zerbe (2001, Chapter 8) concludes that judges ‘act[ ] in accord with a social norm – the norm that judges dispense justice’. Posner, 1992, Economic Analysis, p. 255) notes that ‘traditional legal scholars. . . think judges should apply principles of justice’. Cooter & Ulen (1997, p. 379) note that ‘it can be argued. . . that judges should allocate legal entitlements fairly’. Landes & Posner (1987, p. 9) somewhat gloomily conclude that ‘[m]ost lawyers and law professors still believe. . . that the actual as well as ideal function of tort law is to achieve fairness rather than efficiency’. See also Dworkin (1986, pp. 225–238) and Rawls (1971). 48. The Kaldor-Hicks efficiency criteria have been largely misconstrued in the law and economics literature. Zerbe (2001, Chapter 1) notes that in large part the controversy surrounding economic analysis of law exists because the application of the criteria has not been well thought out. In part, this paper argues that the proper construction of Kaldor-Hicks efficiency contributes to a superior positive theory of the common law. See our discussion in parts (II)(C) and (III). 49. For example, see Posner (1992, Economic Analysis, pp. 263–64). 50. For example, see Posner (1987, pp. 23–24). 51. Kaufman (1980, pp. 77–80) divides the issue into a normative aspect (‘what judges should do’) and a positive aspect (‘what judges do in fact’). 52. The proposition that judges seek to do justice may be taken as an article of faith by many, without reliance on either theoretical or empirical support. Indeed, it has been noted that the ‘proposition. . . may seem so obvious that discussion is unnecessary’ (Kaufman, 1980, p. 83). Nevertheless, in the interest of intellectual comprehensiveness, such a discussion will be briefly pursued here. 53. Indeed, Posner (1992, Economic Analysis, p. 255) feels ‘traditional legal scholars . . . think judges should apply principles of justice’and Kaufman (1980, p. 83) concludes that ‘the purpose of the law is to do justice’. Cooter & Ullen (1997, p. 379) argue that ‘judges should allocate legal entitlements fairly’. Zerbe (2001, Chapter 8) notes that judges act in accord with a social norm – the norm that judges dispense justice. Calnan

190

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

(1997, p. 19) observes that ‘justice’ is ‘that elusive commodity dispensed by judges in our courts of law’. Landes & Posner (1987, p. 9) note with some disdain that ‘[m]ost lawyers and law professors still believe . . . that the actual as well as ideal function of tort law is to achieve fairness rather than efficiency’. With respect to the last point, however, note that efficiency not only need not, but rather can not, properly exclude justice, as we argue in parts (II)(B) and (III). 54. Kaufman (1980, p. 83) notes that ‘people get upset when judges ignore the prevailing standard of justice in the community’ and concluding that ‘society expects that justice is relevant to the job of judging’. 55. On the actual behavior of judges generally see Ryan et al. (1980) and Glick (1990). 56. This view is not universal. See, for example, Everson (1919) and Blanck (1996). 57. Kaufman (1980, pp. 88–89) dismisses this objection as inappropriate: I believe that justice is more accurately understood when one does not start from a preconceived notion of what it should be . . . Instead, one should begin with the data and try to discover what just explanation there could be for different lines of cases or statutes. Over time, one will hammer out principles of justice from which one can induce what the essential nature of justice is.

58. For example, Calnan (1997, p. 19) states that ‘[t]he confusion surrounding the notion of justice begins with its definition’. Englard (1993, p. ix) notes that there are competing ‘substantive questions of human justice’. Indeed, Posner (1992, Economic Analysis, p. 27) has suggested that one ‘meaning of justice, perhaps the most common, is –[traditional Kaldor-Hicks] efficiency’. 59. Note that Aristotle (Ethics, Bk. V) was here describing what he called ‘justice in the special sense’ rather than justice in the general sense, that is, that form of justice concerning ‘acting fairly’ as opposed to that concerned with ‘the observance of established rules’. 60. Consider also the comments of Confucius (Analects, Bk. 14): ‘[w]hen wealth is centralized, the people are dispersed. When wealth is distributed, the people are brought together’. 61. Aristotle described corrective justice as that form of justice ‘that . . . rectifies wrong in men’s dealings with one another’. Aristotle, supra note 83, at Bk. v. See also Calnan, supra note 83, at 99–104. 62. Calnan (1997, p. 85) describes distributive justice as ‘a blueprint for determining the fairest way to pass out things that many people desire’. 63. For a contemporary definition, see Webster’s (1997). 64. As an illustration, assume that G1 is found to be entitled to 40% of the goods in question and G2 is held to be entitled to the remaining 60%; if 1000 goods are available, then S1 is 400 and S2 is 600 (Calnan, p. 85). 65. Although Aristotle stated that the standard of merit will vary across communities, he argued that merit should be defined with respect to the subject of distribution. Thus, for example, if water rights are the subject of distribution, they should be given to the group best able to employ such rights (Fuller, 1949, pp. 39–40, citing Aristotle Politics). The concept of ‘best able’ is, of course, ambiguous in itself. 66. With respect to the fact that traditional Kaldor-Hicks efficiency ignores distributive justice, see, for example, Posner (1992, Economic Analysis, p. 264). With

Of Distributive Justice and Economic Efficiency

191

respect to its capacity to undercut distributive justice, see, for example, Polinsky (1983, pp. 110–113, 1989, pp. 7–10,119–30). 67. A change in the allocation of goods can, for example, be brought about by the judicial enunciation of a legal doctrine (Schwartz, 1979, p. 801). 68. Schwartz (1979, p. 801) notes that ‘[a]ccording to economic analysis, a legal doctrine can have two kinds of consequences: its contribution to efficiency in the allocation of resources and its impact upon what is referred to as the ‘distribution of wealth.’’ 69. Zerbe (1998, ‘Three Rules’, p. 426) notes that a widespread objection to benefit cost analysis is that it does not consider distributional consequences. See also Zerbe (1998, ‘Integration’). 70. For example, Giancoli (1991, p. 7) notes that ‘[t]o determine (or confirm) the form of a relationship, careful experimentatal meaurements are required’. Babbie (1992, p. 2) states that ‘social research (in fact, all science) is organized around two activities: measurement and interpretation’. 71. For a description of the economic concept of ‘utility’, see Mansfield (1994, pp. 81–84). 72. See the works Zerbe (1998, ‘Integration’, p. 30, note 91) cites, and see Byrns & Stone (1992, p. 121). 73. It is not actually necessary to ignore the distribution effect in calculating the efficiency of an allocative change (Zerbe, 1998, ‘Three Rules’, p. 426f; Zerbe, 1998, ‘Integration’, p. 354), as we point out in part (II)(C). 74. A non-traditional version of Kaldor-Hicks efficiency is presented in part (II)(C). 75. As a consequence of taking the existing distribution of wealth as given, ‘[i]f income and wealth [are initially] distributed differently, the pattern of demands might also be different and efficiency would require a different deployment of our economic resources’ (Posner, 1992, Economic Analysis, p. 14). 76. Posner (1992, Economic Analysis, p. 14), however, argues that although traditional Kaldor-Hicks efficiency does ignore ‘distributive justice’ (1) economists ‘can provide value clarification by showing society what it must give up to achieve a noneconomic ideal of justice’ and (2) ‘[a] second meaning of justice. . . is efficiency’. Indeed, Posner (1992, Economic Analysis, pp. 261–262) notes that traditional efficiency is largely consistent with morality, and argues that ‘on balance it would seem that adherence to generally accepted moral principles increases the wealth of society more than it reduces it’. 77. Posner (1992, Economic Analysis, p. 262) notes that ‘the biggest gap between the common law and the sense of justice is the common law’s seeming indifference to distributive considerations’. 78. Nevertheless, Posner (1992, Economic Analysis, pp. 263–64, 460–6), Landes & Posner (1987, pp.17–19) and Polinsky (1989, pp. 9–10,119–30) all argue that this is not necessarily a normatively objectionable result. ‘Unless a society is permeated by envy, increasing the size of the pie deserves at least as much attention as trying to make the slices more equal’ (Posner, 1992, Economic Analysis, p. 264). Note, however, that ‘equitable’ would be a more precise term than ‘equal’ (Polinsky, 1989, p. 7). See also Cooter & Ulen (1997, p. 104) and Walter J. Blum & Harry Klaven, Jr. (1965, p. 65). Nevertheless, Landes & Posner (1987, p. 9) suggest that such a result could derogate the plausibility of the traditional positive efficiency theory.

192

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

79. Posner (1992, Economic Analysis, p. 263) states that ‘[p]robably the major ethical issue posed by the [traditional] efficiency approach to the common law is the discrepancy between [traditional] efficiency maximization and notions of the just distribution of wealth’. 80. That is, using the above notation, G1 and G2 do not receive S1 and S2 respectively. 81. That is, achieving an efficient allocation of goods. 82. Notice that their argument is not explicitly positive, and hence, not expressly a defense against the positive implications of this premise. 83. The costs of redistribution are most likely positive. Indeed, in a related context, Posner (1992, Economic Analysis, p. 460) notes that ‘if [the costs of redistributing income are] substantial, they could swamp the gains in total utility from the redistribution’. See also Polinsky (1989, p. 9). 84. See, for example, Posner (1992, Economic Analysis, pp. 255, 262–264, 460–461), Polinsky (1989, pp. 9–10, 119–127), Landes & Posner (1987, pp. 17–19), and Cooter & Ulen (1997, pp. 104–05). 85. See, for example, Posner (1992, Economic Analysis, pp. 255, 262–264), Polinsky (1989, pp. 9–10, 119–127), Landes & Posner (1987, pp. 17–19), and Cooter & Ulen (1997, pp. 104–105). 86. See, for example, Landes & Posner (1987, pp. 17–19), Posner (1992, Economic Analysis, pp. 255, 264), Cooter & Ulen (1997, pp. 104–105), and Polinsky (1989, pp. 9–10, 119–130). 87. See also Posner (1992, Economic Analysis, p. 255). 88. See also Polinsky (1989, pp.9–10) and Landes & Posner (1987, pp. 18–19, 313). Cooter & Ulen (1997, p. 105) explain why taxation is considered a superior means of redistribution. 89. See also Cooter & Ulen (1997, p. 105). 90. Note also that this recommendation is normative, and, hence, of no consequence to a defense of a positive theory like the traditional efficiency theory. 91. The second premise is not, however, valid with respect to the definition of efficiency advanced in part (II)(C). 92. This is a matter of elementary logic. 93. See Landes & Posner (1987, p. 24) and the works cited by them (1987, p. 24, note 56), who note that ‘other scholars . . . have found significant support for the positive economic theory of tort law –provided the theory is not conceived as asserting a perfect congruence between law and efficiency’. 94. Thus, it would be the alternative theory itself, rather than the distributive justice critique that actually invalidates the traditional theory, and in consequence, the alternative theory would stand, even if the distributive justice critique ultimately fell. Such an alternative theory will be presented in part (II)(C) and tested against the traditional theory in part (III). 95. This standard was introduced by Zerbe (1998, ‘Integration’, p. 354). 96. It may, however, be more accurate to say that integrated efficiency is merely a special version of KHZ efficiency in which certain conditions hold. 97. We discuss this in part (III). 98. Zerbe (2001, Chapter 2) notes that, what is here termed traditional Kaldor-Hicks efficiency is simply a Kaldor-Hicks efficiency measure that does not incorporate the willingness to pay and the willingness to accept distributional effects.

Of Distributive Justice and Economic Efficiency

193

99. The explanation that follows is a simplified, qualitative version of the more detailed and formal explanation presented by Zerbe (3 Rules). 100. We discuss these effects in part (II)(B)(2)(c). 100. See also Byrns & Stone (1992, p. G–6). 101. There are two types of such regard: (1) ‘self-regarding regard’ and (2) ‘alturistic regard’(Zerbe, 2001, Chapter 3). The former is the concern of some for rules applied to others that arises from a concern that such rules may also be applied to them (Zerbe, 1998, ‘Integration’, p. 354). The latter is the concern of some about the justice of rules applied to others that exists irrespective of a concern that such rules will be applied to them (Zerbe, 1998, ‘Integration’, p. 354). 102. See Zerbe (1998, ‘Integration). 103. Compare our analysis with that of Polinsky (1989, pp. 8–9) 104. The change may be measured as other goods are measured, which is by contingent valuation surveys and by observed behavior. 105. We discussed the traditional theory’s definition in part (III). 106. See also part (II)(B)(1). 107. Many of these are no doubt just as ineffectual when applied to the integrated theory as they were when applied to the traditional theory. See part (II)(B)(2)(b). 108. See, for example, Posner (1992, Economic Analysis, p. 264’). 109. This standard was rather informally developed above, in part (II)(B)-(C). 110. 173. Posner (1992, Economic Analysis, p. 23) notes that ‘the great common law [sic] fields of property, torts, crimes, and contracts bear the stamp of economic reasoning’. Posner (1992, Economic Analysis) presents an analysis of the common law based on traditional Kaldor-Hicks efficiency. 111. ‘Technically, rape was not a common-law crime after the Statutes of Westminster in the reign of Edward I. The second of these, adopted in 1285, made it a felony for a man to ‘ravish’ a woman without her ‘assent’; the common-law definition is a gloss on the statute’. Dripps (1992, p.1780, 1780, note 1). 112. Mansfield (1994, p. 14) notes that ‘the most important test of a model is how well it predicts [its subject] phenomena’. Landes & Posner (1987, p. 22) note that an objection to the traditional theory is that its ‘empirical procedure is rationalization rather than prediction’. But see our discussion in part (II)(B)(2)(b). More generally, see, for example, Friedman (1953). 113. Kaufman (1980, p. 79) concludes that ‘[t]he most important premise of the scientific method is that a theory’s validity is tested by its congruence with reality’. Posner (1992, Economic Analysis, p. 17) notes that ‘[a]n important test of theory is its ability to explain reality’. The ability of the theory in question to explain the reality of its subject phenomenon will be referred to as its ‘descriptive capacity’. 114. For example, Mansfield (1994, pp. 15–16) concludes that ‘if a model is the best available it will – and should – be used until a better model appears’. See also Schwartz (1979, pp. 799, 813), Posner (1992, Economic Analysis, p. 26), Landes & Posner (1987, p. 24); and Kuhn (1970). 115. That is, of the traditional Kaldor-Hicks efficiency theory of the common law. 116. The first efforts to apply economic analysis to rape law were made by Posner (1992, Economic Analysis, pp. 67, 357–358), who originally found rape inefficient, and, thus, rape laws efficient; and by Calabresi (Calabresi & Malamed, 1972, pp. 1089, 1125–1126), who presented an analysis which stopped short of the question of efficiency. In an effort ‘to search for the limitations on the reach of economic concepts’,

194

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

Schwartz was apparently the first to argue that rape could be efficient (1979, pp. 799, 813). A ‘deservedly esoteric literature’ has sprung forth (Posner, 1992, Sex, p. 182). . 117. Posner (1992, Economic Analysis, p. 263) recently notes that ‘common law [sic] courts . . . would have no patience for ‘efficient rape.’’ Indeed, not only is every rape, ‘efficient’ or otherwise, both a crime and a tort in all Anglo-American jurisdictions, but the only court to so much as implicitly mention an economic analysis of rape law did so derisively. See U. S. v. Bishop, 66 F.3d 569 (1995). 118. Of course, the traditional theory’s inability to explain rape law is only a grounds for invalidating it to the extent that KHZ provides a superior explanation. At the close of his discussion of rape, Schwartz (1979, p. 812) noted that ‘[i]n mathematics, an equation can be proved invalid by showing that there is one value of ‘x’ for which the equation does not hold’ and that, ‘[i]nsofar as economic principles are dogmatically advanced as theorems’, the example of efficient rape could ‘have that power’. Since the theory of ‘efficient rape’ is founded on a misconception of Kaldor-Hicks efficiency (see part (II)(C)(1)), it may ultimately fall silent under an integrated analysis: a result, interestingly enough, supported by Schwartz’s (1979, p. 813) own argument. 119. Dressler (1987, p. 515) states that ‘the common-law offense of rape has largely been carried over to modern times by statute’. 120. It is, however, more accurate to say that the common law protected a male’s property interest in a female, or at least in her sexuality. See, for example, Brownmiller (1975, pp. 16–30) and Wald (1997). 121. ‘Carnal knowledge’, in turn, meant ‘sexual intercourse’, which implied ‘genital copulation’, (Dressler, 1987, p. 516, note 1), and more specifically and exclusively, ‘penile-vaginal penetration’ (Spohn & Horney (1992, p. 22). As Dressler (1987, p. 516, note 1) observes, ‘Oral and anal intercourse do not constitute rape although they frequently were and are punishable independently’ and ‘sexual emission is neither sufficient nor necessary’ for criminal liability. Note, however, that this definition appears to be derived from the Statutes of Westminster in the reign of Edward I, adopted in 1285 (Dripps, 1992, p. 1780, note 1, citing Maitland, 1968, pp. 490–91). 122. Shockingly, this exception survives to this day in many states, albeit in a diluted form: in many states, a husband cannot be liable for third degree rape of his wife. See WA statute. 123. Rape law, in essence, ‘struck a balance between the interests of males-inpossession and their predatory counterparts’ (Dripps, 1992, p. 1783). Like ferae naturae, the theft of a woman already reduced to possession was a serious crime; but if a woman broke the man’s possession as an animal might escape a pen, his interest in her lapsed. Thus, the notorious common-law procedural rules requiring prompt complaint, resistance to the utmost, corroboration, and good reputation – worked a compromise between competing male interest. (Dripps, 1992, p. 1783, citing Hale 1664, pp. 633–635).

124. It has, however, been argued that properly recast, a property theory of rape may not be objectionable (Wald, 1997, p. 462). Without ‘advocating a literal treatment of rape as theft’, Wald (1997, p. 462) argues that properly ‘understanding the relationship between property and rights of sexual control would contribute to a fuller understanding of why rape is so prevalent and why it is so injurious to women’.

Of Distributive Justice and Economic Efficiency

195

125. The model of such efforts is the Michigan ‘Criminal Sexual Conduct’ statute. For example, Dressler (1987, p. 535), who discusses Mich. Comp. Laws Ann. §§750.520a – .5201 (West Supp. 1986). 126. Dripps (1992, p. 1783) states that ‘the substantive law has not changed much. Consistent with the reformers’ concern for [sexual] autonomy, the gravamen of rape remains [as it was at common law] the conjunction of force and nonconsent’. Indeed, many states continue to fail to punish husbands for some categories of rape. 127. For the proposition that such a market is fundamental to an economic analysis of rape, see, for example, Posner (1985, pp. 1193, 1199). Posner (1992, Economic Analysis, p. 218) assumes the existence of ‘marriage and sex markets’ comparable to ‘explicit markets in goods and services’. Elsewhere, Posner (1992, Economic Analysis, p. 218), discusses the relationship between rape and a ‘market in sexual relations’ and Posner (1992, Economic Analysis, p. 142) expounds the concept of a ‘marriage market’. Schwartz (1979, p. 805) notes that ‘economic analysis asks us to consider the possibility of transactions – contracts – between rapists and victims’ in a ‘market [that] can provide the appropriate sexual encounter’. 128. Given the magnitude of the debate, a large number of works could be cited here, but the two most prominent, pro and con respectively, are probably Posner (1992, Sex) and Calabresi & Malamed (1972). 129. Under KHZ, a good is anything that people care about, whether or not it is tangible (Zerbe, 2001). Thus economics can explain any interaction in which a person sacrifices a portion of something that they care about in order to acquire something else that they care about even more, at least at the margin. For example, there probably aren’t any literal ‘friendship markets’, but friendship is susceptible to economic analysis (Zerbe, 2001). People sacrifice time (something they undoubtedly care about) in order to be with another person and enjoy their company (something else most people care about). This is not to say that we view our friends as mere instruments for maximizing our own wealth, or that we approach the ‘friendship market’ with the same attitude that we approach markets for tangible goods. Indeed, part of the value of friendship would disappear if we viewed it from a reductionist perspective. However, by spending time with one friend instead of another, we are demonstrating a preference for that friendship, whether we consciously realize it or not. Similarly, the fact that we spend time with a friend instead of spending our time on a ‘nonfriendship good’ (such as extra work on a professional project), demonstrates the relative value we attach to ‘friendship goods’ (such as companionship and intimacy) as opposed to ‘nonfriendship goods’ (such as our professional reputation and money). . 130. For example, Dripps (1992, p. 1786) dubs the theory of a market for sexual services ‘the commodity theory’. West (1993, pp. 2413, 2430) calls the sexual market theory presented by Posner in Sex and Reason, a ‘commodification theory of sex’. 131. There are several distinct versions of what is here termed the market for sexual services. Chamallas (1988, pp. 777, 839–841) argues that ‘ideal sex’ clearly ‘exclude[s] sexual encounters in which money, power, prestige or financial security is traded for sexual pleasure and emotional intimacy’, while West (1993, p. 2430), who describes Posner’s version of the market for sexual services as one in which the commodity of sex may be traded for an unrestricted ‘bundle of goods’. Dripps (1992, p.1790) argues that Chammallas’ objection ‘to any bargaining in which one party offers sexual cooperation in exchange for goods that are distributed unjustly according to gender . . . is both incongruous and morally problematic’. And compare both the Chamallas and West-

196

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

Posner-Dripps versions with Schwartz (1979, p. 805) who discusses the possibility of contracts for ‘consensual’ rape. Indeed, perhaps the debate should start here, for the viability of a market for sexual services would seem to depend heavily on the form of market considered. 132. Locke (1962) argues that ‘[t]hough the earth and all inferior creatures be common to all men, yet every man has a property in his own person’. West (1993, p. 2430) describes Posner’s theory of sex markets. 133. In Zelder’s (1993, pp. 1584, 1601), review of Posner’s Sex and Reason, he concludes that Posner is ‘to be applauded . . . for some worthy attempts to bring economic analysis to bear on the question of rape’. Wohl (1997, pp. 68, 69) notes that Posner ‘is the person most closely associated with the school of law and economics’. 134. There is a vast feminist literature on the topic of rape. Among the most influential works are Brownmiller (1975) and Estrich (1987). Dripps (1992, pp. 1780–83) provides a brief summary of the literature. However, it should be noted that one need not be a feminist to maintain this perspective. 135. Posner (1992, Sex, p. 384) holds that ‘[c]ontrary to the view held by many feminists, rape appears to be primarily a substitute for consensual sexual intercourse rather than a manifestation of male hostility toward women or a method of establishing or maintaining male domination’. It may, however, be more precise to state, with respect to rapists for whom ‘the pleasure of [rape] vitally depends on [their] forcible subjugation of [their] victim[s]’ (Schwartz, 1979, p. 806), that there is no market substitute [for rape]. Schwartz (1979, p. 806) states that ‘even if prostitution [and thus, ‘contracts for coerced sex’] were legal, it would not provide an adequate substitute for rape’, for the vital ‘element of pleasure [to these rapists] is automatically lacking when [their] sexual partner[s are] willing, whether this willingness has been achieved by purchase or otherwise’. Landes & Posner (1987, pp. 1193,1199) insist that ‘there are market substitutes for coercive sexual acts’. 136. Zelder (1993, pp. 1601–1602) discusses the empirical evidence supporting this proposition, noting that Posner presents ‘two primary pieces of evidence concerning the economic rationality of rapists: [1] Issac Ehrlich’s finding that rape incidence is deterred by higher penalties and [2] anthropological data showing that rape incidence rises with bride price’. He argues that ‘[o]nly the former is logically defensible evidence [because] bride price data obviously do not control for the intellectual, emotional, and physical attributes of women in the sample’ (Zelder, 1993, p. 1601–1602). See also Pyle (1983). 137. Calabresi & Malamed (1972, p. 1126) note that ‘in the majority of cases we cannot be sure of the economic efficiency of [market by-passing transfers such as] transfer[s] by theft’. They realize that rape may, in essence, be considered theft (Calabresi & Malamed, 1972, p. 1126). 138. West (1993, p. 2430) notes that the commodity theory . . . implies a particular conception of the wrongness of rape: to take the asset of sex by force instead of by consensual transaction short-circuits the standard market mechanism – a-face-to-face consensual bargain – for ensuring that wealth is maximized.

Schwartz (1979, p. 805) notes that ‘[t]he crime of rape, by further preventing the circumvention of the market process, bolsters the protection against inefficiency’. But see West (1986).

Of Distributive Justice and Economic Efficiency

197

139. Although Duxbury (1995, p. 662, note 12) was discussing the concept of incommensurability generally, not the feasibility of a market in sexual relations particularly, he did note that ‘the representation of women – qua prostitutes, wives, and girlfriends – as a sex market for men’ constitutes ‘an inappropriate valuation’. 140. It is, however, always possible to attack a conclusion drawn from an economic analysis of law by arguing that the subject of such analysis has been improperly commodified (Duxbury, 1995, pp. 677–78). Therefore, a claim of incommensurability is, by itself, unpersuasive. Indeed, the viability of such claims ‘depends on whether [they] embody anything more than diverse individual feelings about what ought not to be bought and sold’ (Duxbury, 1995, p. 678). 141. It is worth noting that Posner’s theory could be used to question the efficiency of any criminal law. For example, an efficient murder is theoretically possible if a murderer’s pleasure exceeded a victim’s disutility at having his or her life taken. Unlike efficient rape, this possibility seems to be more than merely theoretical, since it is easy to imagine a scenario with a sadistic killer and a chronically depressed victim. We will show in part (III)(A)(4) that comparing the criminal and the victim’s disutility is universally problematic, and ignores the social cost of leaving crimes unpunished, due to our regard for others. 142. In terms of the Calabresi & Malamed (1972, p. 1092) framework, there are at least ‘three types of entitlements - entitlements protected by property rules, entitlements protected by liability rules, and inalienable entitlements’. Furthermore, ‘[r]ecognition of incommensurability [leads] to the conclusion that certain rights ought to be protected by neither liability nor property rules, but by inalienability rules’ (Duxbury, 1995, p. 683). An ‘inalienable entitlement’ is, itself, one the transfer of which ‘is not permitted between a willing buyer and a willing seller’(Calabresi & Malamed, 1972, p. 1092). And, of course, if transfers are forbidden per se, no values can be assigned to such transfers, and, consequently, interpersonal comparisons of such values become impossible. 143. It should be noted that such valuation can only occur through use of a ‘liability rule’ which would allow the rapist to ‘destroy the initial entitlement [to, in this case, personal integrity] if he or she is willing to pay an objectively determined value for it’ (Calabresi & Malamed, 1972, p. 1092). Epstein (1997, p. 2091) notes that . . . a liability rule denies the holder of the asset [here, the ‘asset’ of ‘sexual services’] the power to exclude others or, indeed, to keep the asset for [her- or] himself. Rather, [she or] he is helpless to resist the efforts of some other individual to take that thing upon payment for its fair value, as objectively determined by some neutral party.

See also Calabresi & Malamed, 1972, p. 1125; Duxbury, 1995, p. 683; Epstein, 1997. Interestingly, Calabresi & Malamed’s (1972, p. 1125) statement, while perhaps a fair expression of policy preference, is not an accurate account of early law. Instead, it seems that many early legal systems employed liability rules as rape laws. Thus, according to the King James Bible, a ‘man’ found to have raped a ‘virgin’, i.e. an unmarried female, was required to ‘give unto the damsel’s father fifty shekels of silver, and she shall be his wife’ (Deuteronomy 22:28–29); while ‘[a]mong aboriginal peoples much customary law was devoted to the payment of compensation to a man whose legal right to control sexual access to his wife or daughter is disregarded [by, inter alia, the rape of his wife or daughter]’ (Dripps, 1992, p. 1781). 144. Duxbury (1995, p. 678) says, ‘[t]o pose the question bluntly: how might we ever determine that it is inappropriate to value something as a commodity?’

198

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

145. See also Radin’s later work (1993). Note that the term ‘personal property’ is used here by Radin (1993, p. 37) to denote property with which one is ‘bound up such a degree ‘that its loss would cause [one pain] which could not be relieved by replacement of the object with other goods of equal market value’. ‘Fungible property’ refers to anything ‘which is held for purely instrumental reasons’ (Radin, 1993, p. 37). 146. For example, Radin (1987, pp. 1921–1925) concludes that ‘[b]odily integrity is an attribute and not an object’. 147. Note, however, that even this proposition is false, for not everything ‘personal’ is, in fact, deemed ‘incommensurable’.(Duxbury, 1995, p. 678). Duxbury (1995, p. 678), notes that . . . the argument that the peculiarly personal nature of certain things may determine that they ought not to be traded . . . is unconvincing for the single reason that there are plenty of things with a distinctively personal dimension such as our favourite literature, music, restaurants, holiday locations and the like which are very obviously and very naturally commodified.

148. Radin (1993, p. 198) attempts to answer this question by suggesting the possibility of ‘a theory of the good and well developed person . . .to tell when objects [should be] treated as personal’ but fails to develop such a theory (Duxbury, 1995, p. 666). 149. Dripps (1992, p. 1786, note 27) and Duxbury (1995, p. 670) note that these theories are inadequate. 150. By use of the word ‘woman’ in lieu of ‘person’ or some comparable genderneutral term, West (1993, 1986) seems to imply that only women can be prompted to ‘consent’ to sex which results in personal disutility. The theory underlying this proposition seems to be that such goods as ‘money, power, prestige [and] financial security’ are distributed unequally so that men have greater access to them than women (Chamallas, 1988, p. 839), and that men desire sex and women power, and thus, that immoral sex-for-power trades are induced. Thus, West (1993, 1986) among others argues that at least some women are forced to consent to sex to obtain power. A prominent problem with this argument is that it demeans women and assumes a tremendous degree of immorality on the part of both genders. It demeans women because it presumes that many are incapable of obtaining power or money by any other means than trades of their sexual services. It assumes the immorality of men by presuming that they would allow, if not force, women to trade such services for power; and yes, it assumes a similar immorality on the part of women by presuming that they would consent to such trades. On the other hand, Dripps (1992, p. 1791) argues that ‘[b]ecause there is no heterosexual reason to suppose that the ex ante distribution of sexual assets is any less arbitrary than the ex ante distribution of other assets’ people should be able to ‘trade their (arbitrarily distributed) attractiveness to improve their holdings of other arbitrarily distributed) goods’. One can’t help but wonder how a reason can be heterosexual or homosexual. 151. If heterosexual practices are morally problematic because they are, in essence, immoral exchanges in which women trade sexual services for money, power, and prestige, then homosexual practices are undoubtedly plagued by the same moral difficulties. It is generally accepted that older people of both genders tend to have more money, power, and prestige than their younger counterparts (Ehrenburg & Smith, 1994, pp. 295–299). Thus, assuming, as does West (1993, 1986), that sex-for-power trades are

Of Distributive Justice and Economic Efficiency

199

induced by unequal distributions of wealth, it seems likely that younger homosexuals will trade sexual services for the money, power, and prestige of older homosexuals, making homosexual practices as morally problematic as their heterosexual counterparts. 152. Landes & Posner (1987, p. 9) conclude that, although critics ‘have argued that a system of law designed to promote efficiency is immoral’, such criticisms are ‘largely misplaced when directed at a positive theory [such as that at issue here]’ for such a theory is ‘interested in explaining, rather than defending, the common law’. 153. ‘[C]urrent scholars generally agree that [sexual conduct contrary to commonly accepted moral or religious standards] should not be punished criminally’. (Malamut, 1993, pp. 45, 47) (citing Model Penal Code 213.2, Comment 2, and 213.6 note on adultery and fornication (1980). Kadish & Schulhofer (1995, pp. 155–67) state that ‘Perhaps the best illustration of the trend in various parts of the U.S. toward removing criminal sanctions from ‘immoral’ conduct is the decriminalization of fornication and cohabitation’ (1985, pp. 111). For the proposition that fornication remains immoral to many people see, for example, Exodus 22:16 and Leviticus 19:20 (King James Bible). However, Hill (1985, p. 123) notes that the fact that many people find fornication immoral does not, by itself, justify punishing those who do it. 154. For example, Mansfield (1994, p. 14) notes that ‘[t]he purpose of a model is to make predictions concerning phenomena in the real world, and in many respects the most important test of a model is how well it predicts these phenomena’. 155. It is true that ‘[l]aw and social morality will constrain much of the same behavior’ but ‘[t]his does not mean . . . that the law will enforce every aspect of morality that concerns preventing harm to others’. In fact, ‘[m]any immoral acts that hurt others are unregulated by the law’ Greenawalt (1995, pp. 710, 711). 240. Indeed, ‘there must remain a realm of private morality and immorality which is, in brief and crude terms, not the law’s business. To say this is not to condone or encourage private immorality’ (Hill, 1985, p.123, quoting Home Office, Scottish Home Department, ‘Report of the Committee on Homosexual Offenses and Prostitution’ (Wolfenden Report)). 156. Duxbury (1995, p. 6789, citing Buchanan, 1985, pp. 101–103) raises the possibility that ‘a market valuation is inept when it degrades that which is being valued’. West (1993, p. 2424) argues that ‘regarding the commodification . . . of sexual and reproductive services as morally problematic sounds not just stylistically odd, but morally deaf’. However, Wald (1997, pp. 459) argues that properly ‘understanding the relationship between property and rights of sexual control would contribute to a fuller understanding of why rape is so prevalent and why it is so injurious to women’. 157. Critics of Posner , who misunderstand the implications of using economics to study sexuality , include West (1993) and Radin (1987). 158. Far from being rhetoric which justifies a tolerant attitude towards rape, any reasonably conducted economic study of rape concludes that the harm to society as a whole and the victim greatly exceeds the gain experienced by the rapist, as we show below. 159. It is, of course, entirely possible that a community’s view of sex degrades one of the participants. The solution, however, is to attempt to change the norm, not to pretend that it doesn’t exist. 160. See also West (1993, p. 2430).

200

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

161. A transfer of wealth may be deemed coercive or voluntary. A coercive transfer is one that occurs involuntarily outside of the market and a voluntary transfer occurs voluntarily within the market. 162. While rape may often be a pure coercive transfer, it may, however, assuming a market for sexual services, also be an impure coercive transfer. This raises the dilemma of efficient rape. West (1993, p. 2430) discusses Posner’s analysis. Note, however, that it is not invariably true that efficiency demands that coercive transfers be criminally forbidden. 163. Most societies have punished rape (Dripps, 1992, p. 1780). Within American society, ‘rape is now a felony in every state and is universally regarded as a tort as well’ (Schwartz, 1979, p. 805). See also Dressler §213.1 (1987), Williams (1983), and Model Penal Code § 213.1 (19XX), Restatement of the Law, Second, of Torts .2d. 164. Posner (1992, Economic Analysis, p. 218) defines an efficient rape as a rape in which ‘the satisfactions of the rapist . . . exceed the victim’s pain and distress’. Thus, such rapes are arguably efficient and, as Posner (1992, Economic Analysis, p. 218) notes, it can be argued that the rapist ‘ought to be permitted to rape, provided only that he derives more pleasure from the act, over and above all substitutes . . . than the pain suffered by his victim’. Thus, in terms of the framework developed above, efficient rape may be considered an impure coercive transfer with a positive net change in social welfare. Hence, even Posner (1985, p. 1198) concludes that rape is a ‘seemingly more problematic example of the concept of crime as pure coercive transfer’. There may be other difficulties as well. For example, Schwartz (1979, pp. 806) argues that the theory is deficient for two reasons. First, it seems to hold that rape law is designed to protect prostitution. Since prostitution is, of course, illegal in itself, the traditional theory is ‘descriptively defective’ (Schwartz, 1979, p. 806), a view shared by Coleman (1988, pp. 153, 162). Second, Schwartz (1979, p. 806) argues that even presuming the legality of prostitution, prostitution could not provide an adequate substitute for rape for often the pleasure of the rapist depends on his forcible subjugation of the victim. Thus, Schwartz (1979, p. 806) seems to argue that rape laws are economically inexplicable because there is no acceptable market transaction they could be designed to promote. Conversely, Landes & Posner (1987, p. 158, citing Posner, 1985, pp. 1198–1199) argue that ‘there are market substitutes for coercive sexual acts - substitutes presumptively more efficient than using the legal system to direct the allocation of resources to sex’. 165. Posner’s position on the subject of efficient rape, however, seems to have fluctuated in its tenacity. At one point he calls ‘the prohibition against rape . . . not only consistent with but entailed by a normative economic analysis of sex’ (Posner, 1992, Sex, p. 182). At another, Posner (1992, Economic Analysis, p. 218) says only that ‘there are practical objections’to ‘a rape license’,but does not compare the strength of these objections with the theoretical gains that might flow from legalizing efficient rape. For other arguments supporting the inefficiency of rape and/or the efficiency of rape law, see Zelder (1993, pp. 1601–02), Ellis (1983), and Fried (1977, pp. 92–93). 166. However, Posner (1992, Sex, p. 386) stresses that this argument is utilitarian, not economic. See also Posner (1985, p. 1197). 167. It is undoubtedly true that fear of future rape is both prevalent and costly, for ‘[m]ost women experience fear of rape as a nagging, gnawing sense that something awful could happen, an angst that keeps them from doing things they want or need to do, or from doing them at the time or in the way they might otherwise do’ (Gordon & Riger (1991, p. 2). However, just as Nietzsche’s murderer ‘thirsted after the bliss of the

Of Distributive Justice and Economic Efficiency

201

knife’ a rapist may experience equally intense pleasure in anticipating an assault (Nietzsche, 1968, pp. 150–151). 168. For example, one could conduct a survey of rape victims, asking them how much they would have demanded before ‘consenting’ to rape (i.e. by not pressing charges) and the second group how much they would pay. It would be profoundly insensitive to ask a rape victim such a question, but it seems highly likely that the number a victim would accept would be high, approaching infinity. WTPr could be roughly estimated by considering the value rapists attach to their freedom (which they sacrifice by comitting a crime) multiplied by their estimate of the probability of arrest and conviction. Given low conviction rates for rape, WTPr is likely low, at least compared with WTAv. 169. Note that while there is a wealth of anecdotal support for this conclusion, no empirical evidence can be found. Such evidence would, of course, be necessary for a proper KHZ analysis. Nevertheless, the lack of any empirical studies probably indicates that rape is so obviously unacceptable that no such study is thought necessary. Indeed, the mere suggestion that such a study into the acceptability of rape be conducted is undoubtedly objectionable to many. 170. Schwartz (1979, p. 808) states that ‘our real judgment is that the act of rape is plainly wrong in any moral sense’; and he notes (1979, p. 807) that ‘the obvious truth is that rape is morally offensive, and deeply so’. West (1993, p. 2424) notes that ‘rape evidences . . . a malignant impulse toward women’ and she (1993, p. 2431, note 76) notes that ‘[a]s common intuition holds, rape is a violation of personhood in the deepest sense imaginable’. She also says (1986, p. 1449) that ‘ I would guess that the average reader, unlike Posner, views rape and prostitution as manifestations of what is wrong with sexual relations in this society and would not view the legal availability of prostitution (along with other sex markets such as dating and marriage) as a reason for the wrongness of rape’. Posner (1992, Economic Analysis, p. 386) himself recognizes that the prohibition against rape is founded in ‘unshakeable moral institutions’. See also Coker v. Georgia, 433 U. S. 584 (1977). 171. Both Louisiana and Georgia have recently passed ‘capital-rape laws’ (Higgins, 1997, p. 30). In upholding the former, the Louisiana Supreme Court noted that similar laws passed [though subsequently invalidated] in Florida and Mississippi may indicate a trend in public opinion ‘favoring such penalties’ (Higgins, 1997, p. 30, citing State v. Wilson, 685 So. 2d 1063 (La., 1996)). Note that such laws may be construed as a reasonable judgment that society is willing to pay more to deny standing to rapists than to grant it. See our discussion in part IV. If so, then any disutility on the part of the victim should be sufficient to find rape KHZ inefficient and, thus, properly illegal. 172. Posner (1992, Sex, pp. 386–387) himself says that rape is ‘not really utility maximizing’, but insists that this is a ‘utilitarian’ rather than ‘economic’ argument. Under KHZ, a community’s sense of outrage and grief at rape is both ‘economic’ and ‘utilitarian’, since the community’s regard for others is an economic concern (Zerbe, 2001). 173. For a discussion of the strength of opinions surrounding the issue of rape, see, for example, Kadish and Schulhofer (1995, p. 315). 174. More precisely, men during the Middle Ages regarded a women’s sexuality as the property of her husband or father. It is hard, if not impossible, to say whether women ever believed this, even during the Middle Ages, since the writers of the age were almost all men.

202

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

175. The lingering, partial exemption of husbands from rape liability may be explained by lingering sexist attitudes, or it may be an example of the written law changing more slowly than social attitudes. 176. Assuming that women did consent to such a grisly exchange, such an exchange can only be explained by a grotesque difference in bargaining power. This demonstrates that ex ante consent does not necessarily legitimate a legal rule. 177. Posner’s (1992, Economic Analysis, p. 218) suggestion that marital rape was legal because the ‘felt impropriety’ was ‘dilute[d]’ is interesting, because it implicitly recognizes the role of the regard for others in determining legal rules. However, Posner likely regards the ‘dilute[d]’ sense of impropriety towards marital rape as a noneconomic factor in determining the law. 178. Posner’s definition of wealth maximization has fluctuated. Although he has said that wealth maximazation is equivalent to Kaldor-Hicks efficiency (Posner, 1987, p. 16; 1986, pp. 12–13; 1992, Economic Analysis, pp. 88–89), he has also said wealth maximization is equivalent to GNP (Posner, 1987, pp. 19–20). At one point he implies that wealth maximization is equivalent with utility (1997, p. 18), but at other points he has said that they are different (1981, p. 60). 179. See also Russ v. Woodside Homes, Inc., 905 P.2d 901, 905 (Utah Ct. App. 1995) (stating that exculpatory clauses ‘relieve one party from risk of loss or injury in a particular transaction or occurrence and deprive the other of the right to recover damages from loss or injury’.). Eisenberg (1993) defines an exculpatory clause as a contract clause which purports to release a party from liability for injury caused by that party’s actions. 180. See Restatement of the Law, Second – Contracts .2d. See also Batsakis v. Demotsis, 226 S.W.2d 673 (Tex. Civ. App. 1949), which recounts that ‘[m]ere inadequacy of consideration will not void a contract’. Farnsworth (1982) concludes that parties are free to make agreements as they wish, and courts will usually enforce any bargained-for exchange without passing on their substance. Note, however, that ‘the broad prohibition against review of ‘adequacy’ has [not] meant that fairness of an exchange is not a legal concern’ (Dawson, Harvey & Henderson, 1993). However, Farnsworth (1982, pp. 40, 66–67) insists that courts’ sole inquiry is whether the parties went through a bargaining process, and that unfair terms are only relevant as evidence of coercion, fraud, sarcasm, or that the ‘exchange’ was really a gift. 181. See also Farnsworth (1982, p. 326) and Schell (1993, p. 441), who states that ‘[d]espite the general need to enforce contracts as written, the law recognizes a number of occasions when judicial review of agreements is warranted. Defenses based on public policy are among the most ancient of these exceptions’. 182. See, for example, Farnsworth (1990, pp. 353–56, 1982, p. 333) and Schell (1993, p. 463). Shell (1993, p. 463) states that ‘contractual risk-shifting may reduce the incentives of parties performing important services to the public at large to take reasonable care in the performance of their contractual duties’. See also Bisso v. Inland Waterways Corp., 349 U. S. 85 (1955), which notes that holding exculpatory clauses unenforceable ‘discourage[s] negligence’. An alternative ground of unenforceability is unconscionability (Dawson, Harvey & Henderson, 1993, pp. 700–726). 183. Exculpatory clauses may purport to excuse [1] liability for intentional torts, [2] strict liability, and [3] liability for negligence. Exculpatory clauses which purport to excuse liability for intentional torts are clearly unenforceable (Farnsworth, 1990, p. 353,1982, p. 333). See Alack v. Vic Tanny Intl. of Missouri, Inc., 923 S.W.2d 330 (Mo.,

Of Distributive Justice and Economic Efficiency

203

1996) (stating that ‘there is no question that one may never exonerate oneself from future liability for intentional torts or for gross negligence’). With respect to clauses purportedly excusing strict liability of a seller of a product, the majority rule is that ‘a seller of a product can[not] exempt [her- or] himself from the strict liability imposed for physical harm caused by its unreasonably dangerous condition’ (Farnsworth, 1982, p. 334) See also Restatement of the Law, Second – Contracts .2d §195(3) and Talbott (1988, p. 30), who notes that although ‘courts have permitted a contractual waiver of a seller’s strict liability for product defects only in cases not involving personal injury’ that nonetheless ‘an exception may be made in cases of a fairly negotiated contract between two merchants for the sale of an experimental product’ (Restatement of the Law, Second - Contracts .2d §195, Comment c. Talbott (1988, p.30) states that courts have ‘uniformly limited such waivers to agreements between commercial entities for commercial losses’. 184. See also Tunkl v. Regents of University of California, 60 Cal. 2d 92 (1963). Farnsworth (1982, pp. 333–34) states that ‘courts have held such an agreement unenforceable because the agreement affects the public interest and the other party is a member of a protected class’. See also Farnsworth (1990, pp. 353–354). 185. However, Farnsworth (1990, p. 354, 1982, p. 334) disagrees. 186. See Garretson v. United 456 F.2d 1017 (9th Cir. 1972). 187. For some relevant work, see, for example, Talbott (1988), Schell (1993) and Posner (1992, Economic Analysis, p. 113–115). 188. Cooter & Ulen (1997, p. 167) state that ‘[i]n general, economic efficiency requires enforcing a promise if the promisor and promisee both wanted enforceability when it was made’. 189. See Tunkl, 60 Cal. 2d. 92 (1963) (stating that because a ‘hospital-patient contract falls within the category of agreements affecting the public interest’ a standardized release from liability for negligence imposed as a condition for admission to a charitable research hospital having superior bargaining power is unenforceable). Talbott (1988, p. 31) notes that ‘the paternalistic defense argues that the person granting the waiver is for all intents and purposes at the mercy of the corporate or institutional party . . . and any such agreements would almost inevitably rebound to the gain of th[at party] and to the individual’s and ultimately the public’s loss’. Talbott (1988, p. 28) also notes that ‘[i]n the development of these now well-established principles of law, the quasi-coercive nature of the bargaining – in order to buy the car or gain admission to the hospital, you must sign the waiver – has played a decisive role’. 190. Although Posner is here discussing the efficiency of standard contracts, the analysis is similar. Consider also Talbott (1988, p. 31, 47 note 16). 191. The Tunkl opinion makes this argument. See Tunkl, 60 Cal. 2d. 92. 306. The exculpatory clause itself could be valued as the probability of an injury to the promisor due to the promisee’s negligence times the probability of a successful suit on that injury times the present value of the expected judgment. 192. Talbott (1988, pp. 25, 30) states that ‘post-Tunkl cases generally consider [inter alia] whether there existed disparity of bargaining power’. See Tunkl, 60 Cal. 2d 92. 193. That is, in cases involving exculpatory clauses executed by an individual promisor in favor of a monopolist promisee. 194. With respect to [1] competition between common carriers consider ‘fare wars’ between airlines, [2] competition between hospitals and doctors consider recent

204

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

developments in the health care industry, [3] competition among ‘public warehousemen’ consider the abundance of private mini-storage companies [cite]; and [4] competition among public utilities, see The Telecommunications Act of 1996, which provides for competition among telecommunications providers ranging from traditional cable companies to municipal electric utilities. 195. Posner (1992, Economic Analysis, p. 114) notes that ‘all the firms in the industry may find it economical to use standard contracts and refuse to negotiate with purchasers. But what is important is not whether there is haggling in every transaction but whether competition forces sellers to incorporate in their standard contracts terms that protect the purchasers’. A similar statement could be made with respect to exculpatory clauses. Here, as elsewhere in this section, the term ‘price’ is defined broadly to include not only the monetary price of the good or service, but also the value of the exculpatory clause. 196. Talbott (1988, p.26) states that ‘[w]here the population of potential accident victims is very large and the probability of injury is very small, it may well be rational for each potential victim to sell her right to compensation at a price that would not begin to compensate [him or] her if [he or] she were unlucky enough to be injured. 197. Talbott (1988, p. 40) concludes that ‘none of the well-known normative political theories explains this result’, including ‘Posner’s theory of Wealth Maximization’. 198. Note our discussion at (II)(C)(1). Consider Zerbe (2001, Chapter 2, 3). 199. Given the regard for others upon which KHZ efficiency is premised, the allocative change will quite possibly be integratively efficient as well. 200. The relevant markets were found to be competitive in the previous section. See part (III)(C)(2)(c). 201. Note, however, that strictly construed, such ‘benefits’ are actually compensation for previous losses. See, for example, Wade (1994) and Cooter & Ulen (1997, pp. 306–311). 202. See also Aristotle (Ethics) and our discussion in part (II). Again, in the absence of proper empirical evidence, anecdotal references and common sense must be the twin polestars of decision. 203. Landes & Posner (1987, pp. 27–28, 315–136) note that the movement toward contribution among negligent tortfeasors is one of the areas of ‘greatest divergence between efficient and actual rules of tort law’. Posner (1992, Economic Analysis, pp. 188–189) is among those who states that no-contribution rules are more efficient than contribution. Keeton et al. (1984) are among those who note the movement towards contribution. 204. However, Landes & Posner (1987, p. 24) argue that the only way to invalidate an existing theory is by presenting evidence that an alternative theory better explains the subject phenomenon than does the existing theory. See also our discussion in part (III)(B). 205. Landes & Posner (1987, p. 91) state that ‘Cases in which the victim’s injury is caused by more than one injurer are of two different types: [1] the simultaneous joint tort and the successive joint tort’. In a simultaneous joint tort, the victim ‘suffers a single or individible injury as a result of the tortious activity of two or more persons’; and in a successive joint tort ‘one tortfeasor aggravates an injury inflicted by the other’ (Landes & Posner, 1987, p. 91). Only simultaneous joint torts are at issue here.

Of Distributive Justice and Economic Efficiency

205

206. The common law recognized two circumstances in which joint and several liability would hold: (1) if the defendants acted together to cause the victim’s harm, or (2) if the fault for the victim’s harm was considered indivisible between multiple tortfeasors (Cooter & Ulen, 1997). And see for example, Restatement of the Law, Second, of Torts .2d §§875, 881. The term ‘‘[j]oint tortfeasor’ has been defined as two or more persons jointly and severally liable in tort for the same injury to persons or property, whether or not judgment has been recovered against them’ (Schwartz, Simpson & Arnold, 1979, pp. 779, 781, note 9). Black’s Law Dictionary (1990) states that the term ‘joint tortfeasors’ ‘refers to two or more persons jointly and severally liable in tort for the same injury to person or property’. 207. The efficiency of joint and several liability, itself, will not be considered here but for insightful economic analyses in support of such liability consider Cooter & Ulen (1997, p. 302) who conclude that ‘[t]here are several economic reasons for joint and several liability’) and Kornhauser & Revesz (1989, pp. 832–833) who set forth an analysis of ‘the properties of several classes of apportionment rules not predicated on a finding of joint and several liability’ which attempts to ‘provide [some] normative guidance as to when such liability is desirable from an efficiency prospective’. However, Menzer (1988, pp. 628, 648–49) argues that joint and several liability induces overdeterrence. However, Kornhauser & Revesz do conclude that ‘[a]lthough rules of joint and several liability are efficient under negligence, they will not produce efficient results under strict liability’ (supra note 331 at 856–58). 208. For an early treatment of the effect of contribution on pre-trial settlement, see James (1941). 209. ‘Indemnity’, according to Keaton et al. (1984), is ‘an order requiring another to reimburse in full one who has discharged a common liability’. The Restatement of the Law of Restitution §77 (1937) states that ‘a person who has discharged a duty that is owed by him but which as between himself and another, should have been discharged by the other is entitled to indemnification. ’ See also Restatement of the Law, Second, of Torts .2d §886B. ‘Although it was occasionally said that contribution ‘is but pro tanto indemnity’ . . . contribution and indemnity were traditionally distinguished in [three] respects’: (1) indemnity is based upon contract, contribution upon principles of equity, (2) indemnity shifts the entire loss to another party, and (3) only in cases of contribution are both parties jointly liable (Higgenbotham & Wiggins, 1990, p. 698). But see American Motorcycle Ass’n v. Superior Court 578 P.2d 899, 907 (Cal., 1978) (stating that ‘the dichotomy between the two concepts [of indemnity and contribution] is more formalistic than substantive, and [that] the common goal of both doctrines, the equitable distribution of loss among multiple tortfeasors, suggests a need for re-examination of the relationship between these twin concepts’.)(superseded by statute as stated in Miller v. Stouffer, 11 Cal Rptr. 2d 454 (1992)). Indeed, American Motorcycle’s recognition of the need for reexamination proved prophetic, in that California subsequently repealed joint and several liability through Proposition 51. See Miller v. Stouffer, 11 Cal. Rptr. 2d 454 (1992). 210. Di Cola (1992, pp. 1543, 1544, note 5, quoting 18 Am. Jur. 2d Contribution §1 (1985) defines ‘contribution’ as ‘a payment made by each person, or by any of several persons, having a common interest or liability, of his share in the loss suffered or in the money necessarily paid by one of the parties in behalf of all the others’. Gregory (1938) adopts a similar definition. See Vickers Petrolium Co. v. Biffle, 239 F.2d 602, 606 (10th Cir., 1956). See Roberts v. Robert V. Rohrman, Inc., 909 F. Supp. 545, 553 (N. D.Ill.,

206

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

1995) (defining ‘contribution’ as the ‘principle whereby a tort-feasor against whom judgment is rendered is entitled to recover proportional shares of the judgment from other joint tort-feasors whose negligence contributed to the injury and who are also liable to the plaintiff’); Avnet, Inc. v. Allied Signal, Inc., 825 F. Supp. 1132, 1138 (D.R. I. 1992) (stating that ‘[a] claim for contribution is a claim in which one liable party seeks to recover from another liable party for the latter party’s share of common liability’); U. S. v. Conservation Chemical Co., 619 F. Supp. 162, 224 (W. D.Mo., 1985)(claim dismissed by U. S. v. Conservation Chemical Co. 1988 U. S. Dist. Lexis 18283 (W. D.Mo. 1988)). See also Blomgren v. Marshall Management Services, Inc., 483 N.W.2d 504, 506 (Minn. Ct. App.1992) (defining ‘contribution’ as a ‘remedy for one who has discharged more than his fair share of common liability for a burden, allowing one to recover a proportionate share from the other liable party’); Somer v. Federal Signal Corp., 593 N.E.2d 1365, 1372 (N. Y. 1992) (noting that ‘[c]ontribution enables a joint tort-feasor that has paid more than its equitable share of damages to recover the excess from other tort-feasors’). 211. See Palmer v. Wick and Pultneytown Steam Shipping Co., A. C. 318 (1894). 212. Keeton et al. (1984, p. 337) argue that ‘the better English view, even before their statute, appears clearly to have been that contribution is not denied in cases of mere vicarious liability, negligence, accident, mistake, or other unintentional breaches of the law’. Consider part (III)(B)(3)(d), which finds a rule of contribution among negligent joint tortfeasors to be KHZ efficient. 213. Keeton et al. (1984, pp. 336–337) argue that ‘the origin of the rule [allowing contribution among negligent joint tortfeasors] and the reason for it were [ultimately] lost to sight’. 214. Higgenbotham & Wiggens (1990, pp. 700–701) note that ‘[t]he English case of Merrywether v. Nixan.. . . is generally cited as the beginning or early restatement of the common law rule [that] there could be no contribution among joint tortfeasors, whether negligent or intentional’. See Merrywether v. Nixan, 101 Eng. Rep. 1337 (K. B. 1898) See Merrywhether, 101 Eng. Rep. 1337. 215. ‘Although the record is sparse, it is believed that the injuries [in Merrywether] were intentionally inflicted’ (Schwartz, Simpson & Arnold, 1979, p. 782). Indeed, the case ‘involv[ed] the intentional tort of trover’ (Schwartz, Simpson & Arnold , p. 782). See Merrywether, 101 Eng. Rep. 1337 (stating that ‘[o]ne Starkey brought an action on the case against [the] plaintiff and defendant for an injury done by them to his reversionary estate in a mill, in which was included a count of trover, for the machinery belonging to the mill’). This was also Keeton’s (1984, pp. 336–337) conclusion. 216. This holding was apparently founded upon the maxim ‘ex turpi causa non oritur actio’, which Leflar (1932) defines as ‘[o]ut of a base [illegal, or immoral] consideration, an action does . . . not arise’. Black’s Law Dictionary 589 (6th ed. 1990). Thus, the ‘traditional justification for the common law rule was that a tortfeasor should not be able to use the judicial system to recover for his own wrongs’(Comment 1968, P. 730). Consider also Schwartz, Simpson & Arnold (1979, p. 782, note 19). 217. For some examples, see Higgenbotham & Wiggins (1990, p. 701), Schwartz, Simpson & Arnold (1979, p. 782) and Stanley (1994, p. 4). Indeed, ‘Lord Kenyon (in Merrywether) explicitly states that the decision ‘would not affect cases of indemnity where one man employed another to do acts, not unlawful in themselves’ (Keeton et al. 1984, pp. 336–337). See Merrywether, 101 Eng. Rep. 1337.

Of Distributive Justice and Economic Efficiency

207

218. Keeton et al. (1984, p. 337) note that ‘[l]ater [English] cases seized upon th[e] limitation [of Merrywether to intentional torts], and held that the rule against contribution did not apply unless the defendant was a willful and conscious wrongdoer’. Higgenbotham & Wiggins (1990, p. 701) note that ‘English cases after Merrywether, and early American cases, noted the distinction between negligent and intentional torts, and allowed contribution in the former’. See also Schwartz, Simpson & Arnold (1979, p. 729). Merrywether, 101 Eng. Rep. 1337. 219. Schwartz, Simpson & Arnold (1979, p. 782) note that ‘[t]he earlier American cases also seemed to draw the distinction between intentional and negligent torts, allowing contribution in negligence actions only’. 220. In commenting on the phenomenon, Reath (1898, p. 177) notes that ‘[i]t is singularly unfortunate, and has led to misunderstanding, that Merrywether v. Nixan should have been continually treated as stating the ‘general rule.’ As a matter of fact that case states not the general rule, but the exception’. See Merrywether, 101 Eng. Rep. 1337. 221. Federal common law, that is, the law applied in diversity cases prior to Erie, also denied contribution in tort cases, absent some express statutory provision in favor of contribution or some indication of Congressional intent to allow such a right (Higgenbotham & Wiggins, 1990, pp. 725–726; Easterbrook, Landes & Posner, 1980, pp. 332–333). 222. 356. Jurisdictions allowing contribution were the District of Columbia, see Knell v. Feltman, 174 F.2d 662 (D. C. Cir. 1949); Nevada, see Weiner v. United Air Lines, 216 F. Supp. 701 (S. D.Cal. 1962) (applying Nevada law); Iowa, see Best v. Yerkes, 77 N.W.2d 23 (Iowa, 1956); Louisiana, see Quatray v. Wicker, 178 La. 289 (1933)(superseded on other grounds by statute as recognized in Cole v. Celotex Corp., 599 So. 2d 1058, 1070 (La., 1992); Maine, see Bedell v. Reagan, 192 A.2d 24 (Me., 1963); Minnesota, see Grothe v. Shaffer, 305 Minn. 17 (1975); Pennsylvania, see Goldman v. Mitchell-Fletcher Co., 292 Pa. 354 (1928); Tennessee, see Davis v. Broad St. Garage, 191 Tenn. 320 (1950); and Wisconsin, see Ellis v. Chicago and N. W. Ry., 167 Wis. 392 (1918). 223. The unpopularity of the no contribution rule is noted by Landes & Posner (1987, p. 192), Keeton et al. (1984, p. 338), and Schwartz, Simpson & Arnold (1979, pp. 783–784). 224. See also Keeton et al. 1984, p. 338; Higgenbotham & Wiggins, 1990, p. 702. 225. Higgenbotham & Wiggins (1990, pp. 701–702) note that ‘[t]o evade the inequitable results occasioned by applying the no-contribution rule, some courts fashioned exceptions’, while other courts either ‘applied a rule that distinguished ‘active’ from ‘passive’ negligence’ or ‘recognized a right to contribution in favor of a tortfeasor to whom another tortfeasor had violated a duty’. 226. Kornhauser and Revesez (1989, p. 841, note 48) note that ‘[b]eginning in the nineteenth century . . . common law courts began to recognize.. . . a right [to contribution]’. 227. This legislative trend is discussed by Higgenbotham & Wiggins (1990, p. 702), Fleming (1983, p. 233), Cavanagh (1987, p.1285–86) and Cooter & Ulen (1997, pp. 301–302). Apparently, the ‘movement toward comparative fault has accelerated the movement toward contribution, since comparative fault statutes commonly provide also for contribution’ (Keeton et al.1984, p. 338, note 17).

208

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

228. The Law Reform (Married Women Tortfeasors) Act §6 (1935), provides that ‘where damage is suffered by any person as a result of a tort . . . any tortfeasor liable in respect of that damage, may recover contribution from any other tortfeasor who is, or would if sued have been, liable in respect of the same damage, whether as joint tortfeasor or otherwise; so, however, that no person shall be entitled to be indemnified by him in respect of the liability of which contribution is sought’. 229. ‘This trend, however, [has] not change[d] the rule that always existed that contribution could not be enforced between willful or intentional tortfeasors’ (Schwartz, Simpson & Arnold, 1979, p. 784). See also Restatement of the Law, Second, of Torts .2d §886A(3), comment j, which provides that ‘[t]here is no right to contribution in favor of any tortfeasor who has intentionally caused the harm’, the Uniform Contribution Among Tortfeasors Act, §1(c) U.L. A. (1955), and Higgenbotham & Wiggins (1990, p. 702). 230. The graph itself is based upon a compilation of two sets of data collected by Eggen (1995, pp. 1701, 1701–02), and Landes & Posner, (1987). 231. The Uniform Contribution Among Tortfeasors Act §1 U.L. A. (1955), provides that ‘where two or more persons become jointly and severally liable in tort for the same injury to person or property or the same wrongful death, there is a right to contribution among them even though judgment has not been recovered against all or any of them’. By 1979, 15 jurisdictions had adopted some version of this uniform act (Schwartz, Simpson & Arnold, 1979, p. 786). By 1990, that number had risen to at least 20 (Higgenbotham & Wiggins, 1990, p. 702). 232. See Uniform Comparative Fault Act, §4, 12 U.L. A. 47 (1979) 233. Restatement of the Law, Second, of Torts .2d §886A(1) holds that, except as otherwise stated, ‘when two or more persons become liable in tort to the same person for the same harm, there is a right of contribution among them, even though judgment has not been recovered against all or any of them’. 234. Restatement of Restitution §81 (1937) provides that ‘[a] person who has discharged more than his proportionate share of a joint duty is entitled to contribution from the other or others except where the payor is barred by the wrongful nature of his conduct’. 235. Eggen (1995, pp. 1701–1702), notes that ‘[n]ot only are the uniform laws inconsistent with one another, but inconsistent applications of identical provisions exist among who states adopting the same uniform act’. 236. ‘The requirement that the [contribution] claimant be ‘liable’ in respect of the plaintiff’s damages may be satisfied not only when he has been actually held liable but also when he has settled’ (Fleming, 1983, p. 236). 237. See also Higgenbotham & Wiggins (1990, p. 714). 238. Interestingly, some careless plaintiffs have signed a ‘release of one joint tortfeasor’ which ‘discharged the liability of all joint tortfeasors’. Therefore, ‘a settlement with one tortfeasor in which a release was given would prevent the plaintiff from going after other joint tortfeasors’ (Easterbrook, Landes & Posner, 1980, p. 333). 378. Conversely, if a settling defendant pays more than that defendant’s share of the liability, that defendant may not seek contribution from the non-settling defendant. See the Uniform Contribution Among Tortfeasors Act §1(d), U.L. A. (1955). 239. Higgenbotham & Wiggins, 1990, p. 714), But see Comment (1968, pp. 730–731) which concludes that ‘the pitfalls [inherent in a rule permitting contribution] can be easily avoided’. Di Cola (1992, pp. 1555–1556) states that ‘[l]aw and economics’

Of Distributive Justice and Economic Efficiency

209

scholars have acknowledged . . . that contribution rules can be structured in a way that avoids the negative impact on settlement’. However, Easterbrook, Landes & Posner (1980, pp. 363–64, 365–366) argue that such rules ‘would probably be very costly to administer’. 240. The Restatement notes that a rule prohibiting contribution from the settling defendant would be both ‘very unfair to the other tortfeasors’ and ‘provide[ ] a clear incentive to collusion between the settling parties’ (Restatement of the Law, Second, of Torts .2d, comment m). Keeton et al. (1984, p. 338, note 13) argue that Penn. Co.v. West Penn. Ry., 144 N. E. 51 (Ohio, 1924) is an example of such collusion. 241. The Restatement of the Law, Second, of Torts .2d , comment m, notes that ‘[t]here are three possible solutions for the situation’ but ‘[i]mportant policy reasons [which] weigh against each of the[m]’. 242. The current (1977) version of the Uniform Act provides that ‘[w]hen a release or a covenant not to sue or not to enforce judgment is given in good faith to one of two or more persons liable in tort for the same injury or the same wrongful death . . .[i]t discharges the tortfeasor to whom it is given from all liability for contribution to any other tortfeasor’. (Uniform Contribution Among Tortfeasors Act, §1, U.Law A, 1955). Thus, the ‘Uniform Act favors settlements’ (Higgenbotham & Wiggins, 1990, p. 714). 243. Restatement of the Law, Second, of Torts .2d §886, Caveat and Comment m, states that ‘[t]he Institute takes no position on the effect of a release of one tortfeasor from liability for the harm or a covenant not to sue him for it upon the right of other tortfeasors to contribution from him’. 244. In the words of Kornhauser & Revesz (1989, p. 831), ‘[t]he apportionment of . . . losses among the various responsible parties raises difficult and controversial questions’. Indeed, Di Cola (1992, p. 1546) notes that the ‘literature discussing contribution [in such areas as] securities, antitrust, and environmental law is extensive’. In the securities area see, for example, Sachs (1994) and Adamski (1981). In the antitrust field, see, for example, Easterbrook, Landes & Posner (1980), and Polinsky & Shavell (1981). In the environmental law sphere, see, for example, Hernandez (1997). 245. However, the trend towards contribution is not universal, and indeed California has abandoned joint and several liability. See Miller v. Stouffer, 11 Cal. Rptr. 2d 454 (1992). 246. This paper concerns only the efficiency of contribution where liability is founded upon simple tort negligence. For an efficiency analysis of contribution where liability is strict, see Kornhauser & Revesz (1989, pp. 834, 858–61), who conclude that ‘while no-contribution rules will not lead to the efficient result under strict liability, rules of contribution, if properly designed, can do so’. And, for an efficiency analysis of where liability is based upon an intentional tort, see Landes & Posner (1987, p. 204), who note that ‘economic analysis of no contribution in the intentional case is symmetrical with that in the unintentional’. 247. According to most commentators, a rule of contribution among negligent joint tortfeasors is traditionally Kaldor-Hicks inefficient, when compared to a rule of nocontribution (Landes & Posner, 1987; Easterbrook, Landes & Posner, 1980; Polinsky & Shavell, 1981; Shavell, 1987, pp. 164–167; Cooter & Ulen, 1997; Stanley, 1994). Posner (1992, Economic Analysis, p. 1553) notes that the ‘primary objection to contribution is that it is economically inefficient’. The commentary is not, however, unanimous. Comment (1968, pp. 730, 735) argues that ‘[c]ontribution. . . increases deterrence [over a rule of no-contribution] by the certainty that each tortfeasor will incur

210

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

some liability . . . rather than by the existence of a chance that some tortfeasors will escape all compensation responsibility’. Landes & Posner (1987, pp. 192–193) counter that this ‘criticism rest[s] on a confusion between ex ante and ex post’ for ‘[i]f, ex ante, each defendant bears a [sufficient expected] cost . . . of liability, each defendant will be deterred, even if ex post all but one pay nothing’. See also Arrow (1973). 248. With respect to incentives for efficient accident avoidance see, for example, Kornhauser & Revesz (1989, p. 837) who state that ‘[f]rom an efficiency prospective, the objective of the liability and apportionment rules is to induce rational actors to produce the socially efficient amount of [harm]’ and who describes and analyzes ‘eight categories of apportionment rules under negligence and two . . . under strict liability’. 249. For a given allocative change to be Kaldor-Hicks efficient, the ‘gainers [from the change must] gain more than the losers lose’. (Katz, 1996, p. 2240) See, for example, Cooter & Ulen (1997, pp. 41–42). A movement from no contribution to contribution results in traditionally Kaldor-Hicks-irrelevant ‘ex post distributional consequences’ (Easterbrook, Landes & Posner, 1980, pp. 349–359). For example, a movement from no contribution to contribution could result in a transfer of wealth from defendants who could have sought contribution to those from whom such contribution would have been procured, and the movement could result in a loss of judicial economy. It, therefore, fails to satisfy traditional conceptions of Kaldor-Hicks efficiency. 250. For example, Dicola (1992, pp. 1556–1558) argues that, while ‘[t]he reliability of the efficiency model as an accurate predictor of result . . .depends on the validity of the assumptions embedded in the analysis . . .Several recent commentators . . .have undermined the validity of some of those assumptions’. 251. For example, see Easterbrook, Landes & Posner (1980, p. 350). Di Cola (1992, pp. 1557–1558) notes the ‘model’s assumption[ ] that all litigant’s have perfect information’. Moreover, in Stanley’s (1994, p. 6) view, both Easterbrook, Landes & Posner (1980) and Polinsky & Shavell (1981) ‘assume[ ] symmetric information and beliefs in reaching their conclusions and recommendations’. 252. Di Cola (1992, pp. 1557–1558, citing Yi, 1990, pp. 18–20) notes the assumption and states that ‘[w]hen some defendants are insolvent. . . the positive effects of a nocontribution rule on settlement will be less than predicted’. 253. For example, see Kornhauser & Revesz (1989, p. 832). Easterbrook, Landes & Posner (1980, p. 351) note that ‘[o]ur conclusion that no-contribution and contribution rules yield the same deterrence assumes that [the actors, here] firms[,] are risk neutral. Note that Easterbrook, Landes & Posner (1980, p. 351) define risk neutrality as ‘indifferen[ce] between paying (receiving a certain sum and paying (receiving) its uncertain actuarial equivalent’). Cooter & Ulen (1997, p. 47) define risk neutrality as ‘constant marginal utility of income’ by which one is ‘indifferent between a certain prospect of income and an uncertain prospect of equal expected monetary value’. 254. Kornhauser & Revesz (1989, pp. 847–850) explore the effects of assuming that all actors are irrational on the analysis of contribution. Cooter & Ulen (1997, pp. 296–297) discuss the ramifications of assuming actor rationality for a microeconomic analysis of law generally. 255. While this assumption is by no means universal, it was made by Landes & Posner (1987, p. 204) for a portion of their work. For example, Kornhauser & Revesz (1997, p. 832) note that ‘these commentators considered the choice between apportionment and contribution only under negligence’. However, Landes & Posner (1987, p. 204), analyze the effects of contribution and no-contribution rules for

Of Distributive Justice and Economic Efficiency

211

intentional torts as well. The assumption of negligence liability is also made throughout this paper. 256. Kornhauser & Revesz (1989, pp. 846–847) assume that ‘the standards of care are set at the socially optimal level’. 257. Kornhauser & Revesz (1989, p. 832–833) note that ‘because there can be no right to contribution without joint and several liability, [several] commentators [have] assumed, without much discussion that rules of joint and several liability would govern’. A similar assumption is made throughout this paper. Joint and several liability has been found to ‘induce[ ] efficient outcomes’ (Kornhauser & Revesz, 1989, pp. 847–851). 258. Although this assumption may be unrealistic (see, for example, Keeton et al. 1984, pp. 469–471), it is probably not vital to the analysis (Cooter & Ulen, 1997, pp. 259–292). 259. This is what Kornhauser & Revesz (1989, pp. 837) call a ‘full liability rule’. Kornhauser & Revesz (1989, pp. 837–39), describe the rule and note that although ‘[c]ommentators are split on whether courts apply a full liability of a partial liability rule’ by which a liable actor is ‘liable only for those losses that . . . are caused by the actor’s negligence . . . most economic models have assumed that [a full liability rule is in force]’. 260. The assumption is made here as well. 261. Easterbrook, Landes & Posner (1980, p. 353) note that they ‘assume[ ] that the antitrust system operates without error’ in that ‘the substantive principles are correctly assessed, where what is meant by correctly is given by economic analysis’. See also Landes & Posner (1987, pp. 197–198). 262. Cooter & Ulen (1997, p. 302) note that in an analysis of contribution versus no contribution, ‘the critical distinction is between situations in which the optimal precaution is unilateral by one of the defendants and those in which it is joint among all of them’. Landes & Posner (1987, p. 190) note that a ‘simultaneous joint tort. . . can involve either joint care or alternative care’. 263. Refer to Cooter & Ulen (1997) for a definition of ‘bilateral care’. Landes & Posner (1987, pp. 60, 190–91) note that ‘in a case of joint care optimal accident avoidance requires both parties (here, both injurers) to take care’. See also Posner (1992, Economic Analysis, pp. 170–171). 264. Landes & Posner (1987, pp. 60, 191, 198) provide a definition of ‘alternative care’. Cooter & Ulen (1997, p. 275) note that ‘unilateral precaution’ cases only require ‘one party to an accident to take precaution against it’. For more details, refer to Posner (1992, Economic Analysis, pp. 170–171). 265. Posner (1992, Economic Analysis, p. 188) notes that ‘[a]t common law. . . the rule in alternative care cases [is] indemnity’. 266. Landes & Posner (1987, pp. 198–201) conduct an efficiency analysis of the rule of indemnity in alternative care cases. 267. Again, it is here assumed that liability is founded upon negligence, (Easterbrook, Landes & Posner, 1980). Legally, negligence may be defined as ‘[c]onduct which falls below the standard established by law for the protection of others against unreasonable risk of harm’. (Black’s Law Dictionary (1990)). Kornhauser & Revesz (1989, p. 836) note that ‘[n]egligence rules define a standard of care’ and that ‘if a[n actor] meets her standard of care, she will bear no portion of the social loss’. Economically, that standard may be said to be equal to that level of precaution, x*,

212

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

which minimizes the expected social cost of an accident, SC, where that cost is defined as the cost of precaution, Cx, plus the cost of expected harm from an accident, p(x)L; that is, SC = Cx + p(x)L. The level of precaution which minimizes the expected social cost of an accident is that at which the marginal cost of an additional unit of precaution equals the marginal benefit of the resulting reduction in the cost of expected harm; thus, the efficient level of precaution is that value of x, x*, for which C = –p⬘(x*)L. Consequently, a potential injurer is negligent if it employs precaution x < x* (Cooter & Ulen, 1997, pp. 273–74; Posner, 1992, Economic Analysis, pp. 163–167). 268. Compare this statement with Kornhauser & Revesz’s (1989, p. 871) analysis. Kornhauser & Revesz (1989, p. 836) note that ‘[t]he efficient amount of [harm] is that which maximizes the social objective function: the sum of the benefits derived by the actors minus the social loss’. 269. For ease of exposition it is here assumed that there are only two potential injurers. This analysis is, however, adaptable to cases of more than two potential injurers (Landes & Posner, 1987). 270. It is, of course, also necessary that the victim not be negligent (Wade et al. 1994, pp. 566–581). 271. Kornhauser & Revesz (1989, p. 842) call this a ‘unitary share rule’ which, in their ’taxonomy’ ‘correspond[s] to rules of joint and several liability’. Kornhauser & Revesz (1989, p.841–842) explain that, under a unitary share rule ‘the negligent actors pay not only for the damages attributable to their own actions but also for the damages attributable to non-negligent actors’ and thus, that ‘the victim recovers the maximum loss allowable whenever at least one actor is negligent, regardless of how much of the damage is caused by non-negligent actors’. Kornhauser & Revesz (1989, p. 841) suggest a ‘fractional share rule’ under which ‘the negligent actors do not pay for the damages attributable to non-negligent actors’. Di Cola (1992, pp. 1556–58) notes that use of that rule could ‘shift’the ‘contribution/no-contribution efficiency calculus’. 272. Specifically, given that SC = Cx + p(x)L (as we say in note 417), the cost of compliance with the legal standard, Cx, is SC ⫺ p(x); and, thus, Cx < SC (Cooter & Ulen, 1997, pp. 270–272). 273. Kornhauser & Revesz (1989, pp. 847–848). Kornhauser & Revesz (1989, pp. 847–848) conclude that ‘[i]f all but one of the actors are non-negligent, it would not be rational for the remaining actor to be negligent’. 274. Again, to simplify the analysis, it is assumed that there are only two potential injurers. This analysis may easily be expanded to account for more than two potential injurers by adding additional terms. 275. For example, Easterbrook, Landes & Posner (1980, p. 345) note that an actor ‘will choose to comply with or violate the antitrust laws depending on whether its anticipated gain from the violation is greater or less than its expected liability’. Kornhauser & Revesz (1989, p. 836) state that ‘[a]n economically rational [actor] . . . seeks to maximize her private objective function: the benefit that she derives from the activity that leads to the production of the [harm] minus whatever share of the social loss the legal regime allocates to her’. 276. Note that it is assumed here, and will be assumed throughout this subsection, that the victim C has an incentive to behave non-negligently by employing precaution z*, and this is indeed the case under either rule of allocation, contribution or nocontribution. Landes & Posner (1987, pp. 194–195) prove that ‘[t]he equilibrium level of victim care is always. . .’

Of Distributive Justice and Economic Efficiency

213

277. ‘Under negligence. . . [f]or a no-contribution rule to be efficient. . . the sum of the probabilities that the actors attach to the risks of being held responsible for the. . . damage must be equal to one’ (Kornhauser & Revesz, 1989, p. 861). See also Landes & Posner (1987, p. 203). This, in turn, requires that each actor have perfect knowledge for without such knowledge, ‘different actors may attach inaccurate probabilities to their risk of liability and the probabilities may not sum to one’ (Di Cola, 1992, pp. 1557–1558). Thus, it has been argued that, because ‘[u]nder a rule of no contribution, each prospective tortfeasor is uncertain what share of the accident cost he will bear. . . misallocations can result’ (Landes & Posner, 1987, p. 203). Landes & Posner (1987, p. 203) analyze and refute this argument. 278. Specifically, injurer A⬘ (or B⬘)s total expected loss depends on whether the other potential injurer is negligent, x(y) < x* (y*) or non-negligent, x(y) = x*(y*). (Landes & Posner, 1987, p. 195). 279. To say that B is non-negligent is to say that B chooses precaution y* (Landes & Posner, 1987, p. 194). 280. If B employs precaution y, B is negligent (Landes & Posner, 1987, p. 194). 281. sAp(x,y,z*)D + A(x) will be less than A(x*) whenever (1) C chooses only to sue B or (2) if C chooses to sue both A and B, A’s share of the collective damages is sufficiently small. 282. This is true because, as discussed above, B will seek to minimize its total expected loss (Kornhauser & Revesz, 1989, p. 836). 283. In the antitrust context, see Easterbrook, Landes & Posner (1980, pp. 344–353). 284. For example, Posner (1992, Economic Analysis, p. 189) concludes that ‘a rule of contribution, which allows a joint tortfeasor made to pay more than his [or her] ‘fair’ share of the plaintiff’s damages to require contribution from the other joint tortfeasors, will also create the right safety incentives for all joint tortfeasors - and this regardless of how the contribution shares are determined (pro rata, relative fault, etc.)’. 285. Easterbrook, Landes & Posner (1980, p. 344) note that ‘any rule apportioning damages produces adequate deterrence if the aggregate damages are properly selected’. 286. Pro rata contribution admits some equitable exceptions, however; thus, equity sometimes requires [1] treating two joint tortfeasors as liable for a single share, or [2] that ‘the share of a tortfeasor who is insolvent or absent from a jurisdiction be borne by others’ (Keeton et al. 1984, pp. 340–341). In addition, ‘[i]n other jurisdictions. . . the distribution of the liability is in proportion to the comparative fault of the defendants’ (Keeton et al. 1984, pp. 340–341). The effect of both the exceptions to the pro rata rule and of the proportional fault rule itself is to alter the values of the ‘individual liability shares’, sA and sB (Easterbrook, Landes & Posner, 1980, p. 350). Such an effect is inconsequential to the analysis presented here. Easterbrook, Landes & Posner (1980, p. 350) note that ‘the specific values of the individual liability shares . . . have no bearing on the analysis (which is the reason no contribution and contribution yield identical effects)’. 287. Again, this is true because, as discussed above, B will seek to minimize its total expected loss (Kornhauser & Revesz, 1989, p. 836). 288. Easterbrook, Landes & Posner (1980, p. 349) employ a similar analysis in the antitrust setting.

214

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

289. Easterbrook, Landes & Posner (1980, p. 357, note 59) conclude that ‘in the multiple-defendant case, it will always be possible to find positive settlement values . . . that make the plaintiff and at least n–1 defendants better off compared to their expected trial outcomes’. 290. By assumption, k < n (Easterbrook, Landes & Posner, 1980, p. 357, note 60). 291. After settlement with n–1 defendants, the expected value of c’s claim against defendant n has been expressed as VECnc = pc(D ⫺ ?Si), where i = 1 to n ⫺ 1 and hence, ?Si = the total settlement received from the settling n–1 defendants. Whether c settles with n depends on whether n’s offer is greater than ECnc (Easterbrook, Landes & Posner, 1980, pp. 354, 355, 357). 292. Di Cola (1992, p. 1555, citing Yi, 1990, pp. 18–20) notes further that ‘[a] nocontribution rule can also create a ‘prisoner’s dilemma’ which tends to increase settlement amounts’ whereas ‘the presence of contribution has a negative effect on the plaintiff’s recovery’. For a more formal proof, see Easterbrook, Landes & Posner (1980, pp. 359–360, 365) who conclude that ‘[a] rule of no contribution creates competition among defendants to settle rather than litigate’. 293. Landes & Posner (1987, pp. 202–203) note that a rule of contribution will ‘discourage defendants from settling’. 294. Di Cola (1992, pp. 1555–1556) notes that ‘contribution rules can be structured in a way that avoids the negative impact on settlement’. 295. A third contribution rule, a ‘carve out’ rule, would ‘reduce[ ] the potential liability of a non-settling defendant by [the] settling defendants’ share of damages (instead of by total settlements)’ (Di Cola, 1992, p. 1556). See also Easterbrook, Landes & Posner (1980, pp. 363–368). 296. Landes & Posner (1987, pp. 202–203) note that ‘the traditional rule in cases where contribution is allowed’ is to allow a joint tortfeasor who loses at trial ‘to get contribution from the [settling defendant]’. Keeton et al. (1984, p. 340) note that ‘[t]he usual holding has been that the [settling] defendant . . . is not released from contribution’ despite the fact that ‘[t]here has been much dissatisfaction with this [rule] because it becomes impossible for a defendant to settle the case’. 297. Easterbrook, Landes & Posner (1980, p. 361) note that what is here termed the traditional rule and the settlement bar rule ‘correspond roughly to the original [1933] and amended [1957] versions of the Uniform Contribution Among Tortfeasors Act’. See also Di Cola (1992, p. 1556). 298. Here it is assumed that a pro rata contribution rule is employed. (Easterbrook, Landes & Posner, 1980, p. 362). 299. Di Cola (1992, p. 1555) notes that ’[i]f [traditional] contribution from settling defendants is allowed, settlement does note bar further liability and . . . each defendant’s incentive to settle decreases’. 300. It seems that a carve-out rule also ‘produces the same incentive to settle as a nocontribution rule’ (Di Cola, 1992, p. 1556). 301. Landes & Posner (1987, p. 203). Landes & Posner (1987, p. 203) note that ‘the settlers may have settled for tiny sums, leaving the nonsettlers to bear the lion’s share of the liability’. See also Di Cola (1992, p. 1556). Similarly, with respect to the carveout rule, it is argued that (1) although those which equally apportion liability are easy to administer, they ‘negatively affect contribution’s fairness objectives’; and (2)

Of Distributive Justice and Economic Efficiency

215

although those which apportion liability according to fault would be fair, they would be ‘very expensive to administer’. (Landes & Posner, 1987, p. 201–203). 302. Indeed, the two carve-out rules may be included in this list as well. However, Stanley (1994, p. 6) notes that ‘[w]hen the analysis is expanded to allow for asymmetric information, previous conclusions do not always hold’. 303. In the federal system and in those state systems modeled after it, this would be accomplished via a cross-claim. 304. Landes & Posner (1987, p. 201) conclude that ‘[n]o contribution is . . . undoubtedly the cheapest to administer’. 305. Di Cola (1992, p. 1553) notes that ‘contribution [is generally thought to both] increase[ ] the cost of litigation’ and ‘discourage settlement . . . without a concomitant increase in deterrence’. 306. We demonstrate the paradoxical nature of the problem in part (III)(B)(3)(a). Landes & Posner (1987, pp. 219–222) hypothesized that ‘contribution [would] be found more often in states that do not weight [traditional] efficiency heavily . . . than in states that do weight it heavily’ but they found only ‘limited support for [their] hypothesis’. 307. See our discussion in part (II)(C)(1). For more details, see Zerbe (2001, Chapter 2, 3). 308. Again, given the regard for others upon which KHZ efficiency is premised, the allocative change will quite possibly be integratively efficient as well, as we noted in part (II)(C)(1). 309. See our discussion in part (II)(C)(1). Recall that the (G-L) term is intended to represent the traditional Kaldor-Hicks efficiency calculation and the (WTP-WTA) term the additional distributional component of KHZ efficiency; that is, the valuation of the change in the distribution of wealth caused by the underlying allocative change. 310. For example, Landes & Posner (1987, pp. 315–316) note that it might be desirable to embed ‘the positive economic theory of tort law’ in ‘a broader economic theory of judicial and legislative action’ because under such a theory ‘such apparent anomalies as the movement[ ] . . . to contribution may disappear’. We argue that the KHZ efficiency theory is just such a ‘broader theory’ (Landes & Posner, 1987, p. 315). 311. Easterbrook, Landes & Posner (1980, pp. 349–350) describe this transfer as ‘only [an] ex post distributional consequence[ ]’ with ‘no offsetting gains in allocative efficiency’. This is, however, a change in the distribution of wealth that is relevant in a calculation of KHZ efficiency. 312. We discussed the increase in administrative cost above. It is here assumed that there is no liability insurance. On the role of such insurance in the contribution/nocontribution debate see Keeton et al. (1984, p. 338). Inclusion of insurance in the KHZ analysis would probably have the effect of increasing the magnitude of L. L, however, would remain positive (Keeton et al. 1984, p. 338). 313. Zerbe (2001, Chapter 3) suggests that the value of distributional changes can be measured in principle as other non-market goods are measured, which is by contingent valuation surveys and by observed behavior. In the absence of contingent valuation surveys, this subsection will rely on anecdotal evidence, which, of course, is simply one form of observed behavior, and, in the present circumstances, the only available form. 314. Landes & Posner (1987, p. 192) note that ‘the rule of no contribution has been denounced by most commentators’.

216

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

315. Di Cola (1992, p. 1563) notes that ‘[t]he impetus behind abrogating the common law proscription of contribution in negligence actions has been a sense that it is unfair and unjust for one tortfeasor to bear the burden for all wrongdoers’. A vast literature details the inequity of no-contribution (Keeton et al. 1984, pp. 337–338). Di Cola (1992, p. 1562) notes that ‘p]erhaps the strongest and most frequent objection to the no-contribution rule is that it is unfair to defendants’. Conversely, contribution is a rule that evolved in equity. See Gomes v. Brodhurst 394 F.2d 465, 467 (3d Cir. 1967)(describing contribution as ‘a rule of reasoned fairness’) Shwartz, Simpson & Arnold (1979, p. 780) note that ‘[b]ecause contribution traces its origin to equity, rather than contract, it serves as a means of assisting in the fair distribution of loss thereby preventing injustice), and which is designed to do justice’. Di Cola (1992, p. 1563) notes that ‘[f]airness considerations have weighted heavily in instances where the courts or Congress have granted a right to contribution’. Comment (1968, p. 736) notes that ‘[t]he values underlying the notion of equal protection or fairness may also support contribution’. Cavanagh (1987, p. 1287) states that ‘[t]he principle rationale for adopting a rule of contribution in antitrust cases is fairness’. Easterbrook, Landes & Posner (1980, p. 340) note that ‘[a] fairness argument from the mouth of the intentional wrongdoer is unappealing because the wrongdoer can avoid his predicament by conforming his conduct to the law’s demands’. However, Higgenbotham & Wiggins (1990, p. 744) hold a contrary view. 316. This, of course, suggests that, contrary to the impost of Di Cola and many other commentators, there is no need to ‘balance anti-contribution efficiency arguments against . . ‘fairness’ [considerations]’. Di Cola (1992, p. 1562) Rather, fairness considerations will, by virtue of the distributional component of KHZ efficiency, often be incorporated in the efficiency analysis itself. 317. For example, Keeton et al. (1984, p. 338) note that ‘[t]he only kind words said by any writer over the last century for the rule denying contribution have been addressed to the proposition that contribution will be used chiefly to permit liability insurance companies to shift a part of the loss’. 318. If the matter of comparative descriptive capacity is to be resolved, it must, in all likelihood, be resolved empirically. There is, of course, a reasonable degree of ambiguity in any statement about a ‘sufficient quantum of future research’, for the term ‘sufficient’ lacks any meaningful definition. 319. That is, as opposed to a non-economic theory. See, for example, Dworkin (1986). We discuss the practical difficulties in part (II)(B)(2)(b).

ACKNOWLEDGMENTS Brian Wasankari wrote the orginal version of this article as his analytic paper for law school under the supervision of Richard Zerbe. It was modified later by Steve Vinyard and Richard Zerbe.

REFERENCES Adamski, M. P. (1981). Contribution and Settlement in Multiparty Actions Under Rule 10b–5. Iowa Law Review, 66, 533–573.

Of Distributive Justice and Economic Efficiency

217

Anderson, E. (1993). Value In Ethics and Economics. Cambridge: Harvard University Press. Armitage, T. (1985). Economic Efficiency as a Legal Norm. Research in Law and Economics, 7, 1–27. Aristotle (1949). Book V of the Nicomachean Ethics of Aristotle ‘On Justice’ In: L. Fuller (Ed.), The Problems of Jurisprudence (pp. 28–64). Brooklyn: Foundation Press. Arrow, K. J. (1973). Social Responsibility and Economic Efficiency. Public Policy, 21, 303–317. Babbie, E. (1989). The Practice of Social Research (5th ed.). Belmont: Wadsworth Publishing Co. Balkin, J. M. (1987). Too Good To Be True: the Positive Economic Theory of Law. Columbia Law Review, 87, 1447–1489. Beutel, F. (1957). Some Potentialities of Experimental Jurisprudence as a New Branch of Social Science. Lincoln: University of Nebraska. Blanck, D. (1996). The Appearance of Justice Revisited. Journal of Criminal Law and Criminology, 86, 887–927. Blum, W., & Kalven, H. Jr. (1965). Public Law Perspectives On a Private Law Problem: Auto Compensation Plans. Boston: Little, Brown. Boadway, R., & Bruce, N. (1984). Welfare Economics. Oxford: B. Blackwell. Bracton, H. de (1968). On the Laws and Customs of England, Vol. 2. Cambridge: Harvard University Press. Brownmiller, S. (1975). Against Our Will: Men, Women, and Rape. New York: Simon and Schuster. Bruce, C. J. (1984). Testing the Hypothesis of Common Law Efficiency: The Doctrine of Informed Consent. Research in Law and Economics, 6, 227–265. Buchanan, A. (1985). Ethics, Efficiency, and the Market. Totowa: Rowman & Allanheld. Buchanan, J. M. (1974). Good Economics – Bad Law. Virginia Law Review, 60, 483–492. Byrns, R. T., & Stone, G. W. (1992). Microeconomics, Vol. 16 (5th ed.). Glenview: Scott. Calabresi, G., & Malamed, A. D. (1972). Property Rules, Liability Rules, and Inalienability: One View of the Cathedral. Harvard Law Review, 85, 1089–1128. Calabresi, G. (1961). Some Thoughts On Risk Distribution and the Law of Torts. Yale Law Journal, 70, 499–553. Calleros, C. (1994). Legal Method and Writing, Vol. 12 (2nd ed.). Boston: Little, Brown. Calnan, A. (1997). Justice and Tort Law. Durhan: Carolina Academic Press. Cavanagh, E. D. (1987). Contribution, Claim Reduction, and Individual Treble Damage Responsibility: Which Path To Reform of Antitrust Remedies? Vanderbilt Law Review, 40, 1277–1336. Chamallas, M. (1988). Consent, Equality, and the Legal Control of Sexual Conduct. Southern California Law Review, 61, 777–862. Coase, R. (1960). The Problem of Social Cost. Journal of Law and Economics, 3, 1–44. Coleman, J. L. (1982). The Economic Analysis of Law. In: J. R. Pennock, & J. W. Chapman (Eds), Ethics, Economics, and the Law (pp. 83–103). New York: New York University Press. Coleman, J. L. (1988). Markets, Morals and the Law. New York: Cambridge University Press. Comment (1968). The Allocation of Loss Among Tortfeasors. Southern Californian Law Review, 41, 728. Confucius (1993). The Analects of Confucius. Trans. Raymond Dawson. New York: Oxford University Press. Cooter, R., & Ulen, T. (1997). Law and Economics (2nd ed.). Glenview: Scott. Cunningham, W. (1992). Testing Posner’s Strong Theory of Wealth Maximization. Georgetown Law Journal, 81, 141–166.

218

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

Dawson, J. P., Harvey, W. B., & Henderson, S. (1993). Cases and Comment on Contracts (6th ed.). Westbury: Foundation Press. Di Cola, E. A. (1992). Fairness and Efficiency: Allowing Contribution Under ERISA. California Law Review, 80, 1543–1545. Dressler, J. (1987). Understanding Criminal Law. New York: M. Bender. Dripps, D. A. (1992). Beyond Rape: an Essay On the Difference Between the Presence of Force and the Absence of Consent. Columbia Law Review, 92, 1780–1809. Duxbury, N. (1995). Law Markets and Valuation. Brooklyn Law Review, 61, 657–701. Dworkin, R. M. (1986). Law’s Empire. Cambridge: Belknap Press. Dworkin, R. M. (1980). Is Wealth a Value? Journal of Legal Studies, 9, 191–226. Easterbrook, F. H., Landes, W. H., & Posner, R. A. (1980). Contribution Among Antitrust Defendants: a Legal and Economic Analysis. Journal of Law and Economics, 23(2), 331–370. Eatwell, J. et al. (Eds) (1987). The New Palgrave: a Dictionary of Economics. London: Macmillan. Eggen, J. M. (1995). Understanding State Contribution Laws and Their Effect on the Settlement of Mass Tort Actions. Texas Law Review, 73, 1701–1777. Ehrenburg, R. G., & Smith, R. S. (1994). Modern Labor Economics. Glenview: Scott. Eisenberg, M. A. (1993). Gilbert Law Summaries: Contracts. Gardena: Harcourt Brace Janovich Legal and Professional Publications. Ellis, D. D. Jr. (1983). An Economic Theory of Intentional Torts: A Comment. International Review of Law and Economics, 3, 45–57. Englard, I. (1993). The Philosophy of Tort Law. Brookfield: Dartmouth. Epstein, R. A. (1997). Property Rules, Liability Rules, and Inalienability: A Twenty-Five Year Retrospective: a Clear View of the Cathedral: the Dominance of Property Rules. Yale Law Journal, 106, 2091–2120. Estrich, S. (1987). Real Rape. Cambridge: Harvard University Press. Everson, G. (1919). The Human Element of Justice. Journal of Criminal Law and Criminology, 10, 90. Farnsworth, E. A. (1982). Contracts (1st ed.). Boston: Little, Brown. Farnsworth, E. A. (1990). Contracts (2nd ed.). Boston: Little, Brown. Flatt, V. B. (1994). The Human Environment of the Mind: Correcting the NEPA Implementation by Treating Environmental Philosophy and Environmental Risk Allocation as Environmental Values Under NEPA. Hasting Law Review, 46, 85–123. Fleming, J. G. (1983). The Law of Torts (6th ed.). Sydney: Law Book Co. Ford, R. L. (1993). The Lawyer’s Bookshelf. New York Law Journal, 209, 108. Fried, C. (1977). Right and Wrong. Cambridge: Harvard University Press. Friedman, M. (1953). Essays in Positive Economics. Chicago: University of Chicago Press. Fuller, L. (Ed.) (1949). The Problems of Jurisprudence. Brooklyn: Foundation Press. Gadre, R. (1994). The Integrity of Efficiency: an Analysis and Synthesis of Law and Economics and Ronald Dworkin’s Law as Integrity. Working paper. Giancoli, D. (1991). Physics (3rd ed.). Englewood Cliffs: Prentice Hall. Glick, H. R. (1990). Courts in American Politics. New York: McGraw-Hill. Gordon, M. T., & Rigger, S. (1991). The Female Fear: The Social Cost of Rape. New York: Free Press. Gregory, C. (1938). Contribution Among Tortfeasors: A Uniform Practice. Wisconsin Law Review, 365–401.

Of Distributive Justice and Economic Efficiency

219

Greenawalt, K. (1995). Legal Enforcement of Morality. Journal of Criminal Law and Criminology, 85, 710–725. Hernandez, M. V. (1997). Cost Recovery or Contribution. Harvard Environmental Law Review, 21(1), 83–136. Higgenbotham, E. R., & Wiggins, K. B. (1990). Contribution and Indemnity. ALI-ABA Course of Study, Trial Evidence, Civil Practice and Effective Litigation Techniques, C491, 697. Higgens, M. (1997). Is Capital Punishment for Killers Only? ABA Journal, 83, 30. Hill, C. G. (1985). Note, Public Employees and Private Conduct: Cohabitation and the Vagueness of Immortality. Journal of Family Law, 23, 111–132. James, F. (1941). Contribution Among Joint Tortfeasors: A Pragmatic Criticism. Harvard Law Review, 54, 1156–1183. Kadish, S. H., & Schulhofer, S. J. (1995). Criminal Law and its Processes (6th ed.). Boston: Little, Brown. Kaldor, N. (1939). Welfare Propositions in Economics and Inter-Personal Comparisons of Utility. Economic Journal, 49, 549–697. Katz, A. (1996). Positivism and the Separation of Law and Economics. Michigan Law Review, 94(7), 2229–2269. Katz, M., & Rosen, H. S. (1998). Microeconomics (3rd ed.). Boston: McGraw-Hill. Kaufman, C. (1980). The Scientific Method in Legal Thought: Legal Realism and the Fourteen Principles of Justice. St. Mary’s Law Journal, 12, 77–112. Keeton et al. (1984). Prosser and Keaton on the Law of Torts (5th ed.). St. Paul: West Publishing Company. Kornhauser, L. A. (1980). A Guide to the Perplexed Claims of Efficiency in the Law. Hofstra Law Review, 8, 591–639. Kornhauser, L. A., & Revesz, R. L. (1989). Sharing Damages Among Multiple Tortfeasors. Yale Law Journal, 98, 831–884. Krepps, D. 1990. A Course in Microeconomic Theory. Princeton: Princeton University. Kuhn, T. K. (1970). The Structure of Scientific Revolutions (2nd ed.). Chicago: University of Chicago Press. Kuperberg, M., & Beitz, C. (Eds) (1983). Law, Economics, and Philosophy, Vol. 3. Totowa: Rowman & Allanheld. Landes, W. M., & Posner, R. A. (1980). Joint and Multiple Tortfeasors, an Economic Analysis. Journal of Legal Studies, 9(3), 517–555. Landes, W. M., & Posner, R. A. (1987). The Economic Structure of Tort Law. Cambridge: Harvard University Press. Lec, S. J. (1962). Unkempt Thoughts. Jacek Galazka, trans. New York: St. Martins Press. Leflar, R. A. (1932). Contribution and Indemnity Between Tortfeasors. University of Pennsylvania Law Review, 81, 130. Locke, J. (1962). Second Treatise on Civil Government: Essays by Locke, Hume, and Rousseau. In: E. Baker (Ed.), Social Contract (pp. 3–143). New York: Oxford University Press. Maitland, F. W., & Pollock, F. 1968. The History of English Law Before the Time of Edward I. London: Cambridge University Press. Manzer, N. L. (1988). 1986 Tort Reform Legislation: a Systematic Evaluation of Caps on Damages and Limitations on Joint and Several Liability. Cornell Law Review, 73(3), 628–652. Malamut, M. (1993). Proposal for Revision of Archaic Statutes Implicating Private Consensual Noncommercial Adult Sexual Conduct. Law and Sexuality, 3, 45–126. Mansfield, E. (1994). Microeconomics, Vol. 13 (8th ed.). New York: W. W. Norton.

220

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

Nietzsche, F. (1968). Thus Spoke Zarathustra. In: Walter Kaufmann (Ed.), Portable Nietzsche. New York: Viking. Okun, A. (1975). Equality and Efficiency: The Big Tradeoff. Washington: The Brookings Institute. Parsons, W. (1983). Note: The Inefficient Common Law. Yale Law Journal, 92(5), 862–887. Polinsky, A. M., & Shavell, S. (1981). Contribution and Claim Reduction Among Antitrust Defendants, An Economic Analysis. Stanford Law Review, 33, 447–471. Polinsky, A. M. (1983). An Introduction to Law and Economics (1st ed.). Boston: Little, Brown. Polinsky, A. M. (1989). An Introduction to Law and Economics (2nd ed.). Boston: Little, Brown. Posner, R. A. (1975). The Economic Approach to Law. Texas Law Review, 53, 757. Posner, R. A. (1979). Utilitarianism, Economics, and Judicial Theory. Journal of Legal Studies, 8, 103. Posner, R. A. (1981). A Reply to Some Recent Criticisms of the Efficiency Theory of the Common Law. Hofstra Law Review, 9, 775. Posner, R. A. (1985). An Economic Theory of the Criminal Law. Columbia Law Review, 85, 1193–1231. Posner, R. A. (1987). The Justice of Economics. Journal of Public Finance and Public Choice, 15, 23–24. Posner, R. (1992). Economic Analysis of Law (4th ed.). Boston: Little, Brown. Posner, R. A. (1992). Sex and Reason. Cambridge: Harvard University Press. Pyle, D. J. (1983). The Economics of Crime and Law Enforcement. London: Macmillan. Radin, M. J. (1987). Market Inalienability. Harvard Law Review, 100, 1849–1937. Radin, M. (1993). Reinterpreting Property. Chicago: University of Chicago Press. Rawls, J. (1971). A Theory of Justice. Cambridge: Belknap Press. Reath, T. W. (1898). Contribution Between Persons Jointly Charged For Negligence – Merrywether v. Nixan. Harvard Law Review, 12, 176. Restatement of the Law, Second, of Torts .2d. 1965. St. Paul: American Law Institute. Restatement of the Law, Second – Contracts .2d. 1981. St. Paul: American Law Institute. Rose, J. (1980). Contribution in Antitrust: Some Policy Considerations. Antitrust Law Journal, 48, 1605–1607. Roth, T. (1991). Law and Economics. University of Chicago Magazine, 83(6), 28. Ryan, J. P. et al. (1980). American Trial Judges: Their Work Styles and Performance. New York: Free Press. Sachs, M. V. (1994). Freedom of Contract: The Trojan Horse of Rule 10b–5. Washington and Lee Law Review, 51, 879–924. Schell, G. R. (1993). Contracts in the Modern Supreme Court. California Law Review, 81, 431–529. Schwartz, G. T. (1979). Economics, Wealth Distribution, and Justice. Wisconsin Law Review, 799–813. Schwartz, S. R., & Simpson, N. H. et al. (1979). Contribution in Private Actions Under the Federal Antitrust Laws. Southwestern Law Journal, 33, 779–822. Shavell, S. (1981). A Note On Efficiency vs. Distributional Equity In Legal Rulemaking: Should Distributional Equity Matter Given Optimal Income Taxation? American Economic Review Papers and Procedings, 71, 414. Shavell, S. (1987). Economic Analysis of Accident Law. Cambridge: Harvard University Press. Spohn, C., & Horney, J. (1992). Rape Law Reform: a Grassroots Revolution and Its Impact. New York: Plenum Press.

Of Distributive Justice and Economic Efficiency

221

Stanley, T. J. (1994). An Analysis of the Rules of Contribution and No Contribution For Joint and Several Liability In Conspiracy Cases. Santa Clara Law Review, 35, 1–122. Talbott, W. (1988). Cost Spreading and Benefit Spreading In Tort Law. Research In Law and Economics, 11, 25–51. Tullock, G. (1980). Two Kinds of Efficiency. Hofstra Law Review, 8, 659–669. Wade, J. W. et al. (1994). Cases and Materials on Torts (9th ed.). Westbury: Foundation Press. Wald, A. (1997). What’s Rightfully Ours: Toward a Property Theory of Rape. Columbia Journal of Law and Social Problems, 30, 459–502. West, R. (1986). Submission, Choice, and Ethics: A Rejoiner to Judge Posner. Harvard Law Review, 99, 1449–1456. West, R. (1993). Sex, Reason, and a Taste For the Absurd. Georgetown Law Journal, 81, 2413–2430. Williams, G. (1983). Textbook of Criminal Law. London: Stevens. Wohl, A. (1997). Paper Trailblazer. ABA Journal, 83, 68–73. Yi, J. G. (1990). Litigations with Multiple Defendants: How to Settle Under Different Apportionment Rules. Stanford University Center for Economic Policy Research Publication No. 255. Zelder, M. (1993). Incompletely Reasoned Sex: a Review of Posner’s Somewhat Misleading Guide to the Economic Analysis of Sex and Family Law. Michigan Law Review, 91, 1584–1608. Zerbe, R. O. Jr. (1998). An Integration of Equity and Efficiency. Washington Law Review, 73, 349–361. Zerbe, R. O, Jr. (1998). Is Cost-Benefit Analysis Legal? Three Rules. Journal of Policy Analysis and Management, 17(3), 419. Zerbe, R. O. Jr. A Foundation For Economy Efficiency In Law and Economics. Aldershot: Edward Elgar (to be published 2001). Zerbe, R. O. Jr., & Dively, D. (1994). Benefit Cost Analysis in Theory and Practice. New York: Harper Collins.

LEGAL CASES Alack v. Vic Tanny Intl. of Missouri, Inc., 923 S.W.2d 330 (Mo. 1996). American Motorcycle Ass’n v. Superior Court 578 P.2d 899, 907 (Cal. 1978). Avnet, Inc. v. Allied Signal, Inc., 825 F. Supp. 1132, 1138 (D.R. I. 1992). Batsakis v. Demotsis, 226 S.W.2d 673 (Tex. Civ. App. 1949). Bedell v. Reagan, 192 A.2d 24 (Me. 1963). Best v. Yerkes, 77 N.W.2d 23 (Iowa 1956). Bisso v. Inland Waterways Corp., 349 U. S. 85 (1955). Blomgren v. Marshall Management Services, Inc., 483 N.W.2d 504, 506 (Minn. Ct. App. 1992). Blyth v. Birmingham Waterworks Co., 156 Eng. Rep. 1047 (1856). Coker v. Georgia, 433 U. S. 584 (1977). Cole v. Celotex Corp., 559 So. 2d 1058, 1070 (La. 1992). Davis v. Broad St. Garage, 191 Tenn. 320 (1950). Ellis v. Chicago and N. W. Ry., 167 Wis. 392 (1918). Garretson v. United, 456 F.2d 1017 (9th Cir. 1972). Goldman v. Mitchell-Fletcher Co., 292 Pa. 354 (1928.

222

BRIAN N. WASANKARI, RICHARD O. ZERBE JR. & STEVEN VINYARD

Gomes v. Brodhurst. 394 F.2d 465, 467 (3d Cir. 1967). Grothe v. Shaffer, 305 Minn. 17 (1975). Knell v. Feltman, 174 F.2d 662 (D. C. Cir. 1949). Merrywether v. Nixan, 101 Eng. Rept. 1337 (K. B. 1898). Mich. Comp. Laws Ann. §§750.520a -.5201 (West Supp. 1986). Miller v. Stouffer, 11 Cal. Rptr.2d 454 (1992). Palmer v. Wick and Pultneytown Steam Shipping Co., A. C. 318 (1894). Penn. Co. v. West Penn Ry., 144 N. E. 51 (Ohio 1924). Quatray v. Wicker, 178 La. 289 (1933). Roberts v. Robert V. Rohrman, Inc., 909 F. Supp. 545, 553 (N. D.Ill. 1995). Russ v. Woodside Homes, Inc., 905 P.2d 901, 905 (Utah Ct. App. 1995. Somer v. Federal Signal Corp., 593 N.E.2d 1365, 1372 (N. Y. 1992). State v. Wilson, 685 So. 2d 1063 (La. 1996). Tunkl v. Regents of University of California, 60 Cal.2d 92 (1963). U. S. v. Bishop, 66 F.3d 569 (1995). U. S. v. Carroll Towing Co., 159 F.2d 169 (1947). U. S. v. Conservation Chemical Co., U. S. Dist. Lexis 18283 (W. D.Mo. 1988). U. S. v. Conservation Chemical Co., 619 F. Supp. 162, 224 (W. D.Mo. 1985). Vickers Petrolium Co. v. Biffle3, 239 F.2d 602, 606 (10th Cir. 1956). Weiner v. United Air Lines, 216 F. Supp. 701 (S. D.Cal. 1962).

THE DURATION OF PERSONAL INJURY LITIGATION Stephen J. Spurr ABSTRACT This paper examines and tests the implications of the court congestion hypothesis of Posner (1972) and Priest (1989). This hypothesis suggests that the effects of delay reduction programs may be largely or completely offset by a resulting increase in demand for litigation. We also analyze the effect on the pace of litigation of certain factors not explicitly considered by the Posner-Priest model. We find that a delay reduction program was successful in reducing the time to settlement, and the time to verdict for tried cases, over the period covered by the data. As for behavioral responses, there is no evidence of an increase in the trial rate; there is, however, strong evidence of an increase in the number of cases filed annually in the court in question. Other results are consistent with the model’s predictions about the effects of the prejudgment rate of interest.

1. INTRODUCTION These days it is increasingly difficult to find areas of legal research untouched by economic analysis. The problem of court delay is no exception to this generalization. The application of the economic approach by Posner (1972) and Priest (1989) has enriched the literature of this subject by providing a new set of issues to be considered and new hypotheses to be tested. Their analysis of court congestion suggests that the effects of delay reduction programs may be Research in Law and Economics, Volume 19, pages 223–246. Copyright © 2000 by JAI/Elsevier Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0308-5

223

224

STEPHEN J. SPURR

only transitory, since initial improvements may be swamped by an offsetting increase in demand for litigation. This paper has two objectives. One is to examine and test the implications of the court congestion hypothesis of Posner and Priest. The other objective (required in order to achieve the first) is to analyze the factors that determine the duration of litigation. We first review the model developed formally by Priest, which concerns the choice the parties make between settling the case and going to trial. The theory has implications for the effect of certain variables, such as the expected time to trial and the rate of pre-judgment interest. We consider how an increase in demand for litigation would be reflected on various intensive and extensive margins. We then analyze the determinants of duration with two data sets, one from Michigan and the other from New York.

2. THE DATA We have used two new data sets to test the court congestion hypothesis and to investigate the factors which affect the duration of litigation. These data sets are complementary, in that each one has variables relevant to duration not available in the other one. The first data set has 7,270 medical malpractice claims filed in Michigan, almost all of which were closed between 1978 and 1990. This data set is of interest because of concerns about delay in medical malpractice cases, which have prompted procedural reforms aimed specifically at this type of litigation. In Illinois, for example, a special trial division was created to enable certain judges of the Cook County Circuit Court to specialize in handling medical malpractice litigation.1 The second data set has 113,965 claims for personal injuries or wrongful death filed in various courts in the First Judicial Department of New York (Manhattan and the Bronx) during the period from 1974 through 1984. This data set includes variables that are rarely available, indicating how the attorney obtained the case, and whether the case was referred by one attorney to another under an arrangement whereby the contingent legal fee is divided between the attorneys. Any lawyer who receives a referral from another lawyer under these circumstances is sure to be a specialist in personal injury litigation.

3. THE PROGRAM TO REDUCE COURT DELAY IN WAYNE COUNTY Forty-five per cent of the claims in the Michigan data set were filed in Wayne County, which accounted for about 25% of the state’s population in 1980.2

The Duration of Personal Injury Litigation

225

During the period covered by this data set, the Wayne County Circuit Court adopted a new scheduling system designed to expedite the disposition of cases. Under the previous system a case was scheduled for a mediation hearing either 18 or 27 months after filing. If the case was not resolved at mediation, it was scheduled for a settlement conference and trial. Normally there would be no intervention by a judge in a case before trial, unless a motion that required a ruling had been filed. Moreover, the date set for trial was lightly regarded by all parties, since continuances were granted as a matter of course. The system has been described as follows: The trial scheduling system placed little emphasis on ensuring that trial dates were firm. Trials were scheduled for about 90 days after the mediation hearing, but the main focus of the first scheduled trial date was on settlement rather than trial. Every scheduled case was first sent to a settlement conference. . . . A significant percentage settled at the conference, but many did not. Most cases that did not settle were continued to a new trial date, usually six months in the future. Not infrequently another continuance would occur because of an overset trial calendar. By about the third trial date, the case would receive priority for trial, and a continuance would be harder to obtain. Finally, faced with the imminence of a trial, the lawyers at last would reach a settlement in most cases.3

In contrast, under the new system each case was assigned to an individual judge, who was responsible for monitoring its progress until its final disposition. Shortly after the case was filed4 the judge would meet with the lawyers at a ‘judicial status conference’, and then issue a scheduling order which specified actual dates for several important events in the litigation process: the witness list exchange, the discovery cutoff date, the date of mediation, and the settlement or final pre-trial conference. The attorneys could choose to put the case on any of four possible tracks. On Tracks number 1, 2 and 3, there is a prescribed interval of time between each event. On Track 1, for example, the witness list exchange is scheduled to occur 175 days after the claim is filed, discovery is to be completed 224 days after filing, mediation is nine months after filing, and the settlement conference is scheduled 42 days after the mediation date. Track 1 is the fastest of the three tracks and Track 3 is the slowest. On Track 4 there is no prescribed schedule; the attorneys can design a schedule tailored to their preferences. This track was originally intended for complex litigation. A court administrator5 informed us that most attorneys seemed to prefer Track 1, because it was the fastest, until Track 4 became an option. Many attorneys then chose Track 4, since they could use it to construct a schedule faster than Track 1. The track system was introduced into the court on January 2, 1987, and was phased in on a judge-by-judge basis.6 The principal difference between the old and new systems was that under the new system, the judge would intervene early in the litigation process

226

STEPHEN J. SPURR

and, after consulting with the lawyers, would impose a rigid time schedule on the major events of the litigation. Somerlot and colleagues (1989) found that the new program was successful in reducing the ‘case-processing time’, i.e. the time from filing to disposition, of all civil cases in the Wayne County Circuit Court from the beginning of the program through 1989.

4. THE COURT CONGESTION HYPOTHESIS The analysis of court congestion by Posner (1972) and Priest (1989) suggests that programs to reduce delay, such as the one adopted in Wayne County, are to some extent self-defeating since initial improvements may quickly lead to an increase in demand for trials.7 It is useful to review the model developed by Priest to understand why a reduction of delay would lead to an increased demand for trials. After we analyze this model, we will consider other factors that may be important in determining the duration of litigation. Priest’s model builds on the economic model of litigation of Landes, Posner & Gould, which is based on the idea that cases go to trial only if the parties have different beliefs about the outcome of the trial. According to this paradigm, the parties will settle the case if the minimum amount the plaintiff is willing to accept is less than the maximum amount the defendant is willing to pay; otherwise the case will be tried. Let Pp be the probability that the plaintiff will win, as the plaintiff sees it, and Pd the probability that the plaintiff wins, as the defendant sees it. X is the amount the plaintiff will recover if he wins at trial. Finally, let Cp and Cd be the costs of trial for the plaintiff and defendant, respectively, and let Sp and Sd be their respective costs of settling the case. Then the plaintiff will accept some settlement payment M if M ⫺ Sp ⭓ PpX ⫺ Cp ⇒ M ⭓ PpX ⫺ Cp + Sp Similarly, the defendant would be willing to settle for some amount M if M + Sd ⭐ PdX + Cd ⇒ M ⭐ PdX + Cd ⫺ Sd The case goes to trial if the maximum amount the defendant will pay is less than the minimum the plaintiff will accept, i.e. if PdX + Cd ⫺ Sd < PpX ⫺ Cp + Sp ⇒ X(Pp ⫺ Pd) > Cd + Cp ⫺ (Sp + Sd)

(1)

Thus the probability of trial increases the greater the divergence of the parties’ opinions about the probability of winning at trial ( the greater Pp ⫺ Pd), and the larger the costs of settlement are relative to the costs of trial. Priest pointed out that the amount of the trial judgment X might be paid by the defendant much later than the parties’ expenses for either trial or

The Duration of Personal Injury Litigation

227

settlement. If certain conditions apply, namely that (A) all expenses for either trial or settlement are made in the present, (B) the judgment X is paid t years later, and (C) there is no provision for prejudgment interest, equation (1) would be replaced by a new condition for trial: X (Pp ⫺ Pd) > Cd + Cp ⫺ (Sp + Sd) (1 + r)t

(2)

Priest argued from equation (2) that delay reduces the likelihood of trial. The larger t is, the less likely it is that the left-hand side of (2) will be larger than the right-hand side. Equation (2) makes a strong assumption about the timing of the judgment in relation to the timing of trial and settlement expenses. If we assume instead that only settlement expenses are made in the present and that all trial expenses are incurred at the time of trial, equation (2) would be modified by dividing Cp + Cd by (1 + r)t. In this case an increase in delay, reflected in an increase in t, would have less of an effect of increasing the probability of trial, especially if trial expenses are large relative to settlement expenses (if settlement expenses were zero, an increase in t would have no effect whatever). Also, as Priest noted, if X the plaintiff is entitled to prejudgment interest, the term must be (1 + r)t X replaced by , where i is the prejudgment rate of interest specified by (1 + r ⫺ i)t law. According to equation (2), if i exceeds r, an increase in delay (t) will increase the probability of trial, since it will actually increase the present value of the amount at stake in the litigation. One qualification to Priest’s thesis should be noted. If damages are determined in current prices, i.e. according to wages and other prices prevailing at the time of trial rather than at the time of the injury, the interest rate which should be used for pre-judgment interest is the real rate of interest rather than the nominal rate. That is, the r in equation (2) should be understood to be the real rather than the nominal rate of interest.8 Another point in support of Priest’s thesis – that delay makes trial less likely – is that litigation expenses tend to increase with delay. As the years pass, witnesses may move away or die, or their recollection of events will fade, making it necessary to incur the costs of recording their testimony; moreover, the lawyers assigned to the case may leave their firms, retire, or die, and will then be replaced by others who must learn anew the facts and the applicable law. See generally Miller (1989). However, a far more important consideration may be the cost to the parties of uncertainty about the outcome, and of

228

STEPHEN J. SPURR

prolonged anticipation of the psychic and emotional costs of a trial. All these factors will increase Cp + Cd in equation (2), thereby reducing the probability of trial. To determine the effects of delay, Priest’s model focuses on the choice the parties make between settlement and trial, for a claim already filed in court. In general, however, it will be important to distinguish between the intensive and extensive margins. On the intensive margin, a dramatic reduction in the time to trial resulting from, say, an increase in the number of judges, will induce some litigants in pending cases to opt for trial rather than settlement. On the extensive margin, plaintiffs’ lawyers who would otherwise have chosen some alternative course will now choose to file an action in the court in which delay has been reduced. These alternatives could include, for example, filing in some other court, arbitration, or not suing at all. The elasticities on these two margins may be quite different. It may be, for example, that suing in court A is not a good substitute for suing in court B, because strict rules on jurisdiction and venue make forum-shopping extremely difficult, i.e. costly. In this case, the elasticity of demand on the extensive margin may be small and a reduction of delay in court A will not cause much of an increase in the number of cases filed. On the other hand, if the parties in many cases pending in court A are almost indifferent about the choice between settlement and trial, the elasticity on the intensive margin will be large, and a reduction in delay will lead to a large increase in the number of trials. In line with the preceding discussion, we consider two margins on which there could be a response to the improvements in the pace of litigation in the Wayne County Circuit Court: (1) on the extensive margin, there could be an increase in the number of cases filed; and (2) on the intensive margin, litigants in pending cases who previously would have settled may now choose to litigate to verdict. With regard to the extensive margin, Table 1 shows that the number of new civil cases filed in Wayne County Circuit Court increased by 26% between 1987, the year the new tracking system began to be phased in, and 1993. This increase, which is highly significant (p-value = 0.0002), is certainly consistent with Priest’s congestion hypothesis, and we do not know of any alternative explanation. The change cannot be attributed to an increase in population, since the population of Wayne County declined by 9.7% between 1980 and 1990.9 For comparison, column 3 shows the number of divorce cases filed in this court during these years; we suspected that the number of these actions was relatively inelastic with respect to delay, i.e. that few persons decide to divorce their spouses because of a reduction in the time to trial. There does not seem to be any time trend in the number of divorce cases filed (pvalue = 0.444), which lends some support to our conjecture that the most

229

The Duration of Personal Injury Litigation

plausible explanation for the increase in filings of civil cases is the reduction in court delay.10 With respect to the intensive margin, Table 1 shows the ratio of total trial verdicts (jury or nonjury) to the total cases settled between 1987 and 1993. There appears to have been a slight decline in the probability of litigating to verdict over this period.11 Column 6 indicates that the annual number of trial verdicts in this court has also declined somewhat since 1987 (p-value = 0.0308). These results suggest that the demand for litigation at the extensive margin is considerably more elastic than the demand for trials at the intensive margin.

5. THE CAUSES OF COURT DELAY A major shortcoming of the Posner-Priest hypothesis is that it attempts to explain dynamic behavior, the length of time until the case is resolved, with a static model, i.e. the choice between settlement and trial. Moreover, the spirit of the hypothesis is that a court reform that brings about a substantial reduction in the time to trial will lead not only to (1) an increase in the frequency of trials, but also to (2) an increase in the number of cases filed, in the court that has seen the improvement, and (3) an increase in the time to settlement, relative to the situation immediately after the improvement. However, the responses of types (2) and (3) are not specified by the static model.

Table 1. Civil Litigation in Wayne County Circuit Court, 1987–1993 New Cases Filed

Year

All Civil Cases

Divorce Cases

Total Civil Cases Pending

1987 1988 1989 1990 1991 1992 1993

16790 16961 18021 18356 20192 21073 21186

11342 11070 11145 11406 10843 11191 11081

61605 54837 51698 58294 44118 44741 43395

Settled Cases

Trial Verdicts in Civil Cases (Jury and Nonjury)

7794 8006 8365 7824 8230 8689 7590

780 806 858 723 641 664 657

Ratio of Trial Verdicts to Settled Cases 0.100 0.101 0.103 0.092 0.078 0.076 0.087

Source: Sally A. Mamo, Director of Docket Management, Third Judicial Circuit of Michigan, 770 City-County Building, Detroit, Michigan 48226.

230

STEPHEN J. SPURR

There are of course many factors outside the Posner-Priest model which determine the duration of litigation. Information about a case is gathered by the parties over time. This information comes from research by the party’s lawyer, through discovery, and through observation of the behavior of the other party. Factors that expedite the collection of information will generally reduce the time to settlement. It appears that the tracking system adopted by the Wayne County Court leads to earlier discovery, which leads to earlier settlements. Delay can also be caused by strategic behavior unrelated to the parties’ information, or by a desire for revenge. In addition, the incentives of the lawyer may be important. Shepherd (1995) found that plaintiffs’ attorneys who were paid on an hourly basis conducted approximately six more days of discovery than plaintiffs’ attorneys working under a contingent fee arrangement. We have noted that delay can result from the deliberate actions of the parties. It is possible to measure the relative importance of this source of delay. Miller (1989) has pointed out that if there is no prejudgment interest, or the rate of interest allowed is substantially lower than r, the market rate of interest, defendants have an incentive to delay litigation. Conversely, if the prejudgment rate of interest exceeds r, plaintiffs have an incentive to delay. If the prejudgment rate of interest does not deviate much from the prevailing market rate, the incentives of the parties to delay litigation should be minimized. In Michigan the rules on prejudgment interest changed during the period covered by our data. For complaints filed before 1987, plaintiffs were entitled to prejudgment interest at a rate of 12%. For complaints filed thereafter, interest was allowed at a rate equal to 1% plus the rate paid at semi-annual auctions of five-year U. S. Treasury notes.12 Table 2 shows that the new legal rates were well below 10% in the two years after 1986; in general the legal rate is still above the nominal rate, but the disparity is much less than before. It is interesting that in 1985, when the prejudgment rate of 12% was in effect, the Michigan Commissioner of Insurance reported an allegation by a large insurance company that this rate led plaintiff’s attorneys to postpone cases repeatedly.13 A reasonable conclusion is that the 1987 change in the rules on pre-judgment interest reduced the incentive of plaintiffs to delay litigation. In New York, on the other hand, interest generally did not accrue in personal injury actions until a verdict was rendered.14 We would infer that defendants had a strong incentive to delay litigation in New York.

6. DETERMINANTS OF THE TIME TO SETTLEMENT It is well known that only a small fraction of cases are litigated to verdict. In the Michigan data set on malpractice claims, 8% of claims were litigated to

231

The Duration of Personal Injury Litigation

Table 2. The Rate of Pre-Judgement Interest Allowed in Michigan Year

Legal Rate of PreJudgement Interest

Nominal Interest Rate

Rate of Inflation

1982 1983 1984 1985 1986 1987 1988 1989 1990

12 12 12 12 12 7.66 and 8.50 9.39 and 9.21 10.005 and 10.105 9.015 and 9.535

10.69 8.63 9.58 7.48 5.98 5.82 6.69 8.12 7.51

6.2 4.1 4.4 3.7 2.6 3.2 3.9 4.4 4.4

Note: the source for the inflation rates is Abel and Bernanke (1995), while the nominal interest rate is the rate for new issues of three-month Treasury bills, reported in Table C–71 of the Economic Report of the President (1996). The source for the legal rates of pre-judgement interest after 1986 is the Office of the Treasurer, Lansing, Michigan; for each year, the first rate quoted applies to the first six months and the second to the last six months.

verdict, 45% were settled, and the rest were abandoned by the plaintiff. Thus an analysis of how long it takes for a case to be settled or abandoned deserves center stage in a study of the duration of litigation. With respect to abandoned claims, it is generally not feasible to determine the date of abandonment with reasonable accuracy. Hence we have focused on the duration of claims which are settled or litigated. In our analysis of the Michigan data, we use two econometric methods to determine the effects of variables on the time to settlement of a claim. First, we do a regression by ordinary least squares, in which the dependent variable is the log of time to settlement. Next, we do a duration analysis with four alternative models of the hazard function – the Weibull, lognormal, log-logistic, and Gamma specifications. Duration analysis is especially useful when some of the data is censored. Broadly speaking, an observation is censored if it provides only partial information about the variable of interest. For our purposes, a case is censored if it does not provide complete information about the time to settlement. This is true for two kinds of cases in our data set: (1) those which are resolved when the mediation award is accepted by both parties,15 and (2) those which are litigated to verdict. From a statistical point of view, we are taking the position that (a) we are unable to observe when these claims would have been settled because of the occurrence of an intervening event, i.e. the parties’ acceptance of the mediation award or the court’s arrival at a verdict, but

232

STEPHEN J. SPURR

that (b) we know that settlement did not occur prior to that event. The advantage of duration analysis is that it exploits the information available in censored claims, while the regression analysis makes use only of settled claims.16 The independent variables used in all specifications are described in Table 3. There are dummy variables for eight different levels of severity of injury of the claimant ranging from severity level one, emotional damage only, to level nine, death. In general, the amount of damages at stake will increase with the severity of the injury, except that claims based on injuries at levels seven and eight, ‘permanent major’ and ‘permanent grave’ injuries, yield recoveries substantially larger than those based on the victim’s death, since these extremely serious but nonfatal injuries generally require large medical and rehabilitation expenses, often including lifelong care.17 We would expect that the time it takes to settle the case will increase with the potential damages. Thus the coefficients of these variables should generally increase with the level of severity, but be largest for levels seven and eight. These coefficients will be interpreted relative to the omitted category, severity of level three – a ‘temporary minor’ injury. The variable ‘MULTIPLE DEFENDANTS’ indicates whether more than one defendant contributed to the plaintiff’s recovery. The presumption is that when more than one defendant is liable, more than two parties are involved in

Table 3. Description of Variables in Michigan Data Set Variable

Description

Multiple Defendants

Equals 1 if more than one defendant contributed to the amount paid in settlement; otherwise it equals 0. The age of the claimant on the date of injury. Equals 1 if the claimant’s medical expenses were paid by Medicaid; 0 otherwise. Equals 1 if the claimant’s medical expenses were paid by private health insurance; 0 otherwise. Equals 1 if the action was filed in a court in 1982; 0 otherwise. The dummy variables for other years of filing are similarly constructed. Equals 1 if the action was filed in a court in Wayne County; 0 otherwise. The product of Filed 82 and Wayne County. The interaction terms for Wayne County and other years are similarly constructed. Level One equals 1 if the level of severity of the claim is one; 0 otherwise. The other dummy variables for levels for severity, Level Two through Level Eight, are similarly constructed.

Age Medicaid Health Insurance Filed 82 Wayne County Wayne County – 82 Level One – Level Eight

The Duration of Personal Injury Litigation

233

settlement negotiations. In general, the costs of reaching an agreement will increase with the number of parties involved. Similarly, we expect that the time required to arrive at a settlement will increase with the number of parties whose agreement must be obtained. Priest (1989) found that in medical malpractice cases the length of a trial increased with the number of defendants.18 There are two dummy variables corresponding to the manner in which the victim’s medical expenses were paid. One indicates whether payment was made by Medicaid, and the other indicates payment by private health insurance. The excluded possibility is payment by Medicare or some other unspecified source. These variables should be related to the victim’s level of income: payment by Medicaid may indicate a low income, and payment by health insurance will generally indicate a high income. We expect that the time required to settle the case will increase with the victim’s income level, since time to settlement should increase with the potential damages, and damages for disability and lost income are greater for those with relatively high incomes. Similarly, we expect age to have a negative effect, since damages are based on foregone earnings, which increase with the victim’s life expectancy. Also included are dummy variables for (1) the year the case was filed, for 1982 through 1988, (2) a dummy variable for Wayne County, and (3) variables representing the interaction of Wayne County and each year of filing. These variables should enable us to sort out the relative influences of several different factors. We know that there were changes in the rules on pre-judgment interest which would have reduced the incentive of plaintiffs to delay litigation after 1986. In addition, the new case scheduling system designed to expedite litigation was adopted by the Wayne County Circuit Court beginning in 1987. Thus, if we believe duration is strongly influenced by the rules on pre-judgment interest, we would expect a statewide reduction in court delay and a negative coefficient for the variables representing 1987 and 1988. Since the Wayne County Circuit Court was a pathological example of delay in civil litigation before 1987,19 we would expect the dummy variable for Wayne County to be positive and significant. Finally, the interaction variables will enable us to determine whether any differential in delay between Wayne County and the rest of the State was reduced by the Wayne County Court’s delay reduction program.

7. RESULTS Results obtained from the Michigan data are set forth in Table 4, which reports ordinary least squares (OLS) estimates from settled cases, and Table 5, which provides estimates from the duration models. In the OLS estimates for settled

234

STEPHEN J. SPURR

Table 4. Effect of Variables on Time to Settlement: O.L.S. Estimates from Michigan Data Variable

Mean

Standard Deviation

Log (Time to Settlement) (the dependent variable)

0.814

0.566

0.610 35.398 0.064 0.156 0.087 0.154 0.144 0.135 0.198 0.076 0.073 0.420 0.033 0.051 0.056 0.064 0.092 0.028 0.031 0.021 0.033 0.076 0.143 0.125 0.083 0.050 0.291

0.488 22.67 0.244 0.363 0.282 0.361 0.351 0.342 0.399 0.264 0.260 0.494 0.179 0.220 0.230 0.245 0.289 0.166 0.174 0.144 0.178 0.266 0.350 0.331 0.276 0.217 0.454

One (Intercept) Multiple Defendants Age Medicaid Health Insurance Filed 82 Filed 83 Filed 84 Filed 85 Filed 86 Filed 87 Filed 88 Wayne County Wayne County – 82 Wayne County – 83 Wayne County – 84 Wayne County – 85 Wayne County – 86 Wayne County – 87 Wayne County – 88 Level One Level Two Level Four Level Five Level Six Level Seven Level Eight Level Nine R2 S.S.E. n

Coefficient

0.674** 0.046** 0.0002 0.061* 0.092** 0.094** 0.053 0.055 –0.041 –0.161** –0.330** –0.493** 0.045 0.026 0.048 0.094 0.079 0.130** 0.049 –0.012 0.069 –0.093* 0.043 0.095** 0.149** 0.226** 0.224** 0.175**

Standard Error

0.030 0.013 0.000 0.026 0.018 0.036 0.031 0.032 0.034 0.031 0.038 0.039 0.034 0.056 0.048 0.048 0.048 0.044 0.058 0.058 0.046 0.038 0.027 0.022 0.023 0.027 0.032 0.019

0.1192 2052.2 7270

Note: (*) denotes significance at the 5% level, and (**) denotes significance at the 1% level. All cases used in this regression were settled for a positive amount. The dependent variable is the log of the time between the date of filing and the date of settlement, expressed in years.

235

The Duration of Personal Injury Litigation

Table 5. Effect of Variables on Time to Settlement: Estimates of Duration Models from Michigan Data Weibull Variable One (Intercept) Multiple Defendants Age Medicaid Health Insurance Filed 82 Filed 83 Filed 84 Filed 85 Filed 86 Filed 87 Filed 88 Wayne County Wayne County – 82 Wayne County – 83 Wayne County – 84 Wayne County – 85 Wayne County – 86 Wayne County – 87 Wayne County – 88 Level One Level Two Level Four Level Five Level Six Level Seven Level Eight Level Nine Scale Shape Log Likelihood

Lognormal

Log-logistic

Gamma

Coef·

Prob·

Coef·

Prob·

Coef·

Prob·

Coef·

Prob·

1.457 –0.168 0.0001 0.056 0.135 –0.152 –0.243 –0.211 –0.323 –0.486 –0.670 –0.830 –0.031 –0.043 0.085 0.036 0.062 0.143 0.092 –0.001 0.153 0.048 0.035 0.086 0.122 0.208 0.214 0.138

0.0001 0.0001 0.728 0.007 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.260 0.338 0.028 0.355 0.107 0.0001 0.047 0.991 0.0001 0.108 0.103 0.0001 0.0001 0.0001 0.0001 0.0001

0.904 –0.122 0.0003 0.056 0.124 0.055 0.003 0.039 –0.003 –0.161 –0.349 –0.470 0.027 0.013 0.082 0.118 0.057 0.122 0.063 –0.036 0.138 –0.037 0.038 0.112 0.165 0.255 0.200 0.196

0.0001 0.0001 0.319 0.035 0.0001 0.141 0.930 0.224 0.918 0.0001 0.0001 0.0001 0.445 0.822 0.101 0.016 0.242 0.007 0.288 0.534 0.002 0.323 0.169 0.0001 0.0001 0.0001 0.0001 0.0001

0.998 –0.105 0.0002 0.057 0.122 –0.030 –0.066 –0.011 –0.074 –0.223 –0.410 –0.545 –0.003 0.046 0.128 0.114 0.080 0.146 0.091 –0.017 0.129 –0.007 0.029 0.094 0.133 0.214 0.175 0.164

0.0001 0.0001 0.523 0.010 0.0001 0.397 0.027 0.728 0.017 0.0001 0.0001 0.0001 0.925 0.367 0.004 0.009 0.066 0.0004 0.078 0.730 0.001 0.827 0.225 0.0001 0.0001 0.0001 0.0001 0.0001

1.376 –0.161 0.0001 0.056 0.137 –0.132 –0.215 –0.178 –0.279 –0.441 –0.627 –0.780 –0.043 –0.013 0.111 0.069 0.083 0.161 0.110 0.014 0.151 0.037 0.034 0.089 0.126 0.214 0.200 0.146

0.0001 0.0001 0.699 0.009 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.127 0.779 0.006 0.081 0.037 0.0001 0.021 0.768 0.0001 0.228 0.128 0.0001 0.0001 0.0001 0.0001 0.0001

0.425

0.582

0.291

–7037

–7727

–7024

0.444 0.819

Note: the dependent variable is the log of the time between the date of filing and the date of settlement, expressed in years. There are 7270 uncensored observations, and 1985 observations censored from the right. A case is censored from the right if it either (1) was resolved when the mediation award was accepted by both parties, or (2) was litigated to verdict. The probability in the right-hand column of each specification represents the probability of obtaining the chi-square value for the sample under the null hypothesis.

236

STEPHEN J. SPURR

cases, the coefficient of MULTIPLE DEFENDANTS is positive; as expected, the time to settlement increases when more than one defendant contributes to the settlement payment. In the duration models, however, when litigated cases and cases resolved by mediation are treated as censored, the coefficient of MULTIPLE DEFENDANTS remains highly significant, but becomes negative. We infer from these results that cases with several defendants are more likely to be settled than litigated, compared to cases with a single defendant; thus the effect of this variable on duration is negative.20 Given that a case is settled, however, the time required to arrive at a settlement increases with the number of defendants because of the greater transaction costs of bringing about an agreement. Returning to the parameter estimates, time to settlement is greater for plaintiffs whose medical expenses were paid by health insurance, probably because persons with health insurance are generally employed on a full-time basis and have a relatively high income in comparison with other (uninsured) victims of medical malpractice. Damages for disability and lost income are greater for those with relatively high incomes, and the time to settlement tends to increase with the amount of damages. MEDICAID also has a positive effect, which is unexpected; however, its estimated coefficient is substantially less than that of HEALTH INSURANCE in the least squares estimates and in the duration models. AGE is not significant. The coefficients for the various years of filing are to be interpreted relative to the omitted years – those before 1982 and after 1988. 715 claims were filed prior to 1982 and 252 after 1988. The mean duration of claims for these omitted years ought to be somewhat less than that of other years, since claims filed after 1988 are not in the data set if they were not closed within two years. This feature of the data set explains why all coefficients for the years of filing are negative in all but one of the duration models (the lognormal). Moreover, the coefficients decline in consecutive years after 1982 indicating that, in general, the duration of claims declined during the period from 1982 through 1988. It is noteworthy that the decline from the previous year is greatest for claims filed in 1987 and 1988. This is consistent with the hypothesis that plaintiffs throughout the state had less incentive to delay litigation after 1986, because of the changes in the rules on pre-judgment interest. In Wayne County, of course, a program was adopted beginning in 1987 for the express purpose of reducing court delay. In the duration models the interaction terms for WAYNE COUNTY and the year of filing variables are positive whenever they are significant, indicating that the time to settlement was greater in Wayne County than elsewhere during the period from 1983 to 1987. The immediate impact of the delay reduction program is reflected in the

The Duration of Personal Injury Litigation

237

decline of the size of the interaction term from 1986 to 1987; the coefficient declines from 0.14 to 0.09 in the Weibull and log-logistic estimates, and from 0.16 to 0.11 in the Gamma estimates. The time to settlement is still greater in Wayne County than in other Michigan counties, but the differential has declined substantially. The effects of the delay reduction program were also reflected in the time to verdict for tried cases. Table 6 shows that there was a substantial decline in the time to verdict for cases tried in Wayne County during the two years after 1986, when the delay reduction program was in effect. Returning to Tables 4 and 5, the estimates of the effects of the severity of injury on duration are to be interpreted relative to the omitted category, severity of level three, a ‘temporary minor’ injury. As expected, there are significant positive coefficients for levels five through nine, with the largest coefficients being estimated for level seven (permanent major) and level eight (permanent grave). The more severe the injury is, the greater are the damages, and the longer it takes the parties to agree on a settlement payment.

8. ANALYSIS OF THE NEW YORK DATA To test our findings from the Michigan data and investigate the effect of other variables on duration, we decided to analyze a large data set of cases from New York City closed between 1974 and 1984.21 In all these cases an attorney was hired by a claimant on a contingent fee basis, and virtually all the cases were claims for personal injuries.22 The information in this data set was obtained from ‘retainer statements’, filed by a lawyer when each case was filed in court, and ‘closing statements’, filed when the case was resolved, whether by settlement, litigation to verdict, or abandonment by the plaintiff. The characteristics and history of this data set gave us great confidence in its accuracy and completeness.23 We used ordinary least squares to estimate the effect of covariates on the duration of settled cases. The dependent variable is the log of time from filing to settlement. The dummy variable ‘REFERRAL’ indicates whether the case was referred by one attorney to another, under an arrangement whereby the contingent legal fee is divided between the attorneys. Attorneys to whom cases are referred tend to be specialists in litigating cases of the type that are referred to them. They are well suited to determine the value of a claim and to obtain a recovery in an efficient manner. One would expect that referral of a claim to a specialist would shorten the time to settlement appreciably. The use of this variable is motivated by mounting evidence that the quality of a lawyer is a factor of crucial importance in every aspect of the market for legal services.

238

Table 6.

A Comparison of Time to Settlement with the Time to Verdict, for Different Courts in Michigan and New York

Court

Wayne County Circuit Court (Michigan): before 1987 1987 and later New York Supreme Court New York City or Municipal Court District Court of Nassau or Suffolk (Long Island) Federal District Court (New York)

Number of Cases Settled

2504 548 69346 12887 195 3465

Time to Settlement (in months) Mean

Standard Deviation

34.66 19.81 21.50 17.21 15.17 19.20

13.29 7.82 15.60 13.13 11.45 12.56

Number of Cases Litigated

387 82 1092 256 7 196

Time to Verdict (in months) Mean

Standard Deviation

45.46 20.67 25.59 20.47 17.18 25.87

19.20 9.09 18.20 14.83 16.73 16.18

STEPHEN J. SPURR

Note: The track system was introduced into the Wayne County Circuit Court on January 2, 1987, and phased in on a judge-by-judge basis. The first row of the table applies to cases filed in this court before 1987, whereas row 2 applies to cases filed thereafter.

The Duration of Personal Injury Litigation

239

Spurr (1987) found that differences in compensation, promotion and turnover across law firms, were substantially driven by differences in lawyer quality. Sloan & Hoerger (1991) found that plaintiffs with medical malpractice claims who were represented by specialist attorneys received about twice as much compensation for their economic loss as those represented by nonspecialists. Another variable, WALK-IN, indicates how the case came to the attorney. WALK-IN equals 1 if there is no indication that the lawyer was recommended to the client or was previously known to the client, and the case was not referred to the lawyer by another lawyer. Thus, the omitted category relative to the WALK-IN and REFERRAL variables consists of cases where the lawyer was either previously known to the client or recommended to the client. If we consider a case where the client ‘walks in’ as being like a random assignment of client to lawyer, we would expect these cases to be handled less efficiently than cases where the assignment is made on the basis of some market information, e.g. cases where the attorney was recommended to the client. On the other hand, a case might be coded as a ‘walk-in’ when the client was responding to the lawyer’s advertising. Information about the quality of lawyers conveyed by advertising might be as good as, or better than, information about lawyers obtained from friends or relatives. Accordingly, we have no particular a priori expectation about the effect of the WALK-IN variable. The literature on court administration has identified a number of factors affecting the pace of litigation which vary across courts. These factors include the pending caseload per judge; the composition of the caseload, e.g. the percentage of tort cases; the extent to which the court’s judges specialize in handling cases of a particular type; procedural rules applied in the court; and the type of case management procedures which the court has adopted, such as early court intervention in scheduling case events and the use of disposition time goals. (See generally Goerdt 1991.) Accordingly, we have included dummy variables for the nine different categories of New York courts represented in the sample. Results are set forth in Table 7. The time to settlement is greater for larger claims and much less for cases referred by one lawyer to another. At the sample mean, the fact that a case is referred would reduce its duration from 20.8 months to 11.6 months. Duration is also less for the cases of clients who walk in, although the magnitude of this effect is much less than that of a referral. Our conjecture is that this result reflects the fact that many of the clients classified as ‘walk-ins’ were actually directed to their lawyers by advertising. Thus, the lawyers who handled their cases were likely to be specialists in personal injury litigation. Finally, the results for the court variables show that it takes longer to

240

STEPHEN J. SPURR

Table 7. Effect of Various Factors on the Time to Settlement in New York Variable

Coefficient

Standard Error

Intercept Log of Settlement Payment, in real (1982–84) dollars Walk-in Referral Court outside the State of New York County Court New York City or Municipal Court Small Claims Court District Court of Nassau or Suffolk (Long Island) Surrogate Court New York City Court outside the five boroughs Federal District Court

–1.17** 0.13** –0.18** –0.44** 0.26** 0.21 0.14** –0.95 0.15 0.39 0.002 0.15**

0.033 0.004 0.015 0.010 0.074 0.157 0.013 0.599 0.096 0.292 0.335 0.023

R2

0.026

Note: (**) denotes significance at the 1% level. All cases used in this regression were settled for a positive amount. The dependent variable is the log of the time between the date of filing and the date of settlement, expressed in years. There are 113,965 observations. These cases were handled by lawyers in the First Judicial Department of New York (Manhattan and the Bronx) and were closed during the period from 1974 through 1984. The settlement payments are adjusted for inflation. We deflated by the Consumer Price Index, using 1982–84 as the base period.

settle cases in federal district court, city and municipal courts, and out of state courts than it does in New York Supreme Court.

9. SUMMARY AND CONCLUSION The court congestion hypothesis has made a valuable contribution to the literature on court administration by providing a model which can be used to evaluate proposals for reducing court delay. This theory implies that the decisions of litigants on a number of margins are affected by factors such as the expected time to trial, the rate of pre-judgment interest, and the expected costs of litigation. Analysis of both the Michigan and New York data showed rather conclusively that the time to settlement increases with the amount at stake. Analysis of the Michigan data with regression and duration models enabled us to determine the effects of a delay reduction program in Wayne County, under which there is early intervention by a judge to impose on the parties a schedule for case events. We found that the program was successful in reducing the time

The Duration of Personal Injury Litigation

241

to settlement and the time to verdict for tried cases, over the period covered by the data. Although the time to trial has been greatly reduced, there is no evidence of an increase in the trial rate; there is, however, strong evidence of an increase in the number of cases filed annually in the Wayne County Circuit Court. This increase is certainly consistent with the Posner-Priest hypothesis that a reduction in time to trial will attract more litigants to a court, and we do not know of a plausible alternative explanation. Finally, the decline in time to settlement after 1986 is consistent with the hypothesis that plaintiffs had less incentive after 1986 to delay litigation because of a reduction in the rate allowed for pre-judgment interest. Regression results based on a large data set of personal injury claims from New York showed that referral of the case to a specialist had a very large effect in reducing the time to settlement. This finding should be considered in conjunction with (A) the evidence that plaintiffs in medical malpractice cases who are represented by specialists recover twice as much, in relation to their economic loss, as those represented by nonspecialists (Sloan and Hoerger 1991), and (B) the argument that because of court delay, those most seriously injured by medical malpractice get the smallest settlements in relation to their economic loss (Weiler 1991). Our results provide additional support for the recommendation by Sloan and Hoerger for consideration of ways to improve the flow of information to personal injury victims about the quality of lawyers. Such measures might include the amendment of laws or of bar association rules that inhibit referrals between lawyers, and the abolition of any remaining restraints on lawyer advertising, or on the dissemination of information about lawyer quality.

NOTES 1. Priest (1989, pp. 546–547). 2. Between 1980 and 1990 the population of Michigan increased from 9,262,078 to 9,295,297, while the population of Wayne County declined from 2,337,891 to 2,111,687 (1990 Census of Population and Housing, Bureau of the Census, August 1992). Thus the share of the state population in Wayne County was 25.2% in 1980 and 22.7% in 1990. 3. Somerlot, Solomon & Mahoney (1989, p. 12). 4. The period of time between the date of filing and the judge’s initial intervention was changed during the program, but the status conference was to occur 91 days after filing under a court-wide plan adopted in October 1988. Somerlot, Solomon & Mahoney (1989, p. 15). 5. Gary S. Wolfe, Deputy Administrator for Trial Services, Third Judicial Circuit of Michigan. 6. Somerlot, Solomon & Mahoney (1989, p. 14).

242

STEPHEN J. SPURR

7. The Posner-Priest hypothesis bears more than a passing resemblance to the Lucas (1976) critique of interventionist macroeconomic policy. Both models are based on the idea that the behavioral responses of individuals can offset or nullify altogether the intended effects of government intervention. 8. The real interest rate is the nominal interest rate minus the rate of inflation. If in its assessment of damages the court takes into account inflation between the time of the injury and the time of verdict, interest on the damage award should accrue at the real rate of interest. 9. Between 1980 and 1990 the population of Wayne County declined from 2,337,891 to 2,111,687 (1990 Census of Population and Housing, Bureau of the Census, August 1992). 10. One might ask why the number of cases filed in the Wayne County Court continued to increase through 1993, given that the delay reduction program was fully implemented by the end of 1989. The reason would seem to be that the response of the legal community is not instantaneous, and that it takes time for information about a new program to be fully disseminated. Even if a lawyer knew that a delay reduction program had been adopted, he or she might wait to see through experience how effective the program was before making any decision to file claims based on the program. 11. In a regression of the log probability on the year and an intercept, the negative coefficient for year has a p-value of 0.0373. 12. M.C.L.A. 600.6013 (2) and (4). An additional complication was that under the new rules the rate of pre-judgment interest could be changed if either party made an adequate settlement offer which was rejected by the other party. These rules, which became effective in 1987, were designed to use the rate of pre-judgment interest as a means to encourage settlement rather than litigation. Under these rules a plaintiff would not be entitled to pre-judgment interest if he had previously rejected a ‘bona fide, reasonable’ offer of settlement from the defendant. Conversely, if the defendant had rejected a ‘bona fide, reasonable’ offer of settlement from the plaintiff, and judgment was subsequently entered for the plaintiff, the interest rate after the date of rejection would be 2% higher than the normal rate. The definition of a bona fide, reasonable offer was based on the outcome of the litigation. The interest penalty was imposed on the party who rejected the offer of settlement unless the judgment was more than 10% better for that party than the settlement offer would have been. M.C.L.A. Sec. 600.6013(7)–(13). 13. Baerwaldt (1985, p. 16). 14. C.P.L.R. Sec. 5001(a). However, in wrongful death actions interest accumulated at a rate of 6% from the date of the decedent’s death. C.P.L.R. Sec. 5004. N.Y. Laws 1962, Ch. 310, Sec. 112. 15. One might ask why we do not regard a case as ‘settled’ when the mediation award is accepted by both parties. We take this position for two reasons: (1) the mediation procedure in Michigan differs from settlement negotiations in that there is an element of coercion. Under certain conditions a party who does not accept the mediation award will be liable for the post-mediation legal expenses of the other party. (2) Again, unlike settlement negotiations, each party decides whether to accept the mediation award without knowing the response of the other party. Parenthetically, only about 9% of filed medical malpractice claims were resolved by acceptance of the mediation award during the period of our data.

The Duration of Personal Injury Litigation

243

16. Although duration analysis has been used to study the pace of litigation (see Joel B. Grossman et al. (1981) and the articles by Grossman & Herbert Kritzer in Adler et al. (1982)), we have not found any previous work estimating the effect of covariates. 17. For these three levels of injury severity, the number of paid claims, the mean recovery and the standard deviation are as follows: level seven (permanent major): 826, $284,305, and $491,960; level eight (permanent grave): 431, $614,243, and $1,211,649; level nine (death): 2800, $148,561, and $204,311. Recoveries are expressed in real (1982–84) dollars. 18. With another data set, he investigated whether the time from filing to trial increased with the number of defendants involved, but did not find a significant effect. 19. See table 6 of this paper; Mahoney, Sipes & Ito (1985, pp. 7–9); Goerdt et al. (1989); and Somerlot et al. (1989). 20. There is theoretical support for the idea that settlement is more likely when there are multiple defendants. Easterbrook, Landes & Posner (1980) have considered the question whether defendants will settle or litigate under different rules of contribution among joint tortfeasors. They found that defendants have a maximum incentive to settle when there is a rule of no contribution. In this case there is a competition among defendants to settle, since no defendant will want to be in a position to be sued after other defendants have settled with the plaintiff. The law in effect in Michigan was the common law rule that there is no contribution among defendants jointly and severally liable for medical malpractice. Thus, it seems reasonable that a settlement was more likely to occur in cases where more than one defendant was liable. 21. These cases were all filed in the First Judicial Department of New York (Manhattan and the Bronx) during the period from 1974 through 1984. 22. 99.7% of the 113,912 cases were actions for personal injuries or wrongful death. 23. A detailed account of the origins of this data set is provided in Rosenberg & Sovern (1959).

ACKNOWLEDGMENTS I am grateful to The Richard J. Barber Foundation for support of this Chapter’s research.

REFERENCES Abel, A. B., & Bernanke, B. S. (1995). Macroeconomics (2nd ed.). Reading, Mass.: AddisonWesley. Adler, J. W., Felstiner, W. F., Hensler, D. R., & Peterson, M. A. (1982). The Pace of Litigation: Conference Proceedings. Santa Monica, Calif.: Rand Corporation Institute for Civil Justice. Baerwaldt, N. A. (1985). Medical Malpractice Issues. Michigan Insurance Bureau, Department of Licensing and Regulation. Boden, L. I. (1990). Dispute Resolution in Workers’ Compensation. Working Paper, Boston University School of Public Health. Bovbjerg, R. R. (1989). Legislation on Medical Malpractice: Further Developments and a Preliminary Report Card. University of California, Davis Law Review, 22, 499–556.

244

STEPHEN J. SPURR

Church, T. W. Jr., Carlson, A., Lee, J.-L., & Tan, T. (1978). Justice Delayed: The Pace of Litigation in Urban Trial Courts. Williamsburg, Va.: National Center for State Courts. Cooter, R. D., & Rubinfeld, D. L. (1989). Economic Analysis of Legal Disputes and Their Resolution. Journal of Economic Literature, 27, 1067–1097. Coyte, P. C., Dewees, D. N., & Trebilcock, M. J. (1991). Canadian Medical Malpractice Liability: An Empirical Analysis of Recent Trends. Journal of Health Economics, 10, 143–168. Coyte, P. C., Dewees, D. N., & Trebilcock, M. J. (1991). Medical Malpractice: the Canadian Experience. The New England Journal of Medicine, 324, 89–93. Danzon, P. M. (1983). Contingent Fees for Personal Injury Litigation. Bell Journal of Economics, 14, 213. Danzon, P. M. (1984). The Frequency and Severity of Medical Malpractice Claims. Journal of Law and Economics, 27, 115. Danzon, P. M. (1986). The Frequency and Severity of Medical Malpractice Claims: New Danzon, P. M. (1985). Medical Malpractice. Cambridge: Harvard University Press. Danzon, P. M. and Lillard, L. A. (1982). The Resolution of Medical Malpractice Claims: Modeling the Bargaining Process – Research Results and Policy Implications. Report No. R2792-ICJ. South Monica, Calif.: Rand Corporation Institute for Civil Justice. Danzon, P. M., & Lillard, L. A. (1983). Settlement Out of Court: The Disposition of Medical Malpractice Claims. Journal of Legal Studies, 12, 345. Evidence. Law and Contemporary Problems, 49, 57–84. Dewees, D. N., Trebilcock, M. J., & Coyte, P. C. (1991). The Medical Malpractice Crisis: A Comparative Empirical Perspective. Law and Contemporary Problems, 54, 217–251. Dungworth, T., & Pace, N. M. (1982). Statistical Overview of Civil Litigation in the United States. Santa Monica, Calif.: Rand Corporation Institute for Civil Justice. Easterbrook, F. H., Landes, W. H., & Posner, R. A. (1980). Contribution among Antitrust Defendants: A Legal and Economic Analysis. Journal of Law and Economics, 23(2), 331. Farber, H. S., & White, M. J. (1991). Medical Malpractice: An Empirical Examination of the Litigation Process. Rand Journal of Economics, 22, 199–217. Fleming, R. W. (1985). Final Report to Gov. James J. Blanchard on the Subject of Health Care Provider Malpractice and Malpractice Insurance. Unpublished. Fleming, R. W. (1965). The Labor Arbitration Process. Urbana, Illinois: University of Illinois. Fournier, G. M., & Zuehlke, T. M. (1989). Litigation and Settlement: An Empirical Approach. Review of Economics and Statistics, 71(2), 189–195. Garves, B. R. (1988). Michigan Medical Malpractice Law Update. Michigan Bar Journal, 597–600. General Accounting Office (1990). Medical Malpractice: Few Claims Resolved through Michigan’s Voluntary Arbitration Program. GAO/HRD–91–38, B–239639. Goerdt, J. A. (1991). Explaining the Pace of Civil Case Litigation: The Latest Evidence from 37 Large Urban Trial Courts. Justice System Journal, 14, 289–342. Goerdt, J. A., Lomvardias, C., Gallas, G., & Mahoney, B. (1989). Examining Court Delay: The Pace of Litigation in 26 Urban Trial Courts. Williamsburg, Va.: National Center for State Courts. Goerdt, J. A., & Lomvardias, C. (1991). Re-examining the Pace of Litigation in 39 Urban Trial Courts. Williamsburg, Va.: National Center for State Courts. Gould, J. P. (1973). The Economics of Legal Conflicts. Journal of Legal Studies, 2, 279. Greene, W. H. (1991). Limdep 6.0 – User’s Manual and Reference Guide. Bellport, New York: Econometric Software Inc.

The Duration of Personal Injury Litigation

245

Grossman, J. B., Kritzer, H. M., Bumiller, K., & McDougal, S. (1981). Measuring the Pace of Litigation in Federal and State Trial Courts. Judicature, 65, 86. Harvard Medical Practice Study (1990). Patients, Doctors, and Lawyers: Medical Injury, Malpractice Litigation, and Patient Compensation in New York. Report of The Harvard Medical practice Study to The State of New York. Cambridge, Mass.: President and Fellows of Harvard College. Hughes, J. W. (1989). The Effect of Medical Malpractice Reform Laws on Claim Disposition. International Review of Law and Economics, 9, 57–58. Landes, W. M. (1971). An Economic Analysis of the Courts. Journal of Law and Economics, 14, 61. Hughes, J. W., &Snyder, E. A. (1989). Policy Analysis of Medical Malpractice Reforms: What Can We Learn from Claims Data. Journal of Business and Economic Statistics, 7, 423–431. Lucas, R. E. Jr. (1976). Econometric Policy Evaluation: A critique. In: K. Brunner, & A. Meltzer (Eds), The Phillips Curve and the Labor Market Vol. 1 of Carnegie-Rochester Conferences in Public Policy, a supplementary series to the Journal of Monetary Economics. Amsterdam: North-Holland. Mahoney, B., Aikman, A., Casey, P., Flango, G., Gallas, G., Henderson, T., Ito, J., Steelman., & Weller, S. (1988). Changing Times in Urban Courts: Caseflow Management and Delay Reduction in Urban Trial Courts. Williamsburg, Va.: National Center for State Courts. Metzloff, T. B. (1991). Resolving Malpractice Disputes: Imaging the Jury’s Shadow. Law and Contemporary Problems, 54, 43–115. Michigan Department of Licensing and Regulation (1986). A Review of the Bureau of Health Service’s Response to the Medical Malpractice Problem. Michigan Bureau of Health Services. Miller, G. P. (1989). Some Thoughts on the Equilibrium Hypothesis. Boston University Law Review, 69, 561–568. Payne, B. C., Van Harrison, R., & Harel, Y. (1985). Malpractice Trends in Michigan: Commercially Insured Claims and Awards, 1978–1984. University of Michigan Institute for Social Research. Posner, J. R. (1986). Trends in Medical Malpractice Insurance. Law and Contemporary Problems, 49, 37–56. Posner, R. A. (1972). An Economic Approach to Legal Procedure and Judicial Administration. Journal of Legal Studies, 2, 399. Posner, R. A. (1986). Economic Analysis of Law (3rd ed.). Boston: Little, Brown and Company Powsner, R. M., & Hamermesh, F. (1987). Medical Malpractice Crisis – the Second Time Around. Journal of Legal Medicine, 8, 283–304. Priest, G. L. (1989). Private Litigants and the Court Congestion Problem. Boston University Law Review,69, 527–559. Priest, G. L. (1985). Reexamining the Selection Hypothesis: Learning from Wittman’s Mistakes. Journal of Legal Studies, 14, 215. Priest, G. L., & Klein, B. (1984). The Selection of Disputes for Litigation. Journal of Legal Studies, 13, 1–55. Rosenberg, M., & Sovern, M. I. (1959). Delay and the Dynamics of Personal Injury Litigation. Columbia Law Review, 59 1115. Schoonmaker, J. R. (1977). The Medical Malpractice Arbitration Program in Michigan. Insurance Law Journal, 370–375. Shanley, M., & Peterson, M. (1983). Comparative Justice: Civil Jury Verdicts in San Francisco and Cook Counties. Rand Corporation, Report No. R–3006-ICJ.

246

STEPHEN J. SPURR

Shelton, D. E., Bishop, L. R., & Blaske, T. H. (1987). 1986 Tort Reform Legislation. Michigan Bar Journal, 252–260. Shepherd, G. B. (1995). The Economics of Pretrial Discovery: An Empirical Study. Working paper, Emory University School of Law. Shuart, K. L., Smith, S., & Planet, M. D. (1983). Settling Cases in Detroit: An Examination of Wayne County’s ‘Mediation’ Program. Justice System Journal, 8(3), 307–324. Sloan, F. A. (1990). Experience Rating: Does it Make Sense for Medical Malpractice Insurance? AEA Papers and Proceedings, 80, 128–133. Sloan, F. A. (1985). State Response to the Malpractice Insurance ‘Crisis’ of the 1970’s: An Empirical Assessment. Journal of Health Politics, Policy and Law, 9, 629–647. Sloan, F. A., & Bovbjerg, R. R. (1989). Medical Malpractice: Crises, Responses, and Effects. Health Insurance Association American Research Bulletin. Sloan, F. A., Mergenhagen, P. M., & Bovbjerg, R. R. (1989). Effects of Tort Reforms on the Value of Closed Medical Malpractice Claims: A Microanalysis. Journal of Health Politics, Policy and Law, 14, 663–689. Snedecor, G. W., & Cochran, W. G. (1967). Statistical Methods (6th ed.). Ames, Iowa: Iowa State University Press. Somerlot, D. K., Solomon, M., & Mahoney, B. (1989). Straightening out Delay in Civil Litigation. The Judge’s Journal, 28(4), 10–48. Spurr, S. J. (1987). How the Market Solves an Assignment Problem: The Matching of Lawyers with Legal Claims. Journal of Labor Economics, 5(4), 502–532. U.S. General Accounting Office (1987). Medical Malpractice: Characteristics of Claims Closed in 1984. Washington, D.C.: U.S. Government Printing Office. Weiler, P. C. A Measure of Malpractice. Cambridge, Mass.: Harvard University Press. Forthcoming. Weiler, P. C. (1991). Medical Malpractice on Trial. Cambridge, Mass.: Harvard University Press. Zeisel, H., Kalven, H., & Buchholz, B. (1959). Delay in the Court. Boston: Little, Brown and Company.

THE EFFICIENCY OF MEDICAL MALPRACTICE LAW: A NEW APPRAISAL Reed Neil Olsen ABSTRACT This paper analyzes the perception by researchers, public policy makers, and physicians that the medical malpractice system is in disarray and in need of reform. The perception of a medical malpractice crisis arose because of what was viewed as sudden and dramatic increases in physician liability for malpractice. Contrary to the common perception, however, previous research has shown that historical growth rates in physician liability are similar in magnitude to current growth rates. This paper focuses on explaining the conditions under which increased physician liability would be optimal. According to the theoretical model, increased physician liability would be optimal when (1) physicians become more adept at curing patients, especially by increased technological ability, (2) the costs of physicians’ time increases, or (3) the cost to physicians of defending against malpractice claims decreases. The paper carefully examines the available historical evidence that indicates that these three reasons account for much of the increased liability of physicians in the United States. The finding that much of the historical An earlier version of this paper was presented at the Western Economic Association’s annual meetings in Lake Tahoe in July of 1993. Research support was granted by the Graduate Studies and Research Office, Southwest Missouri State University. Research in Law and Economics, Volume 19, pages 247–273. Copyright © 2000 by JAI/Elsevier Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0308-5

247

248

REED NEIL OLSEN

increases in physician liability are consistent with the model, further questions the existence of a medical malpractice crisis.

I. INTRODUCTION The occurrence of medical malpractice claims before the 1960s has been perceived to be quite rare and to have had an insignificant impact on the practice of medicine (Danzon, 1985, 1988; Sloan et al., 1991; Weiler, 1992). However, beginning in the 1960s the frequency with which patients filed malpractice suits, the proportion of suits won by patients, and the average size of the dollar award for paid malpractice claims all began to rise at what was thought to be unprecedented rates (Danzon, 1985; Olsen, 1996; Peterson, 1987; Weiler, 1991). As a result of these changes, physician liability for malpractice rose dramatically. For example, by 1984 medical malpractice premiums, which had averaged less than 1% of a physician’s total costs in the 1950s, averaged 9% of a physician’s total costs (Sloan et al., 1991, GAO, 1986b). These increases in physician liability led researchers, public policy makers, and the medical profession to conclude that the medical malpractice system was in disarray and in need of reform.1 In fact, numerous medical malpractice reforms have been proposed in the past three decades and a significant number of reforms have been passed by state legislatures.2 By 1986, for example, all states had enacted at least one major medical malpractice reform and had, on average, enacted four out of ten major reforms (Olsen, 1996). Currently, medical malpractice reform is being debated at the federal level (White House, 1993). Recent research has demonstrated, though, that the impetus behind the reforms, the perception of a crisis in the malpractice system, may be incorrect. Although it appears clear that physician liability has reached an unprecedented high over the past three decades, recent growth rates in malpractice claim frequency and the average size of paid claims are consistent with the history of medical malpractice in the United States (Olsen, 1996). Given that neither malpractice litigation nor large growth rates in physician liability are recent phenomena, the explanation for both recent and past increases becomes somewhat puzzling. It seems implausible that the medical malpractice system could have been dysfunctional throughout most of United States history. This paper develops an alternative scenario to the prevailing view of a malpractice system in crisis. Rather, the theoretical model developed below focuses on predicting the conditions under which increased physician liability would be optimal. Assuming that enhanced liability deters medical accidents, the model demonstrates that physician liability should increase, for example,

The Efficiency of Medical Malpractice Law: A New Appraisal

249

when physicians become more technologically adept at treating patients. It is also optimal for physician liability to increase when physicians’ time becomes more valuable or when their cost of defending against a claim of malpractice falls. The paper also examines available historical evidence and finds that both the existence of physician liability for malpractice and its growth historically has been generally consistent with the theoretical predictions. For example, prior to 1835 the few malpractice cases in the United States were filed for those procedures, mostly amputations of dislocated and fractured limbs, for which physicians had some technological ability to affect cures (DeVille, 1990). Likewise, the historical evidence clearly indicates that as physicians have become more technologically adept at affecting cures in the past 150 years, their liability, as predicted, has also increased. For example, even as physicians became more adept at treating fractures and dislocations with new technological innovations, such as x-ray technology and plaster-of-paris dressings, their liability for adverse outcomes also increased. Similarly, as surgery and obstetric treatments advanced technologically, so did physician liability in these areas.

II. A THEORETICAL MODEL OF MEDICAL MALPRACTICE Consider the exchange that occurs in a health services market. Inherent in the consumption of a physician’s services is the risk that an accident will occur. However, both patient and physician can, by using resources, reduce the probability of an accident occurring. Patients can, for example, take precautions against adverse outcomes by exercising, following nutritional guidelines, and by following the physician’s directions. Physicians have two major methods by which they can reduce the probability of an accident. For example, they can simply take extra precaution when diagnosing and treating an illness. However, physicians can also avoid adverse outcomes by learning about and using technological advances in diagnosis and treatment. Note that these two methods are related. Taking extra precaution reduces the probability of an accident directly through the use of more resources, such as the physician’s time, while improved technology tends to increase the marginal returns to the physician’s time. Both failure to use improved technology and failure to use it properly may constitute grounds for malpractice. Formally, assume that the probability of an accident occurring is given by P = p(x, y) where x and y are the physician’s and patient’s precaution levels.

250

REED NEIL OLSEN

The probability function is assumed to have the normal properties; increased precaution by both parties reduces the probability of an accident occurring but at a decreasing rate (px, py < 0; pxx, pyy > 0). For simplicity, I assume that the two inputs are neither substitutes nor complements (pxy = 0).3 Medical malpractice is modelled under the assumption that the law chooses a particular share rule and, once an accident has occurred, incurs the costs necessary to enforce the division of accident losses consistent with the rule. This assumption both simplifies the analysis of medical malpractice and, as shown in the following section, yields numerous testable hypotheses. Brown (1973) made a similar assumption in his treatment of comparative negligence rules. Haddock and Curran (1985) correctly point out that this treatment is incorrect when courts have complete information about actual precaution levels ex post. In fact, when courts have complete and costless information, all negligence liability rules result in optimal precaution being taken. However, I am considering the more interesting case here of costly and, hence, incomplete information by the courts. In this more realistic scenario, courts may simply allocate liability by a share rule as I assume. In fact, White (1989) presents evidence that courts and juries do share accident losses in this manner. For example, under negligence rules a non-negligent defendant should technically not bear any of the losses. However, White finds that such defendants actually bear some, but not all of the losses. Likewise when the defendant is negligent and the plaintiff is non-negligent, all negligence rules would have the defendant bearing the entire liability yet White finds that such defendants only pay a fraction of actual accident losses. Medical malpractice rules must be enforced in the event that an accident occurs. The enforcement of the tort liability rule is, of course, costly. I assume that there are three sources of such enforcement costs: the enforcement costs borne by the patient (Ep), the enforcement costs borne by the doctor (Ed), and the enforcement costs borne by the court (Ec).4 Under the exogenous tort liability rule, t, expected values to physician and patient are given by: Vd = f ⫺ (1 ⫺ t)p(x, y)A ⫺ p(x, y)Ed ⫺ wxx ⫺ c

(1)

Vp = V ⫺ f ⫺ tp(x, y)A ⫺ p(x, y)Ep ⫺ wyy

(2) 5

where V is the monetary value to the patient of the physician’s services, f is the fee paid by the patient to the physician (determined by the market), t is the share of expected accident losses borne by the patient under medical malpractice law, A represents accident losses, wy and wx are per-unit input costs for y and x respectively, and c is the marginal cost of producing the physician’s services.

The Efficiency of Medical Malpractice Law: A New Appraisal

251

Maximization by each party of their respective expected values will result in input levels xd and yp and satisfy the following first-order conditions: ⫺ [(1 ⫺ t)A + Ed]px(x, y) = wx

(3a)

⫺ [tA + Ep]py(x, y) = wy

(3b)

Both parties have an incentive to shirk on their use of inputs relative to the optimal solution. The societal wealth maximization problem is to: max Nt = V ⫺ p(x, y)[A + E] ⫺ c ⫺ wxx ⫺ wyy x, y

(4)

where E represents total enforcement costs (E = Ep + Ed + Ec). The optimal input levels (x* and y*) satisfy: ⫺ [A + E]px(x*, y*) = wx

(5a)

⫺ [A + E]py(x*, y*) = wy

(5b)

Comparison of equations 5a and 5b to equations 3a and 3b demonstrate that precaution is undersupplied by both parties. Shirking occurs because each party bears only a fraction of societal losses if an accident occurs. There are two components of societal losses both of which contribute to the shirking problem. First, societal losses include the actual accident losses, A, which must be borne when an accident occurs. The malpractice share rule, t, causes each party to bear only a fraction of the expected accident losses which induces shirking by both parties. Second, societal losses include total enforcement costs, E, which must also be borne when an accident occurs. Additional shirking occurs because each party pays only its own, rather than total, enforcement costs. In accordance with the efficiency theory of the common law it is assumed that medical malpractice rules are chosen to maximize societal wealth.6 Of course, it would be much too costly for the courts to establish a completely different malpractice rule for each different physician-patient relationship. As a result, a single malpractice rule is chosen for a large number of relationships. However, the costs of not being able to set a single rule for each relationship are minimized by setting different rules for different homogeneous classes of relationships. Obviously the medical malpractice rule for a given class of relationships has a large impact on the parties’ input choices and, hence, on societal wealth. As a result, the share rule (t) must be chosen optimally to maximize wealth subject to the constraints that each party will shirk on their own input use. The societal wealth maximization problem is to: Max N = V ⫺ p(xd, yp)[A + E] ⫺ c ⫺ wxxd ⫺ wyyp t

(6)

252

REED NEIL OLSEN

The optimal rule, t*, satisfies the following first-order condition: ⫺ (px[A + E] + wx)xt = (py[A + E] + wy)yt

(7)

Equation 7 demonstrates that the optimal share rule, t*, must be set such that the marginal distortion in the patient’s precaution exactly cancels the marginal distortion in the physician’s precaution. Three factors affect the optimal share rule: enforcement costs (Ed and Ep), per-unit input or precaution prices (wx and wy), and each party’s marginal productivity (–px and –py). For example, increases in the patients’ enforcement costs will decrease their share of accident costs (t*). As patients’ enforcement costs rise, their willingness to supply precaution rises since they bear a larger portion of total enforcement costs. In order to also reduce shirking by physicians, their share of expected accident costs must also increase. Likewise, increases in the physicians’ enforcement costs will decrease their share of accident costs (1 ⫺ t*). A party whose precaution costs are higher has a larger incentive to shirk on the use of the more costly precaution. Thus, as a party’s price of precaution (wx or wy) rises its share of expected accident losses will also rise. Finally, either party’s shirking becomes more costly as that party’s precaution is more productive in reducing the probability of accidents. Thus, for example, as the physician’s productivity increases relative to the patient’s, the physician’s liability (1 ⫺ t*) is also expected to increase. As noted above, one method by which a physician becomes more productive on the margin is by taking advantage of improved technology. Table 1 summarizes the predicted impact of enforcement costs, precaution prices, and productivity on the medical malpractice rule (t*). The model predicts that these factors should affect medical malpractice rules in two separate and distinct manners. First, as these factors change over time the Table 1. Summary of Theoretical Predictions

Factor Physician’s Enforcement Costs (Ed) Patient’s Enforcement Costs (Ep) Physician’s Precaution Price (wx) Patient’s Precaution Price (wy) Physician’s Marginal Productivity (–px) Patient’s Marginal Productivity (–py)

Expected Impact on t* + + +

The Efficiency of Medical Malpractice Law: A New Appraisal

253

malpractice rule is expected to change consistent with these predictions as well. Second, the rule in effect is expected to be different for different types of relationships. For example, it would be expected that the malpractice liability of specialists, who are presumably more productive in avoiding accident loss, would be higher than that of general practitioners. The model can also be applied to the malpractice liability of hospitals, nurses, and other health care providers and professionals. As enforcement costs, precaution prices, and productivity vary for these different types of relationships it is expected that malpractice liability will vary as well.

III. EMPIRICAL APPLICATIONS OF THE MODEL The hypothesis that the legal rules governing medical malpractice in the United States respond to relative changes in productivity, precaution costs, and enforcement costs has never been empirically tested. Inadequate data is responsible for much of the lack of empirical testing; this study faces the same problem. Inadequate data, however, is not the full story. Part of the deficiency in empirical testing may arise from an inability or unwillingness to carefully consider the full implications of the theoretical tort model as applied to medical malpractice. A careful examination of the available data reveals that medical malpractice rules respond as the model predicts. Productivity Defined The first step in understanding the empirical impact of physician productivity is to understand what constitutes increased physician productivity. As noted in the theoretical section, physician productivity measures the marginal impact of the physician’s precaution (x) on the probability of a medical accident occurring (P) and is given by –px. Although this definition of physician productivity seems straightforward, the point of this section is to examine the empirical impact of increased physician productivity. Hence, it is necessary to first carefully define which empirical circumstances will likely lead to increased physician productivity. Given that increased physician productivity occurs, for a given amount of precaution, when the probability of a medical accident is decreased, the first step must be to examine what constitutes a medical accident. The most common view of an accident is one where the injurer, in this case the physician, does something which directly causes harm (e.g., operates on the wrong limb). However, medical malpractice includes a second type of medical accident in addition to this straightforward type of accident. For virtually all cases of

254

REED NEIL OLSEN

treatment, the doctor has not only a duty to avoid directly causing harm, but an affirmative duty to take actions to reduce the harm that will naturally occur, from disease or injury, in the absence of such treatment. The affirmative duty that a physician has to avoid causing harm has been enforced by holding physicians to a standard of care that is often, but not always, given by the prevailing practice of the profession. Physicians whose care fell below that standard would be found liable for accident losses.7 In other words, physicians can cause harm through incorrect treatment, given current technology and the current state of medical knowledge, but also through the absence of treatment. Thus, the failure to provide treatment that meets the current standards of the medical specialty constitutes negligence under medical malpractice law and, hence, is considered a ‘medical accident’ in terms of the theoretical model. This view of medical accidents implies that anything that increases the physician’s marginal productivity of treating patients’ illnesses or injuries also increases the physician’s marginal productivity of avoiding medical accidents. In brief, anything that increases the ability for the physician to do harm to the patient, either directly or indirectly, also increases the marginal productivity of the precaution the physician takes to avoid such harm. To illustrate this principle, consider physician productivity when physicians’ actions cannot cause much harm, either directly or indirectly. In this case, the productivity of taking care to be attentive while performing a medical procedure, for example, or taking the time to make a careful diagnosis is low simply because doing so has little ability to reduce the probability of harm.8 However, when physician ability to cause harm rises, then the productivity of paying attention during a procedure or taking care to make a careful diagnosis clearly rises, as harm can now be avoided by taking such care. It is relevant to ask, then, what would increase physician productivity in treating diseases and injuries (i.e. increase the ability of the physician to cause harm). Obviously, both improved medical technology and expanded medical knowledge would do so and, hence, would increase the physician’s marginal productivity of avoiding harm. In essence, increases in either technology or medical knowledge will increase the ability of physicians to cause harm through, for example, machines which have increased treatment potential but which, if not correctly used, can directly cause harm. Likewise, increased ability to cause harm can be caused by not using the treatment at all.9 The Impact of Productivity The evidence that the relative productivity of physicians impacts medical malpractice as predicted by the model is compelling. As noted above,

The Efficiency of Medical Malpractice Law: A New Appraisal

255

improvements in medical technology increase the relative productivity of physicians and are an important factor in explaining increased physician liability. In fact, the impact of increased technology may well explain much of the increased liability physicians have incurred over the history of medical malpractice in the United States.10 At the very least, it appears that the impact of productivity, especially that of improved technology, has had a stronger impact on malpractice rules than have changes in either enforcement or precaution costs.11 For example, prior to 1835 very few medical malpractice cases were filed or tried (DeVille, 1990, Olsen, 1996). Furthermore, the cases that did exist generally had two characteristics in common. First, most cases prior to 1835 were brought only for the most severe and disabling injuries including, but not limited to, injuries causing death. Second, the cases most often involved medical treatment that exhibited some proficiency in affecting cures. According to DeVille (1990), the three treatments with the most advanced technical expertise – vaccination, obstetrics, and amputations – also generated the majority of the scattered malpractice cases prior to 1835.12 Thus, even before 1835, the productivity of the physician in avoiding adverse outcomes was of paramount importance. After 1835 cases began to be brought for less severe injuries, as well as for a wider range of treatments. Seventeen different types of malpractice cases were found in a sample of cases gathered from 1835 to 1865.13 This widening appears to have also been driven by increases in physician productivity. To illustrate the importance of physician productivity consider the most common source of malpractice cases in the years after 1835: deformities caused by the treatment of fractures and dislocations.14 Before 1835, the threat of infection and eventual death was so large that, although some compound fracture cases existed where physicians had saved limbs, the common prescription was amputation.15 In subsequent years, however, new techniques were developed by physicians to treat fractures and dislocations. For example, physicians began to remove jagged bone ends with saws and rough the exposed bone ends to increase the probability of the bone healing into a strong union. Physicians also devised methods to combat postoperative infections which often followed compound fractures (DeVille, 1990). Treatment practices for fractures and dislocations continued to experience technological innovations. New technology included the introduction of plaster-of-paris dressings in the 1870s, aseptic surgery techniques in the 1890s, and x-ray technology in 1895. With each technological advance, physician liability also increased and malpractice cases were filed at increasing rates.

256

REED NEIL OLSEN

From 1946 to 1956, 25% of all appellate malpractice cases resulted from the treatment of fractures and dislocations (Sandor, 1957). The importance of the new technology in generating suits is especially illustrated by Sandor’s data. At least 38% of the fracture and dislocation cases resulted from the new innovations; 22% from alleged improper use of x-ray technology, 14% from alleged improper casting. The treatment of fractures and dislocations is not an isolated instance of increased physician liability following on the heels of enhanced physician productivity. In fact, ‘[t]echnological advancement and the accompanying expectations of complete proficiency have always helped fuel significant increases in malpractice rates’.16 An additional example of this general trend is the increased frequency of surgery as a source of malpractice cases during the early to mid-1900s. Surgery cases overtook fracture and dislocation cases as the largest single source of malpractice cases in the mid-1940s. From 1946 to 1956, 25% of all appellate malpractice cases were fractures and dislocations while 33% were surgery cases (Sandor, 1957). In the late-1800s, however, major surgery was rarely practiced. Rather, the practice of surgery was very limited in scope and focused mostly on external diseases and accidents (English, 1980; DeVille, 1990). In surgery, as with fractures and dislocations, fear of infection often kept surgeons from performing operations. Besides adverse outcomes caused by infection, an incomplete understanding of hemorrhage and shock prevented most body cavity surgery during the 1800s. However, during the late-1800s and continuing into the twentieth century, innovations allowed surgeons to become more productive. The antiseptic surgery ideas of Joseph Lister, originally developed during the 1860s, eventually found wide acceptance by surgeons during the late-1800s.17 More effective antiseptic surgery techniques and other innovations in surgical techniques were developed at this time and helped to increase the effectiveness of surgeons.18 By the 1940s, additional innovations had reduced the problems of shock and infection even further. However, each increase in surgical expertise resulted in additional malpractice cases. By 1984 only 10.7% of all physicians were surgeons but they generated 26.3% of all malpractice cases (GAO, 1987a). As a general rule, it appears clear that increased productivity by physicians tends to generate more malpractice cases and higher liability levels for physicians. Fracture and dislocation treatment, and surgery are two examples of the relationship between increased physician productivity and physician malpractice liability. Further, these are not isolated instances. The 1984 GAO (1987a) survey of closed malpractice claims, for example, finds that

The Efficiency of Medical Malpractice Law: A New Appraisal

257

malpractice claims were filed against 43 different types of specialists, although only 17 specialties generated 86% of all claims. Most of these specialties did not exist 40 years ago and treatments in all have benefitted from technological advances. Reproductive impairment serves as an additional example of how recent technological advances have fueled malpractice litigation. Reproductive impairment refers to medical malpractice cases arising because of the increased technological ability of the medical profession to track the effects of medical care given to a patient upon that patient’s unborn child. Werthmann (1984) notes four different types of reproductive impairment cases. In preconception negligence cases, a physician’s negligence is alleged to have caused the longterm damage of the plaintiff’s reproductive capacity. This damage is the subsequent cause of damage to a child conceived after the alleged negligence took place. A wrongful conception case is filed when it is alleged that a negligent sterilization subsequently allowed conception to occur. In a wrongful birth case, the parents of a genetically impaired child allege that with proper (non-negligent) genetic counselling they would have terminated the pregnancy. In a wrongful life case, damages are sought to support the genetically impaired child of a wrongful birth case.19 Although the evidence is convincing, the brief review of the relationship between physician productivity and liability provided in this paper is nevertheless incomplete. A complete review would require an analysis of all of the technological advances in medicine which occurred during the last two centuries. Since the advancement of medicine has been so large over this time period, such a review is beyond the scope of this paper. However, other researchers have also noted the direct and positive relationship between physician productivity and their legal liability.20 In fact, physicians themselves are well aware of the relationship between productivity and malpractice suits. For example, Frank Edwards, M.D. (1989, p. 154) notes that: “[e]very new medication, surgical technique, or diagnostic machine multiplies the probability for an error significant to the life of a given patient. The more conditions we can treat, the more opportunity for mishaps to occur.” The Impact of Precaution Costs As noted previously, both the frequency and the severity (i.e. average size of awards) of medical malpractice cases has risen dramatically during the history of the United States. Part of the increase can be attributed to increased physician productivity caused by technological innovations in medical practice.

258

REED NEIL OLSEN

Technological innovation, however, is not the only possible explanation for these increases. For example, consider the expected impact of precaution costs on physician liability. As physicians’ per-unit cost of precaution rises, physicians will tend to shirk more in an attempt to reduce their overall higher precaution costs. To reduce the increased incentive to shirk on their use of precaution, physician liability must also be increased. Thus, increases in precaution costs may also explain increased physician liability in the United States.21 To understand this argument one must carefully distinguish between the overall costs of precaution and the per-unit cost of precaution. For example, an increase in our medical knowledge base might allow physicians to spend less overall time avoiding harm per treatment. As a result, the overall precaution costs of treating a given patient would fall, which might suggest that liability should efficiently decrease as well. However, it is not a decrease in overall precaution costs which, according to the theory, would lead to an efficient decrease in physician liability. Rather it is a decrease in the per-unit costs of precaution which is predicted to lead to decreased physician liability. In fact, the example above constitutes, not a decrease in the per-unit cost of precaution, but rather an increase in the per-unit productivity of precaution which, as discussed above, is predicted to lead, not to decreased, but increased, physician liability. Table 2 presents data on physician earnings and fees which measure physicians’ opportunity cost of providing precaution. First note that physicians’ earnings are as much as six times the earnings of manufacturing workers. Further, from 1930 to 1988, physicians’ net income rose at an annual average rate of 5.7% while earnings for manufacturing workers rose at 4.9% annually. In terms of opportunity cost, it has become more costly for physicians to supply precaution relative to manufacturing workers.22 Likewise, the use of physician fees to measure opportunity costs also indicates that physician precaution costs have risen. For example, from 1935 to 1988, physicians’ fees have averaged a 4.5% annual increase while prices in general have only risen at a 3.8% annual rate. Further, the largest increases in fees occurred subsequent to 1970. The Harvard Practice Study (1990) also provides further evidence that increased precaution costs are associated with increased liability. They find a positive and generally significant relationship between total hospital costs per discharge and total claims per discharge. Thus, these results support the theoretical predictions to the extent that costs per discharge correctly measure precaution costs.23

Date

Physician Average Annual Net Incomea

Average Annual Growth Rateb

Average Annual Earnings in Manufacturingc

Average Annual Growth Rateb

Index of Physicians’ Fees (1967 = 100)d

Average Annual Growth Rateb

Consumer Price Index (1967 = 100)

Average Annual Growth Rateb

1930 1940 1950 1960 1970 1980 1988 1930 to 1988

$4,870 $4,441 $12,324 $22,100 $41,500 $83,700 $117,800 NA

NA –0.9% 10.7% 6.0% 6.5% 7.3% 4.4% 5.7%

$1,300 $1,152 $2,937 $4,550 $6,771 $14,500 $21,300 NA

NA –1.3% 9.8% 4.1% 4.1% 7.9% 4.9% 4.9%

39.2 39.6 55.2 77.0 121.4 269.3 492.1 NA

NA 0.2% 3.4% 3.4% 4.7% 8.3% 7.8% 4.5%

41.1 42.0 72.1 88.7 116.3 250.0 355.0 NA

NA 0.4% 5.6% 2.1% 2.7% 8.0% 4.5% 3.8%

Sources: United States Department of Commerce (1975, 1988, 1990). a after 1960 equals median annual net income. b [(Numbert/Numbert ⫺ n)1/n ⫺ 1] * 100. c the number listed for 1930 is from 1929; the number listed for 1940 is from 1939. d the number listed for 1930 is from 1935.

The Efficiency of Medical Malpractice Law: A New Appraisal

Table 2. Physician Earnings and Fees Since 1930

259

260

REED NEIL OLSEN

The Impact of Enforcement Costs Available evidence indicates that both increased technology and precaution costs help to explain historical, as well as recent, increases in physician liability. However, decreases in the relative cost of physicians defending a malpractice claim may also help. In the theoretical model presented above such physician defense costs were termed enforcement costs. The theoretical model predicts that as physician defense costs or enforcement costs decrease physician liability should increase. Historically, physicians have always searched for methods to reduce their cost of defending against malpractice suits. A number of methods were used by physicians in the nineteenth century to reduce defense costs (DeVille, 1990). For example, in the early-1800s many physicians, mostly due to intense competitive pressures, were willing to testify against fellow physicians. Since expert testimony is generally needed to establish a claim, the availability of expert witnesses decreases the cost of prosecuting a claim and increases defense costs. However, by the late-1800s, physicians had established strengthened professional organizations which were crucial in reducing plaintiffs’ use of expert witnesses. In addition, literature explicitly for the purpose of reducing the costs of malpractice defenses also became available in the 1800s and proliferated thereafter.24 In the 1880s, physicians began to band together and formed mutual defense organizations which were used to reduce defense costs. Each member physician would contribute to a legal defense fund for organization members. The association would then place an experienced law firm on retainer with the obtained funds to defend members accused of malpractice. Group defense organizations such as these were the first to institute forms of malpractice insurance (DeVille, 1990). After the full evolution of medical malpractice insurance, insurers themselves have filled the role of group defense and have increasingly been involved with defense efforts as litigation rates have risen. For example, by 1984 insurers spent an average of $10,985 dollars per filed claim both to investigate the validity of the claim and to defend their insured physicians (GAO, 1987a). By far the majority of these costs (91%) were payments for legal counsellors and expert witnesses.25 Variations in Malpractice Liability The data presented above provides evidence that medical malpractice liability has responded to technology, precaution costs, and enforcement costs as

The Efficiency of Medical Malpractice Law: A New Appraisal

261

predicted by the model. Much of the focus has been on explaining historically significant growth rates in U. S. malpractice frequency and severity. However, significant cross-sectional variations in medical malpractice frequency and severity also exist. The case for the empirical validity of the theoretical model is strengthened to the extent that the model also explains these cross-sectional variations. In addition to jurisdictional variations in claim frequency and severity, significant variations have always been found between different medical specialists, as well as between different schools of medical practice. In the early-1800s, for example, no unanimity existed on correct treatment methods for most injuries and diseases. As a result, different schools of practice developed with radically different treatment practices. Regular physicians based their treatment upon medical science similar to much of current medical practice. However, other schools existed and competed with regular physicians for patients. The most notable alternate schools of practice were Thomsonians, homeopaths, and hydropaths. Thomsonians based their treatments on regimes made popular by Samuel Thomson. Thomsonians were self-taught and believed in the use of ‘natural’ cures, such as natural purgatives used to induce vomiting. Homeopathy was based on the doctrine that ‘like cures like’. Homeopaths administered minute amounts of substances that caused similar symptoms to the malady being treated. Hydropaths stressed the curative values of water. Additional schools of thought included natural bonesetters, mesmerists, root doctors, and phrenologists.26 Before 1860, trial judges often allowed members of different schools to testify against alternate practitioners. However, early appellate court decisions consistently overturned verdicts based upon such testimony. By 1860, it was firmly established that there: is no particular system of medicine established or favored by the [law]. . ..[t]he people are free to select from the classes of medical men. . ..While a regular physician is expected to follow the rules of the old school in the art of curing, the botanic physician must be equally expected to adhere to his adopted method.27

The adoption of different rules and liability levels for different schools continued into the twentieth century. McCoid (1959) noted the continued existence of the rule in the late-1950s. However, to avail themselves of the rule physicians must adhere to a school which has a regular mode of practice.28 The existence of the ‘school of practice’ rule has a two-fold explanation. First, even with large technological advances, medical treatment was, and is, by no means certain. The rule merely reflected the fact that medical science itself was not certain of the ‘correct’ treatment in many instances (McCoid, 1959). As the

262

REED NEIL OLSEN

productivity of a certain treatment has increased, the older schools have gradually given way to the more productive mode of treatment. Second, the rule appears to reflect inherent differences in productivity between the different schools as predicted by the theoretical model. For example, ‘regular’ medicine has historically proven itself to be more effective than alternate modes of practice in many, if not most, instances. Thus, if variations in the legal rule were productivity based, one would expect to observe lower liability for the alternate modes of practice. As expected, during the heyday of the alternate practitioner, suits against such practitioners were quite rare (DeVille, 1990). Even more telling, as suits against regular physicians increased into the twentieth century, suits against alternate practitioners slowly died away (McCoid, 1959; DeVille, 1990). In addition to explaining the school of practice rule, the theoretical model also helps explain why liability varies by medical specialty. Table 3 illustrates the existence of this variation by specialty, in both the risk of suit and the expected awards, using data from the GAO’s (1987a) 1984 survey of closed malpractice claims. For example, although obstetricians and gynecologists represent only 5.2% of all physicians in the United States, in the GAO (1987a) sample they generated over 15% of all claims. Given a claim was filed, OB/ GYNs also had an expected award of $93,171; the second highest expected award of any specialty. Other high-risk specialties as measured by expected awards include pathologists, pediatricians, and orthopedic surgeons. The fact that expected liability levels vary by specialty does not necessarily mean that the variation is explained by the theoretical model. Independent evidence does exist, however, of the relevancy of the model. For example, both the frequency of claims and the probability of claims being paid are higher in those specialties which have exhibited the largest productivity increases.29 Table 4 demonstrates that board-certified specialists averaged more than twice the number of cases than did non-certified specialists in the 1940s and 1950s. The GAO’s (1987a) 1984 survey of malpractice insurers also found that boardcertified physicians generated more cases; 3.5 times the number of cases as did non-certified physicians. Given that board certification is an adequate proxy for productivity, it appears that increased productivity does enhance a specialist’s chance of being sued.30 The GAO’s (1987a) data also indicates that physicians with previous malpractice claims are more likely to have current claims against them – another indication that productivity is an important determinant of physician liability. The importance of physician productivity is further illustrated by changes in legal doctrines which affect expected liability levels. Consider, for example, historical changes in the locality rule. The locality rule allows standards of care

Physician Claims, Awards, and Expected Awards in 1984 by Specialtya Claims

Physician Specialty Obstetrics/Gynecology General surgery Orthopedic surgery Internal medicine General practice Family practice Radiology Emergency Medicine Anesthesiology Plastic Surgery Urology Pediatrics Ophthalmology Neurosurgery Otolaryngology Psychiatry Pathology Totald

Awards for Paid Claims

Physician Distribution in the United Statesb

Total Claims

Claims Paid

Percent Paid

Percent of Total Claims

Median Awards

Average Awards

Expected Award

Total Physicians

Percent of Total

5,165 3,952 3,283 1,288 2,076 2,400 520 826 1,851 1,998 912 1,640 1,260 828 632 957 195

2,711 1,173 1,493 629 1,071 717 231 364 884 618 771 492 578 247 419 637 195

52.49% 29.68% 45.48% 48.84% 51.59% 29.88% 44.42% 44.07% 47.76% 30.93% 84.54% 30.00% 45.87% 29.83% 66.30% 66.56% 100.00%

15.62% 11.95% 9.93% 3.90% 6.28% 7.26% 1.57% 2.50% 5.60% 6.04% 2.76% 4.96% 3.81% 2.50% 1.91% 2.89% 0.59%

$75,000 $49,000 $25,000 $10,000 $45,000 $15,000 $19,500 $7,000 $3,000 $22,500 $625 $195,000 $10,000 $10,000 $23,500 $25,000 $250,000

$177,509 $120,889 $80,059 $42,757 $50,264 $40,339 $53,101 $22,640 $42,680 $70,172 $14,896 $198,684 $55,593 $65,226 $23,264 $34,914 $197,652

$93,170.75 $35,881.27 $36,408.19 $20,880.55 $25,930.99 $12,051.28 $23,589.10 $9,976.95 $20,383.10 $21,704.85 $12,593.00 $59,605.20 $25,502.19 $19,457.51 $15,423.44 $23,239.52 $197,652.00

25,234 31,308 14,572 60,118 29,399 31,195 19,893 7,811 16,845 3,193 7,889 28,027 13,281 3,498 NA 27,303 12,502

5.20% 6.45% 3.00% 12.39% 6.06% 6.43% 4.10% 1.61% 3.47% 0.66% 1.63% 5.78% 2.74% 0.72% NA 5.63% 2.58%

33,068

14,749

44.60%

100.00%

$25,000

$85,179

$37,991.56

485,123

100.00%

263

Source: GAO (1987a) Table 4.2 and Table 4.4 a Data only includes paid claims involving only one physician. Claims against multiple physicians are not included in the data. b Data as of December 31, 1981, the year most claims were filed. c Expected Awards Given a Claim has been filed. Equals Percent of Claims Paid x Average Awards. d Details do not add to totals due to the exclusion of 26 low-risk specialties.

The Efficiency of Medical Malpractice Law: A New Appraisal

Table 3.

264

REED NEIL OLSEN

Table 4. Number of Appellate Malpractice Suits by Type of Practice and Certification 1946 to 1956a

Certified

Non-Certified

Ratio of Certified to Non-Certified

Surgicalb Orthopedics Obstetricsd Urology Neuropsychiatry Otolaryngology Internal Medicine

28 27 15 6 8 7 5

20 4 4 3 4 0 5

1.40 6.75 3.75 2.00 2.00 NA 1.00

Total

96

44

2.18

Specialty

Source: Sandor (1957), Table 2. a Includes only specialties with more than 5 cases. b Includes general, plastic, and neurosurgery. c Includes orthopedics and radiology. d Includes gynecology.

and, hence, expected liability levels to vary by location. The early application of the locality rule was clarified by Judge Nathan Weston in an 1821 malpractice case: It is not to be expected of a Surgeon or a Physician in a country or obscure village, that he will possess the skill of a surgeon in the city of London, or any large city – this would be unreasonable to expect. . . all that is required is ordinary skill according to the general state of medical science in the section of the country in which he lives.31

Although the locality rule was used sporadically by some trial court judges in the early-1800s, it was not supported by appellate court decisions until the 1860s and 1870s (DeVille, 1990.) The increased use of the locality rule in the late-1800s corresponded to the technological innovations in medical practice which arose during the same time period. Thus, as productivity increased some, usually rural, physicians were insulated from the increased liability which inevitably followed the innovations. However, as DeVille (1990) noted, the application of these innovations was dependent upon education, place of practice, and dissemination of the technology. Since access to improved technology with enhanced physician productivity varied widely between different locales, the locality rule also allowed liability levels to vary. Widespread differences in the application of new innovations are alone, of course, not a sufficient justification for the development of the locality rule. In

The Efficiency of Medical Malpractice Law: A New Appraisal

265

the late-1800s and early-1900s, however, variations in medical training and education, as well as in opportunities to apply the new technology to difficult cases, justified the use of the locality rule.32 The locality rule did not allow physicians in small localities to completely ignore technological innovations. As these innovations were disseminated and became standard practice in small localities, the locality rule required their use by all local physicians (DeVille, 1990). Thus, the locality rule was used to insulate the less-educated and lesstechnologically-adept physicians, who nevertheless provided valuable services to rural areas, from the same liability levels that their more productive colleagues faced. Recently, the locality rule has been modified, or even abolished, in most jurisdictions (Danzon, 1985; Weiler, 1991). The most common modification has been to widen the locality to which local physicians are compared. For example, many states have adopted a state-wide common practice standard. Other states set the standard to be one as practiced in the current or similar localities. Arguably, the recent widening in the locality rule mirrors technological innovations and the increased access of rural physicians to technological advances.

IV. CONCLUSIONS The perception that the United States medical malpractice system is out of control and desperately in need of renovation is pervasive in the malpractice literature. Although this view is often taken for granted, there exists no common prescription on the malpractice system’s repair. Proposals range from outright abolition of the system by moving to a contractual basis for malpractice to various no-fault insurance alternatives. Recent increases in malpractice frequency and severity serve as the major impetus for the perception that the malpractice system is in disarray. These increases are so large and unprecedented, it is claimed, that they demonstrate the need for serious reform of the malpractice system. However, historical statistics, as well as a careful examination of the history of medical malpractice, do not substantiate these claims. Recent growth rates in malpractice claim frequency and severity are similar in magnitude to historical growth rates. Further, available evidence indicates that these growth rates are a result of an efficient adjustment of physician liability levels. As physicians’ productivity has increased with technological innovations, their liability has increased as well. Increases in physicians’ liability can also be traced to increases in physician precaution costs and decreases in the costs of defending malpractice claims. Variations in physician productivity, precaution

266

REED NEIL OLSEN

costs, and defense costs also appear to explain much of current, as well as historical, variations in liability between jurisdictions, specialties, and schools of practice. The idea that medical malpractice rules efficiently adjust to changes in the medical market is an intriguing possibility in need of further study. At the very least, the evidence presented in this paper suggests the need for caution in applying ‘quick fix’ reforms to the medical malpractice system. In fact, legislative reforms of the malpractice system have already been enacted in many states over the past two decades. However, the enactment of these reforms may also reflect jurisdictional variations in productivity, precaution costs, or enforcement costs. Such a prospect has never before been seriously entertained or empirically tested. The idea that human institutions, especially the common law, respond efficiently to maximize societal welfare is not new; it had just not been previously applied to medical malpractice.33 Once applied, though, the possibility cannot be lightly dismissed especially in light of its apparent empirical relevancy. Since the beginning of malpractice in the United States, physicians have criticized its existence, called for its reform, and fought against its expansion. Many early physicians ‘maintained that the only sensible solution was to refuse to treat certain classes of patients and injuries’.34 Yet, even as physicians fought against both the existence and the expansion of medical malpractice, with inevitable regularity each technological innovation, each expansion in precaution costs, and each reduction in defense costs have increased physician liability as predicted by the model.

NOTES 1. See for example, Bovbjerg (1989), Blackman & Bailey (1990), Danzon (1985), GAO (1986a, 1986b, 1987a, 1987b), Harvard Medical Practice Study Group (hereafter referred to as Harvard Practice Study) (1990), Sloan, et al. (1991), U.S. Dept. of Health, Education, and Welfare (1973), and U.S. Dept. of Health and Human Services (hereafter referred to as HHS Report) (1987), and Winter (1988). 2. For example, Epstein (1976, 1978, 1984, 1986), Robinson (1986b), and Havighurst (1983, 1986) have proposed we move to a contractual basis for malpractice liability. Others have proposed a no-fault medical malpractice system (AMA 1987; Danzon, 1985; Johnson et al., 1989; Weiler, 1991). Bovbjerg (1989) and Olsen (1996) overview the major malpractice reforms enacted by state legislatures. 3. This assumption is not needed to show that an equilibrium exists. Cooper & Ross (1985) show that an equilibrium exists for all of the three possible assumptions about the sign of pxy. The assumption does, however, make the comparative statics of the model more tractable. 4. Enforcement costs are similar to Calabresi’s (1970) administration costs and include both direct and indirect legal costs borne by each of the parties. Each party bears

The Efficiency of Medical Malpractice Law: A New Appraisal

267

only their own enforcement costs while court enforcement costs are generally borne by society as a whole. According to Prosser & Keeton (1984) this is generally consistent with rules in American courts. 5. Both Landes & Posner (1985) and Shavell (1980) consider the quantity of the good being exchanged to be a choice variable. I make the simplifying assumption that quantity is not a choice variable. I make this assumption because I am not, as Shavell was, primarily interested in analyzing the impact that activity levels have upon tort liability rules. 6. Although of importance, the process by which the common law chooses efficient liability rules is beyond the scope of this paper. For relevant literature see Rubin (1977), Priest (1977), Landes & Posner (1984), and Posner (1981). 7. See Weiler (1991), Danzon (1985) and McCoid (1959). 8. Both being attentive during a medical procedure or taking the time to make a careful diagnosis are examples of what Grady (1988) terms non-durable precaution, since this type of precaution does not endure for future treatments. 9. Grady (1988) makes a related argument with, however, some fundamental differences. For example, I am arguing that increased technology increases the ability of the physician to cause harm. As a result, malpractice law efficiently responds by increasing physician liability thereby giving physicians an incentive to take more care. Grady, however, views increased technology as increasing physician liability, not because of an efficient change in the law, but because the law has not changed and applies old standards to the new technology. Hence, increased physician liability, especially the increased frequency of malpractice cases due to what he terms nondurable precaution, results from an inefficient application of existing law to advances in technology. 10. Of course, increased technology may be combined with increased expectations of performance by patients which may well explain part of the increased tendency to file malpractice cases (e.g. Reder, 1976). However, as long as those increased expectations are reasonable, given the new medical technology or knowledge, any increased tendency to file lawsuits is directly related to enhanced physician productivity and fits within the theoretical model. Only unreasonable expectations would lie outside the model’s predictions and, as Reder (1976) notes, quantifying either the existence or the extent of such patient attitudes is problematic. Further, unreasonable expectations would only lead to increased claim frequency and not to either the increased probability of winning claims nor the increased average severity of paid claims that have accompanied both historical and recent increases in claim frequency (Olsen, 1996; Weiler, 1991; DeVille, 1990; Danzon, 1985). Hence, the hypothesis that unreasonable patient expectations explain why increased technology has historically been associated with higher physician liability is both unquantifiable and inconsistent with the empirical data. 11. A large number of authors have noted the relationship between improved medical technology and increased medical knowledge and increased physician liability (see note 16 infra). Some researchers have noted that a competing explanation for this relationship is the impact that increased technology, especially the increased complexity of new treatments which rely more upon machines than personal care, may have upon the willingness of patients to sue physicians (Danzon, 1984, 1985, 1986; Robinson, 1986a; Reder, 1976). Thus, this explanation hinges upon the two related concepts that patients are more likely to file lawsuits as medical care becomes more impersonal and

268

REED NEIL OLSEN

that enhanced technology tends to depersonalize medical care. Supporting this hypothesis, Danzon (1984, 1986) finds that the most important factor explaining increased claim frequency is urbanization which, she speculates elsewhere (Danzon, 1985), may be related to more impersonal care in urban areas. However, a number of other factors tend to discredit the hypothesis. First, Danzon (1986) also finds that increased complexity of medical care, which according to this theory should depersonalize medicine, has no impact on malpractice claim frequency. Second, she finds that urbanization tends to decrease claim severity and, thereby, actually decrease physician liability. Third, as Danzon (1985) and others note (Reder, 1976; Robinson, 1986a), it is not possible to directly test this hypotheses and they relegate it to ‘unquantified – and perhaps unquantifiable – speculation’ (Danzon, 1985, p. 83). Fourth, and more importantly, this competing hypothesis cannot explain the entire history of the increased physician liability that has resulted from new technology. As shown by DeVille (1990) and briefly discussed below, new technology explains not only the recent rise in physician liability, which may well be tied to more impersonal types of medical care, but also historical increases which have not generally been associated with more impersonal care (e.g. advances in the treatment of infection through new antiseptic techniques). 12. The technical advances in these specialties were, by modern standards, tenuous at best. Physicians did assert technical expertise in these areas and it appears that improvements were made in treatment over previous standards. Vaccinations, for example, had been proven as treatment for diseases in Europe and statistics from Europe convinced Americans of their value (DeVille, 1990; Cassedy, 1984). However, such advances were still subject to a large amount of uncertainty by current standards. 13. Cases were brought for: (1) fracture (2) hernia (3) amputation (4) laceration (5) abandonment (6) obstetric (7) dislocation (8) vaccination (9) calomel (10) aneurism (11) patent medicine (12) death (13) bleeding (14) misdiagnosis (15) chloroform (16) ocular, and (17) tonsils. See DeVille (1990), Table 2. 14. Smith’s (1860) survey of malpractice cases in The American Journal of Medical Science (see DeVille, 1990, p. 33), found that fracture and dislocation cases accounted for 63 percent of the cases. Other contemporary sources were in general agreement with this figure. For an overview of this literature, see DeVille (1990). 15. Cooper (1813) as quoted by DeVille (1990, p. 95) notes that the most common method of treating a compound fracture was amputation because although ‘apparently desperate cases [of compound fracture] are sometimes cured, . . . every man also knows, that such escapes are very rare to admit of being made precedents and the majority of such attempts fail’. DeVille gives a number of additional contemporary sources which confirm that standard practice prior to 1835 was amputation for compound fractures. 16. DeVille (1990, p. 104). Weiler (1991), Edwards (1989), Giesen (1988), Grady (1988), Robinson (1986a), Danzon (1985), Werthman (1984), Reder (1976), Mechanic (1975), McCoid (1959), and Sandor (1957) all note the direct relationship between increased productivity (i.e. better technology) and the increased frequency and severity of medical malpractice suits. 17. See DeVille (1990), Rothstein (1972), and Brieger (1966). Lister’s theories were built on Louis Pasteur’s work which found that infections were caused by microorganisms. He used this theory to develop a somewhat effective system of cleansing wounds by the use of carbolic acid.

The Efficiency of Medical Malpractice Law: A New Appraisal

269

18. See Rothstein (1972). Additional techniques included sterilization of surgical instruments and the use of sterile caps, gowns, masks, and rubber gloves. Additional innovations included the invention of more sophisticated surgical instruments, the use of finer needles and thread to control bleeding, and the development of better analgesics and anesthetics. 19. See Werthmann (1984) for more details on these recent advances in the law. She provides more complete United States case studies and analysis of the technological innovations which underlie reproductive impairment cases. For an overview of nonUnited States cases see Giesen (1988). 20. In addition to DeVille (1990), see note 16 for a more complete list of references. 21. A common claim, especially by physicians and the popular press, is that patients file lawsuits primarily because of the ‘deep pockets’ of physicians. As shown by the data discussed below, physician incomes have risen substantially in real terms over time. Further this increase in physician income at least partially results from their increased productivity caused by increased technology. As noted above (see notes 10 and 11 supra) claims about the attitudes patients take, which might explain increased physician liability, are generally unquantifiable and, hence, untestable. In this particular instance, though, it is well established in the general law and economics literature that increased defendant ability to pay would, theoretically, be expected to result in a higher claim frequency (e.g. Shavell, 1982). Rather than a claim competing against the theoretical model, however, this claim fits within its confines. That is, the model also predicts that ‘deep pockets’ (i.e. increased physician precaution costs) should result in increased liability as is discussed in more detail in this section. 22. Notice that after 1970 average annual earnings growth rates are lower for physicians than for manufacturing workers. However, these numbers may understate the true growth in average physician income since median income figures must be used after 1960. 23. See Harvard Practice Study (1990) Tables 10.6, 10.7, and 10.8. These results are somewhat tentative especially as applied to the theoretical model in this paper. First, their results are sensitive to functional form. Second, it is not clear that hospital costs per discharge are an adequate measure of precaution costs. Hospital costs are generally measured as charges to the patient or insurance company. However, the Harvard Practice Study’s data has been adjusted in an attempt to truly reflect costs rather than charges using a hospital specific cost/charge ratio. 24. See DeVille (1990) for a discussion of this trend in the 1800s. Numerous physician guides to medical malpractice are currently available. For examples, see Blackman & Bailey (1990) and Kramer & Kramer (1983). 25. Of course, the increase in physician incomes, presented in Table 2, would tend to increase physicians’ personal time costs of defending against a malpractice claim, at least to the extent that physicians spend their own time, rather than experts such as lawyers and expert witnesses, defending against such a claim. Which of these two competing impacts, increased time costs or the decreased defense costs as described above, is largest is problematic. At the very least, however, physician efforts to reduce their defense costs is an indication that at least part of the historical increases in physician liability may result from such efforts. 26. See DeVille (1990) and McCoid (1959) for descriptions of a number of the different schools of practice.

270

REED NEIL OLSEN

27. Bowman v. Woods, 1 Green 441 (Iowa 1848). 28. For example, McCoid (1959) lists cases where ‘clairvoyant’ physicians, magnetic healers, and Chinese herb doctors were denied the right to exclude alternate practitioners as expert witnesses because of a lack of a regular mode of practice that potential patients could evaluate. 29. Obstetrics/gynecology and surgery have, as noted above, exhibited enormous productivity increases. Of course, changes in productivity might not be the only explanation for higher expected awards for some specialties. For example, some specialties may be more intrusive with resultant increases in the actual accident losses. Of course, for any given liability rule as accident losses increase so would the physician’s liability. Thus, part of the relatively large expected award for surgery cases undoubtedly stems from the more intrusive nature of surgery. 30. This conclusion, however, is somewhat tentative as the ratio of certified to noncertified specialists is unknown both for Sandor’s data and for the GAO survey. 31. Lowell v. Faxon & Hawks as quoted in Adams (1825) and DeVille (1990, p. 18). 32. The appellate judge explicitly acknowledged this in Small v. Howard 128 Mass. 131 (1880). According to DeVille (1990, p. 212) the appellate judge recognized that it was: ‘a matter of common knowledge that physicians in small country towns and villages could not possess the same degree of skill as their big city counterparts, who had more opportunities to observe practice’. 33. For examples see Rubin (1977), Landes & Posner (1984, 1985), Posner (1981, 1986), and Barzel (1989). 34. DeVille (1990, p.197) paraphrasing an earlier anonymous commentator on medical malpractice.

REFERENCES Adams, J. Jr. (1825). Report of a Trial of an Action, Charles Lowell against John Faxon and Micajah Hawks, Doctors of Medicine, Defendants for Malpractice in the Capacity of Physicians and Surgeons at the Superior Judicial Court of Maine Held at Machias for the County of Washington June, 1824 before the Hon. Nathan Weston. Portland, Maine: Printed for James Adams Jr., by David and Seth Paine. American Medical Association, Report of the Specialty Society Medical Liability Project (1987). A Proposed Alternative to the Civil Justice System for Resolving Medical Liability Disputes: A Fault-Based, Administrative System. Chicago, Illinois: American Medical Association. Barzel, Y. (1989). Economic Analysis of Property Rights. New York: Cambridge University Press. Blackman, N., & Bailey, C. (1990). Liability in Medical Practice: A Reference for Physicians. New York: Harwood Academic Publishers Bovbjerg, R. R. (1989). Legislation on Medical Malpractice: Further Developments and a Preliminary Report Card. U.C. Davis Law Review., 22, 499–556. Brieger, G. H. (1966). American Surgery and the Germ Theory of Disease. Bulletin of the History of Medicine, 41, 135–145. Brown, J. P. (1973). Toward an Economic Theory of Liability. Journal of Legal Studies, 2, 323–349.

The Efficiency of Medical Malpractice Law: A New Appraisal

271

Calabresi, G. (1970). The Costs of Accidents: A Legal and Economic Analysis. New Haven: Yale University Press. Cassedy, J. H. (1984). American Medicine and Statistical Thinking, 1800 – 1860. Cambridge, Massachusetts: Harvard University Press. Cooper, R., & Ross, T. W. (1985). Product Warranties and Double Moral Hazard. Rand Journal of Economics, 16, 103–113. Danzon, P. (1985). Medical Malpractice: Theory, Evidence and Public Policy. Cambridge, Massachusetts: Harvard University Press. DeVille, K. A. (1990). Medical Malpractice in Nineteenth-Century America: Origins and Legacy. New York: New York University Press. Edwards, F. (1989). Medical Malpractice: Solving the Crisis. New York: H. Holt. English, P. C. (1980). Shock, Physiological Surgery, and George Washington Creel: Medical Innovation in the Progressive Era. Westport, Connecticut: Greenwood Press. Epstein, R. (1976). Medical Malpractice: The Case for Contract. American Bar Foundation Research Journal, 1976, 87. Epstein, R. (1978). Medical Malpractice: its Cause and Cure. In: S. Rottenberg (Ed.), The Economics of Medical Malpractice. Washington, D.C.: American Enterprise Institute for Public Policy Research. Epstein, R. (1984). In Defense of the Contract at Will. University of Chicago Law Review, 51, 947–982. Epstein, R. (1986). Medical Malpractice, Imperfect Information, and the Contractual Foundation for Medical Services. Law and Contemporary Problems, 49, 201–212. Giesen, D. (1988). International Medical Malpractice Law: A Comparative Study of Civil Liability Arising from Medical Care. Boston, Massachusetts: Kluwer Academic Publishers. Haddock, D., & Curran, C. (1985). An Economic Theory of Comparative Negligence. Journal of Legal Studies, 14, 49–72. Harvard Medical Practice Study Group (1990). Patients, Doctors, and Lawyers: Medical Injury, Malpractice Litigation, and Patient Compensation in New York: a report by the Harvard Medical Practice Study to the State of New York. [Cambridge, Massachusetts?]: President and Fellows of Harvard College. Havighurst, C. C. (1983). Decentralizing Decision Making: Private Contract Versus Professional Norms. In: J. A. Meyer (Ed.), Market Reforms in Health Care. Washington: American Enterprise Institute for Public Policy Research. Havighurst, C. C. (1986). Private Reform of Tort-Law Dogma: Market Opportunities and Legal Obstacles. Law and Contemporary Problems, 49, 143–172. Johnson, K. B., Phillips, C. G., Orentlicher, D., & Hatlie, M. S. (1989). A Fault-Based Administrative Alternative for Resolving Medical Malpractice Claims. Vanderbilt Law Review, 42, 1365–1406. Kramer, C., & Kramer, D. (1983). Medical Malpractice. New York: Practicing Law Institute. Landes, W., & Posner, R. (1984). The Positive Economic Theory of Tort Law. Georgia Law Review, 15, 851–924. Landes, W., & Posner, R. (1985). A Positive Economic Analysis of Products Liability. Journal of Legal Studies, 14, 535–583. McCoid, A. H. (1959). The Care Required of Medical Practitioners. Vanderbilt Law Review, 12, 549–632. Mechanic, D. (1975). Some Social Aspects of the Medical Malpractice Dilemma. Duke Law Journal, 1975, 1179–1196.

272

REED NEIL OLSEN

Olsen, R. N. (1996). The Reform of Medical Malpractice: Historical Perspectives. The American Journal of Economics and Sociology, 55, 257–275. Peterson, M. (1987). Civil Juries in the 1980s: Trends in Jury Trials and Verdicts in California and Cook County, Illinois. Santa Monica, California: The Rand Corporation, RAND Working Paper No. R–3466-ICJ (1987). Posner, R. (1981). A Reply to Some Recent Criticisms of the Efficiency Theory of the Common Law. Hofstra Law Review, 9, 775–794. Posner, R. (1986). Economic Analysis of Law. Boston: Little, Brown and Company. Priest, G. L. (1977). The Common Law Process and the Selection of Efficient Rules. Journal of Legal Studies, 6, 65–82. Prosser, W., & Keeton, R. (1984). A Handbook on the Law of Torts. St. Paul, Minnesota: West Publishing Company. Reder, M. W. (1976). Medical Malpractice: An Economist’s Viewpoint. American Bar Foundation Research Journal, 1976, 511–563. Robinson, G. O. (1986a). The Medical Malpractice Crisis of the 1970’s: a Retrospective. Law and Contemporary Problems, 49, 5–36. Robinson, G. O. (1986b). Rethinking the Allocation of Medical Malpractice Risks Between Patients and Providers. Law and Contemporary Problems, 49, 173–200. Rothstein, W. G. (1972). American Physicians in the Nineteenth Century. Baltimore: Johns Hopkins University Press. Rubin, P. (1977). Why is the Common Law Efficient? Journal of Legal Studies, 6, 51–63. Sandor, A. A. (1957). The History of Professional Liability Suits in the United States. Journal of the American Medical Association, 163, 459–466. Shavell, S. (1980). Strict Liability Versus Negligence. Journal of Legal Studies, 9, 1–25. Shavell, S. (1982). Suit, Settlement, and Trial: A Theoretical Analysis Under Alternative Methods for the Allocation of Legal Costs. Journal of Legal Studies, 11, 55–81. Sloan, F. A., Bovbjerg, R. R., & Githens, P. B. (1991). Insuring Medical Malpractice. New York: Oxford University Press. United States Department of Commerce, United States Bureau of the Census (1975). Historical Statistics of the United States: Colonial Times to 1970. Washington, D.C.: Government Printing Office. United States Department of Commerce, United States Bureau of the Census (1988). Statistical Abstract of the United States 1988. Washington, D.C.: Government Printing Office. United States Department of Commerce, United States Bureau of the Census (1990). Statistical Abstract of the United States 1990. Washington, D.C.: Government Printing Office. United States Department of Health, Education, and Welfare (1973). Report of the Secretary’s Commission on Medical Malpractice. Washington, D.C.: Government Printing Office. United States Department of Health and Human Services (1987). Report of the Task Force on Medical Liability and Malpractice. Washington, D.C.: Government Printing Office. United States General Accounting Office (1986a). Medical Malpractice: Six State Case Studies. Washington, D.C.: Government Printing Office. United States General Accounting Office (1986b). Medical Malpractice Insurance Costs Increased but Varied among Physicians and Hospitals. Washington, D.C.: Government Printing Office. United States General Accounting Office (1987a). Medical Malpractice: Characteristics of Claims Closed in 1984. Washington, D.C.: Government Printing Office. United States General Accounting Office (1987b). Medical Malpractice: A Framework for Action. Washington, D.C.: Government Printing Office.

The Efficiency of Medical Malpractice Law: A New Appraisal

273

Weiler, P. C (1991). Medical Malpractice on Trial. Cambridge, Massachusetts: Harvard University Press. Werthmann, B. (1984). Medical Malpractice Law: How Medicine is Changing the Law. Lexington, Massachusetts: Lexington Books. White, Michelle (1989). An Empirical Test of the Comparative and Contributory Negligence Rules in Accident Law. Rand Journal of Economic, 20, 308–330. White House (1993). Health Security: The President’s Report to the American People. Washington, D.C.: Government Printing Office. Winter, R. A (1988). The Liability Crisis and the Dynamics of Competitive Insurance Markets. Yale Journal on Regulation, 5, 455–499.

This Page Intentionally Left Blank

THE EFFECTS OF HOSPITAL COMPETITION ON NONPROFIT AND FOR-PROFIT HOSPITAL MEDICAID SHARE Sheri L. Eisert ABSTRACT This paper empirically tests whether the nonprofit hospital is influenced by the community benefit standard by analyzing differences in nonprofit and for-profit hospital inpatient Medicaid share, while controlling for the effects of hospital market competition. Hospital specific data for this study are from the 1991 American Hospital Association Annual Survey of Hospitals and the 1991 Medicare Minimum Cost Report by the Health Care Financing Administration. The sample includes 192 for-profit, 899 nonprofit and 221 government-owned general acute care hospitals. Key variables in the multivariate analysis include hospital inpatient Medicaid share as the dependent variable and Medicaid demand, market competition and input quality as independent variables. The principal findings are that nonprofit hospitals serve a larger share of Medicaid inpatients than for-profit hospitals although the greater the presence of for-profit hospitals in the market (one measure of market competition) the smaller the nonprofit hospital Medicaid share. It is concluded that nonprofit hospitals reduce their level of community benefit, where community benefit is

Research in Law and Economics, Volume 19, pages 275–294. 2000 JAI/Elsevier Inc. ISBN: 0-7623-0308-5

275

276

SHERI L. EISERT

measured as a hospital’s percent Medicaid share, in markets with a greater presence of for-profit hospitals.

I. INTRODUCTION This paper explores the impact of the differences in the objectives of nonprofit and for-profit hospitals, and the effects of market competition, on their decision to accept Medicaid patients. Providing an alternative for unwanted demand by nonprofit and for-profit hospitals, government-owned hospitals are treated as ‘hospitals of last resort’. For the purposes of this paper, hospitals are community acute care hospitals, where nonprofit hospitals are synonymous with private not-for-profit hospitals and government-owned hospitals refer to state and local hospitals, including tax-district hospitals. The literature on nonprofit organizations contains a wide range of hypothesized objective functions. Economists have distinguished the objective function of nonprofits from for-profits as involving the maximization of utility of the manager/owner constrained either by output or a budget rather than maximizing the profits of a firm.1 Others have modeled nonprofit objectives as maximizing profit (net revenue), which includes an additional term for social benefits,2 while others assume nonprofits maximize an expected social welfare function.3 Some theorists argue that it is difficult to develop a general nonprofit model because there is lack of a singular objective.4 One similarity among each of these specifications of nonprofit objectives includes a factor representing the community’s interests, where the differences lie in defining community interests and the weight placed on maximizing net revenue. In this study, community benefit is measured by the nonprofit hospital’s provision of inpatient Medicaid services, which is reimbursed at less than average cost in most states.5 Factors which are expected to influence nonprofit and for-profit inpatient Medicaid share are Medicaid reimbursement rates; Medicaid demand; and market competition and hospital input quality, where a hospital’s inpatient Medicaid share is defined as the percent of a hospital’s inpatient days which are for Medicaid patients. Unlike other studies analyzing the effects of hospital market competition using the Herfindahl Hirschman Index (HHI), this study considers the effects of the presence of for-profit hospitals in the market. Therefore, the focus of this study is to first determine whether nonprofit hospitals provide a larger share of their inpatient services to Medicaid patients than for-profit hospitals, providing a benefit to the community in exchange for tax-exempt status. Second, this study considers whether hospital competition has an influence over a nonprofit hospital’s decision to provide this community benefit. With an increased emphasis on

Hospital Competition on Nonprofit and For-profit Hospital Medicaid Share

277

hospital competition as a means to control health care costs, the effects of competition on nonprofit hospital behavior may be different from the effects on for-profit hospital behavior. The motivation for this study stems from the evolving interpretation of charitable as it applies to the tax-exempt status of nonprofit hospitals. The courts and the IRS have steered away from specifying an amount of free care nonprofit hospitals must provide to maintain their tax-exempt status and have considered other benefits to the community, such as providing underreimbursed care to those insured by Medicaid.6 Studies which have evaluated differences in the provision of free or uncompensated care by nonprofit and forprofit hospitals have had mixed results. Some studies have found that nonprofit hospitals provide more uncompensated care than for-profit hospitals, but these rates vary by state.7 One study that compared nonprofit and for-profit provision of charity care in California found no significant difference in the provision of charity care.8 Studies which have evaluated differences in Medicaid share have also been inconclusive. Another study using California data found that forprofits, on average, had only one percent less net revenue from MediCal patients than nonprofits.9 The authors attributed these results to the generous MediCal reimbursement rates at the time of the study. A study using a national sample of 160 hospitals (primarily California, Texas and Florida) compared the percent of nonprofit and for-profit Medicaid inpatient days and found no significant difference between these two organizational forms.10 The shortcoming with these studies includes the lack of control for market characteristics, such as poverty rates and the degree of market competition, as well as individual hospital characteristics. This study controls for these factors. Nonprofit hospitals contend nonprofit status does not dictate that they must provide an amount of free care or below-cost reimbursed care to the point where they are operating at a loss in the long run. With increased hospital competition for patients through contractual arrangements from private insurers, nonprofit hospitals are pressured to compete with the prices of forprofit hospitals while, at the same time, they are under increasing scrutiny to provide a community benefit. Historically, this community benefit has been interpreted as the provision of free care or below-cost reimbursement care. If nonprofit hospitals do not have the financial flexibility to provide a community benefit to justify their tax-exempt status, then are we ready to accept a hospital industry dominated by the for-profit motive? Is society willing to accept the tradeoff between the efficiencies gained from increasing hospital market competition and the loss of community benefit provided by the nonprofit hospital?

278

SHERI L. EISERT

II. THE ECONOMIC THEORY OF HOSPITAL DECISION MAKING Competition in the hospital industry has been characterized differently from other industries where price competition and profit maximization describe firm behavior. Empirical studies on hospital competition have shown that greater hospital competition leads to higher hospital costs.11 These results are consistent with the theory that hospitals compete on quality dimensions rather than price. However, recent studies have focused on the transition from hospital quality competition to price competition.12 These studies have generally found that while quality competition still exists, there are indications that hospitals are moving toward price competition strategies. Price competition does not describe hospital behavior for services provided to patients insured by government-Medicare and Medicaid. The price received from government payers is not driven by market competition since reimbursement rates are established by government based on average costs.13 Hospital revenue from Medicare and Medicaid accounts for approximately 50% of total hospital revenue; the remaining revenue is generated from private payers, where price is endogenous to hospital decision making. For these private-pay patients, the hospital market has been moving away from a patient-driven to a payer-driven form of competition. Nonprofit and for-profit urban hospitals compete for inpatient market share unlike rural hospitals, which often exist in monopolistic markets. A hospital’s decision to provide inpatient care depends on whether the marginal revenue from a patient is at least equal to the marginal cost of providing care, irrespective of the level of hospital market competition. In a segmented market with private pay and Medicaid patients, profit-maximizing hospitals have an incentive to serve Medicaid patients as long as the marginal revenue from the Medicaid patient is greater than the marginal revenue from the private-pay patient and more than marginal cost. A. Nonprofit Hospital Behavior Nonprofit hospital behavior can be distinguished from that of the for-profit hospital by the legal requirements which define tax-exempt status, specifically the community benefit standard and the private inurement constraint. A nonprofit hospital may choose to provide various benefits to the community other than the provision of under-reimbursed care for Medicaid patients. For instance, a nonprofit hospital may support health education programs or provide free care without the expectation of payment. Nonprofit hospitals can

Hospital Competition on Nonprofit and For-profit Hospital Medicaid Share

279

also be distinguished from for-profit hospitals by the financial advantages related to tax-exempt status, such as access to philanthropic contributions. It is not known whether the revocation of tax-exempt status is a credible threat to nonprofit hospitals which do not provide a significant benefit to the community. The IRS and the State courts have avoided establishing an acrossthe-board quantification of community benefit, because the interpretation has been vague and widely varied. Since 1969 the interpretation of federal tax policy has been somewhat controversial as to whether the provision of hospital services to Medicaid patients should be required for hospital tax-exempt status. Prior to 1969 the Internal Revenue Service considered a hospital tax-exempt under Revenue Ruling 56–185, 1956 C. B. 202 if it “were operated to the extent of its financial ability for those not able to pay for the services rendered and not exclusively for those who are able and expected to pay.”14 After the introduction of Medicare and Medicaid in 1966, the IRS issued Revenue Ruling 69–545, adopting the community benefit standard for providing the charitable tax exemption. The ruling held that the promotion of health is, in itself, considered beneficial to the entire community. This ruling is worded ambiguously and does not mention Medicaid explicitly, although the ruling strongly implies that maintaining an emergency room open to all is a condition of exemption.15 Then in 1991, General Counsel Memorandum 39862 provided explicit IRS guidelines regarding the provision of Medicaid services in order for a hospital to maintain tax-exempt status. GCM 39862 requires IRS auditors to compare a nonprofit hospital’s proportion of Medicaid patients to the population of Medicaid beneficiaries in the hospital’s service area. State courts have varied widely in their interpretation of the community benefit standard as it relates to local property tax-exemption and may even be considered more aggressive about enforcement than the federal courts. Two states, Utah and Pennsylvania, have revoked a nonprofit hospital’s tax- exempt status, based on a narrow interpretation of community benefit as the provision of free care.16 While other states have not revoked tax-exempt status, a decision primarily based on a nonprofit hospital’s provision of under-reimbursed care through Medicaid and Medicare.17 Irrespective of how a nonprofit hospital attempts to justify its tax-exempt status, nonprofit hospital behavior is expected to be different from for-profit hospital behavior. In particular, it is expected that nonprofit hospitals serve a larger proportion of Medicaid patients than for-profit hospitals. Nonprofit hospitals also differ from for-profit hospitals in that they have access to philanthropy, which provides a significant alternative revenue source. This additional source of revenue creates another area for nonprofit hospital competition other than hospital price competition for the privately insured.

280

SHERI L. EISERT

Both nonprofit and for-profit hospitals receive lower prices for privately insured patients the more competitive the market. Therefore, with lower profits generated in markets with greater competition there is less surplus to subsidize below-cost reimbursement by Medicaid. On the other hand, nonprofit hospitals also compete with other nonprofit hospitals for philanthropic dollars. In markets with fewer nonprofit hospitals, there is less competition for philanthropic dollars in the market.18 Depending on the objectives of the nonprofit hospital, Medicaid share may increase or decrease. With an increase in the supply of philanthropic dollars, nonprofit hospitals have a greater financial ability to subsidize below-cost reimbursement from Medicaid patients. With fewer nonprofit hospitals in the market, nonprofit hospitals may have an incentive to serve more Medicaid patients. On the other hand, with fewer nonprofit hospitals in the market there is less incentive to distinguish themselves by providing a community benefit in order to receive philanthropic funding. Thus, nonprofit hospitals may serve fewer Medicaid patients with fewer nonprofit hospitals in the market. B. Factors Affecting a Hospital’s Decision to Accept Medicaid Patients The following factors are expected to influence a nonprofit or for-profit hospital’s decision to accept Medicaid patients, where both have an incentive to maximize profit. To the extent nonprofit hospitals behave differently than profit maximizing for-profit hospitals, due to adherence to a community benefit standard, it is expected that nonprofit hospitals will provide care to a greater percentage of Medicaid inpatients than for-profit hospitals. 1. Medicaid Reimbursement Rate and Medicaid Eligible: The higher the Medicaid reimbursement rate, the greater a hospital’s incentive to accept Medicaid patients, whether nonprofit or for-profit. Hospitals located in markets with a relatively high proportion of Medicaid eligible population would serve a greater proportion of Medicaid patients than a hospital located in an area with a relatively small proportion of Medicaid eligible population. 2. Presence of Government-Owned Hospitals: The presence of a governmentowned hospital within the nonprofit hospital’s service area would reduce community pressure regarding a nonprofit hospital’s provision of services to Medicaid patients, whether nonprofit or for-profit. 3. Hospital Input Quality: Variation in hospital input quality can also influence a hospital’s decision to accept Medicaid patients.19 Marginal cost would not

Hospital Competition on Nonprofit and For-profit Hospital Medicaid Share

281

change for quality improvements requiring a fixed investment, independent of the number of inpatient days, such as a MRI. Marginal costs would increase for those quality improvements that are dependent on the number of inpatient days, such as an improvement in the food served or an increase in the wages for highly trained medical staff. With an increase in hospital quality independent of output, the profit-maximizing hospital will increase its market share for private pay patients and decrease its market share for government-insured patients. But, when a hospital increases quality that is dependent on output, the profitmaximizing hospital will still increase its market share for private pay-patients, although the decrease in share of government-insured patients will be relatively greater than when quality increase is independent of output.20 4. Market Concentration: The hospital in the less concentrated (more competitive) market charges a lower price to private-pay patients and thus has more paying patients than a hospital in a concentrated market. With more paying patients paying a price close to marginal cost, the hospital cannot afford to serve as many Medicaid patients in a competitive market as compared to a concentrated market. Therefore, as hospital competition increases, a hospital’s Medicaid share will decrease, ceteris paribus. 5. Market Occupancy Rate: A low market occupancy rate may be an indicator of intense market competition for private-pay patients which decreases hospital Medicaid share and increases hospital private-pay share. Conversely, a low hospital occupancy rate is associated with low marginal cost and may create incentives for hospitals to accept Medicaid patients. The effect of the market occupancy rate on a hospital’s Medicaid share is therefore considered ambiguous. 6. Presence of For-Profit Hospitals in Market: The greater presence of forprofit hospitals (greater nonprofit hospital competition) may increase or decrease a nonprofit hospital’s Medicaid share. It is expected that nonprofit hospitals, located in markets with a relatively large number of nonprofit hospitals compared to for-profit hospitals, will experience more competition for philanthropic funding. This will provide nonprofit hospitals an incentive to distinguish themselves from other nonprofits by providing a greater share of services to Medicaid patients. On the other hand, with relatively more nonprofit hospitals in the market, a nonprofit hospital has less need to provide services to Medicaid patients. 7. HMO Market Penetration: It is assumed that the greater the degree of HMO penetration the more competitive the market. As competition increases, a

282

SHERI L. EISERT

hospital’s share of private-pay patients increases and its share of Medicaid patients decreases.

III. METHODS A hospital’s Medicaid share, measured by the percent inpatient days that are Medicaid, is a function of Medicaid cost reimbursement, Medicaid market demand, market competition and hospital input quality. If nonprofit hospital behavior cannot be distinguished from that of for-profit hospitals, each type of hospital will provide similar shares of inpatient hospital services to Medicaid patients. If the nonprofit hospital is influenced by the community benefit standard it is expected that it will provide a greater share of inpatient services to Medicaid patients than for-profit hospitals. If nonprofit hospitals behave like profit maximizing for-profit hospitals it is also expected that the above factors will have a similar influence on the hospital’s decision to accept Medicaid patients. The empirical analysis will first establish whether nonprofits provide a greater share of inpatient services to Medicaid patients than for-profits. Then it will be determined whether there is a difference between nonprofit and for-profit hospitals, concerning the influence of the above factors on a hospital’s inpatient Medicaid share. To test these hypotheses, an interactive dummy regression equation is specified which allows for direct comparison of nonprofit and for-profit hospital behavior. The interactive dummy variable approach has the distinct advantage in that it indicates the difference in the regression coefficients when a condition is met. In this case, the dummy variable equals 1 if the hospital is nonprofit and 0 if for-profit. The intercept will test the hypothesis that nonprofit hospitals provide a larger share of Medicaid inpatient days that for-profits. The remaining interactive dummy variables will indicate whether there is a significant difference between nonprofit and for-profit hospital behavior. A. Sample and Data Hospital-specific data for this study are from the 1991 American Hospital Survey Annual Survey of Hospitals by the American Hospital Association (AHA) and the 1991 Medicare Minimum Cost Report by the Health Care Financing Administration (HCFA). The AHA hospital survey includes hospital service, expense and utilization data for over 90% of the hospitals in the United States; and the HCFA includes hospital revenue data and other balance sheet information for all Medicare participating hospitals in the U.S. Demographic characteristics by metropolitan statistical area, from the 1990 Census, are also

Hospital Competition on Nonprofit and For-profit Hospital Medicaid Share

283

included in the analysis, in addition to state level insurance information from the Prospective Payment Assessment Commission (ProPAC), Group Health Association of America (GHAA) and InterStudy Center for Managed Care Research. Both hospital level data sets were merged by hospital Medicare identification number, resulting in a data set representing approximately 6,800 hospitals in the U.S.. Since the AHA data is from a survey of hospitals with a response rate of about 90%, and the HCFA data set includes all Medicare and Medicaid participating hospitals, the AHA data set had fewer hospitals than the HCFA data set. Of the hospitals that did not report to AHA, the lowest response rate was from hospitals with fewer than 24 beds and from for-profit hospitals. The resulting number of excluded hospitals from the sample was few and from small hospitals, so no significant bias is expected to result from this merging. Since hospital market behavior is of primary interest, hospitals were then selected based on similar product and geographic markets. The relevant product market is defined as hospital inpatient services provided by community acute care hospitals. These hospitals accounted for about 75% of all hospitals in the U.S. in 1991. This definition of the product market excludes veterans’ hospitals, hospitals that specialize in one type of care, and hospitals that are units of institutions. Outpatient care is also excluded since there are substitutes available for this type of care provided outside of the hospital environment. The geographic market includes urban community acute care hospitals only, which are defined as those hospitals located in metropolitan statistical areas (MSAs). Although rural hospitals could be included to add a monopolistic market dimension to the analysis, rural hospitals generally provide a narrower range of services than urban hospitals. Rural hospitals also are generally smaller and may transfer patients to urban hospitals because a service is not available, not because they are influenced by profit-maximizing factors. Monopolistic hospital behavior is included in the empirical analysis without including rural hospitals since 38 MSAs have only one hospital. Large MSAs, which are defined as having populations greater than one million, are also excluded from the study. These MSAs, which include such cities as Los Angeles and New York City, are considered too large to be defined as one market. When hospitals were located outside the major city of the MSA, a geographic map was consulted to limit the hospitals defined in a geographic market to approximately a 25 mile radius from the major city. The empirical analysis includes 899 nonprofit, 192 for-profit and 221 government-owned community acute care hospitals located in 289 of the 341 metropolitan statistical areas, representing all 50 states in the U.S., but excluding

284

SHERI L. EISERT

Washington D.C.. For-profit hospitals are represented in 103 of the MSAs, nonprofits in 270 MSAs and government-owned in 131 of the MSAs.21 B. Model Variables The dependent variable is the percent of a hospital’s inpatient days that are Medicaid inpatient days. Only nonprofit and for-profit hospitals are included as dependent variable observations, while government-owned hospitals are included in the model as an independent variable to the extent they represent hospitals of last resort. Government-owned hospitals are also included in the calculation of the market concentration index. Table 1 lists the variables Table 1. Variables in Multivariate Model Variable Dependent Variable: MAID Explanatory Variables: Pm Demand POV EL POP EMER TEACH GOVBD Market Competition HHI HMO FPBD MKTOCC Input Quality RN OBS MRI OH SIZE

Description

Exp. Sign Level

% Medicaid Inpatient Hospital Days

Hospital

Medicaid reimbursement rate.

+

State

Percent MSA population in poverty. Percent state population Medicaid eligible. MSA Population Does hospital operate an emergency room? (dummy = 1 for yes) Is the hospital a teaching hospital? (dummy = 1 for yes) Government-owned hospital beds per capita.

+ + + ?

MSA State MSA Hospital

?

Hospital

-

MSA

Herfindahl Hirschman Index. HMO penetration rate. For-profit hospital beds per capita. Market bed occupancy rate.

+ ? ?

MSA State MSA MSA

FTE/Total hospital inpatient days. Does hospital provide services for all complicated cases in obstetric unit? (dummy = 1 for yes) Does hospital provide hospital-based MRI services? (dummy = 1 for yes) Does the hospital provide open-heart surgery? (dummy = 1 for yes) Hospital beds set up and staffed

?

Hospital Hospital

-

Hospital

-

Hospital

-

Hospital

Hospital Competition on Nonprofit and For-profit Hospital Medicaid Share

285

utilized in the empirical model and the expected relationship with the dependent variable. Table 2 provides the descriptive statistics. The independent variables include a Medicaid price variable, a vector of Medicaid demand variables, market competition variables and input quality variables. The price variable for Medicaid patients is defined by the Medicaid reimbursement rate. This rate is calculated as hospital payments as a percentage

Table 2. Descriptive Statistics Nonprofit (N = 899) Mean

S.D.

Mean

Dependent Variable: MAID 12.7% 10.3% Explanatory Variables: Pm 81.2% 11.7% Demand POV 12.3% 4.4% EL 10.4% 2.3% POP 416,110 272,192 EMER 94.2% TEACH

MRI OH SIZE

32.6% 33.4% 264

Gov.-Owned (N = 221)

Total Hospitals (N = 1,312)

S.D.

Mean

S.D.

Mean

S.D.

9.5%

7.7%

20.1%

15.6%

13.5%

11.5%

79.3%

9.0%

80.2%

9.3%

80.8%

11.0%

15.2% 10.5% 453,022

6.1% 2.5% 277,895

15.1% 11.0% 409,824

4.7% 2.4% 269,729

13.2% 10.5% 420,453 90.9%

4.9% 2.3% 272,756

76.0% 1.0%

7.2%

GOVBD 0.00035 Market Competition HHI 3308 HMO 13.0% FPBD 0.00026 MKTOCC 66.8% Input Quality RN 0.00418 OBS 12.2%

For-Profit (N = 192)

90.5% 12.7%

0.00063

0.00091

0.00115

0.00143

0.00117

7.2% 0.00062

0.00093

1887 7.7% 0.00049 8.9%

2854 9.4% 0.00117 62.5%

1425 7.4% 0.00078 7.1%

3381 11.4% 0.00055 64.4%

1933 9.3% 0.00068 7.8%

3254 12.2% 0.00044 65.8%

1841 8.0% 0.00066 8.6%

0.00351

0.00479

0.00755

0.00398

0.00254

0.00424

0.00423

179

2.6% 22.4% 20.3% 164

116

15.4% 32.1% 25.8% 203

177

11.4% 31.0% 30.2% 240

175

Note: For the dummy variables, numbers represent percent of hospitals. Government-owned hospitals are included in the descriptive statistics for informational purposes only and are not included as dependent variable observations. Government-owned hospitals are specified in the empirical model only as independent variables, one representing government beds per capita in the market and the other as part of the market share measure in the HHI calculation.

286

SHERI L. EISERT

of costs for Medicaid patients, at the state level.22 Even though Medicaid price data is not available at the hospital level and is not distinguished by inpatient and outpatient care, there exists significant state level variation in Medicaid reimbursement rates to capture the effects on hospital behavior.23 Medicaid demand variables include the percent state population Medicaid eligible and the percent population below poverty within a hospital’s MSA.24 The total population within an MSA, which measures market size, is also included to control for any influence larger cities may have on the pool of Medicaid eligibles. The presence of government-owned hospitals in the market is also expected to have an impact on a hospital’s decision to serve Medicaid patients. Whether or not a hospital operates an emergency room is expected to have an effect on hospital Medicaid demand. A hospital emergency room is a point of entry for the poor and uninsured to receive emergency and non-emergency care. Hospitals that provide this emergency room service have a greater probability of serving Medicaid patients. On the other hand, hospitals that operate emergency rooms may be of higher quality than those that do not. Therefore, the expected net-effect of this variable on hospital Medicaid share is unknown since the above explanations have opposite effects on hospital Medicaid share. Whether or not a hospital is a teaching hospital is also included as a demand variable. Teaching hospitals differ from other community acute care hospitals in that they may receive a greater proportion of their revenues from philanthropy and that they focus on medical research and the application of advanced technology. Since these hospitals also train physicians, they may also have more patients without physician attachments for residents to work on. A larger share of philanthropic contributions may or may not have an effect on hospital Medicaid share. These contributions would provide the hospital with greater flexibility in serving Medicaid patients; on the other hand these contributions may be earmarked for a specific purpose other than serving Medicaid patients. It could also be argued that a teaching hospital is a signal of high quality where it is expected that higher hospital quality reduces a hospital’s share of Medicaid inpatient days. It is expected that the more price competitive the market, the smaller a hospital’s Medicaid share. The variables include a market concentration index, degree of HMO penetration and the market occupancy rate. The market concentration index (HHI) is a proxy for the degree of hospital price competition, where the hospital’s market share is measured by the hospital’s percent of inpatient revenue relative to the total market. The smaller the HHI, the less concentrated the market and the greater the hospital market

Hospital Competition on Nonprofit and For-profit Hospital Medicaid Share

287

competition. It is also expected that the greater the degree of HMO penetration, the greater the market competition. The greater the presence of for-profit hospitals in the market (fewer nonprofit hospitals) is included as a nonprofit hospital competition variable. Nonprofit hospitals may increase or decrease their Medicaid share the greater the presence of for-profit hospitals in the market.25 The market hospital bed occupancy rate or capacity utilization is also considered an indicator of hospital competition. The lower the market bed occupancy rate, the greater the competition in the market to fill those beds and the smaller a hospital’s Medicaid share. On the other hand, for those markets with very low occupancy it may be profitable for a hospital to admit Medicaid patients where marginal costs would be low relative to the Medicaid reimbursement rate. With these two opposing effects the net influence of this variable is uncertain. Hospital quality is also expected to have an influence on a hospital’s decision to serve Medicaid patients. Hospital quality is characterized by inputs into production rather than outputs to production (mortality, infections, readmissions, etc.) since hospital outputs are sensitive to differences in hospital patient case-mix and to the differences among patients in compliance with doctor’s orders. The hospital is assumed to be unable to change general hospital quality in the short run. The prevalence of labor contracts and the long term process involved with capital acquisition limits hospital flexibility in changing hospital quality in the short run. Indicators of hospital quality included in the empirical model are RN staffing levels, level of obstetric services, whether a hospital operates a MRI or provides open-heart surgery, and hospital bed size. Some may argue that hospital bed size is a poor quality indicator. This measure was included in the model as a proxy for the scope of services offered by a hospital where the broader the scope of services offered, the higher the quality of hospital. Also, many other variables may be considered measures of hospital input quality but these variables were chosen as proxies for indicators of general hospital quality since they are considered indicators which are familiar and observable to consumers. In general, hospitals of higher quality are expected to serve a smaller share of Medicaid patients than hospitals of lower quality. Of these hospital quality indicators, the level of obstetric care may or may not have a negative influence on a hospital’s decision to accept Medicaid patients. The general population value high quality obstetric care and the general public are also more familiar with this type of service relative to other services provided by a hospital. On the other hand, it may also be argued that the provision of high- quality obstetric care is a demand variable since in 1991,

288

SHERI L. EISERT

32% of all births in the United States were Medicaid births.26 The empirical results, if significant, will reveal which one of these interpretations dominates since an increase in Medicaid demand is expected to have a positive impact on a hospital’s Medicaid share and an increase in hospital quality is expected to have a negative impact on a hospital’s Medicaid share.

IV. EMPIRICAL RESULTS Table 3 contains the regression results for the dependent variable MAID, a hospital’s percent inpatient days that are Medicaid. The insignificant interactive dummy coefficients were eliminated since leaving insignificant interactive dummies in the equation distorts the parameter estimates for the relevant noninteractive variables. The null hypothesis that for-profits and nonprofits provide the same relative share of inpatient care to Medicaid patients can be rejected. Nonprofit hospitals provide a significantly greater share of inpatient hospital services to Medicaid patients than for-profit hospitals; about 8% more of nonprofit inpatients days are attributed to Medicaid patients. This result indicates that the community benefit standard has a significant effect on nonprofit hospital behavior. When interpreting the interactive dummy coefficients it is important to keep in mind that nonprofits already provide a greater share of hospital services to Medicaid patients than for-profit hospitals, therefore there is a difference in nonprofit and for-profit hospital behavior. The differences lie not only in hospital share of Medicaid inpatient days but also the degree to which the presence of for-profit hospitals and the ratio of RNs to inpatient days have an impact on a hospital’s decision to serve Medicaid patients. Nonprofit hospitals will serve fewer Medicaid patients the greater the presence of for-profit hospitals in the market and the greater the ratio of RNs to inpatient days (higher quality). The greater the presence of for-profit hospitals, or the relatively fewer nonprofits in the market, the smaller the nonprofit hospital Medicaid share. Therefore, nonprofit hospitals are less motivated to provide services to Medicaid patients when there is less competition for philanthropic contributions in the market. The higher quality nonprofit hospital also has a smaller share of Medicaid patients than the lower quality nonprofit hospital where the community benefit standard has less influence over their decision to accept Medicaid patients. The presence of for-profit hospitals and the ratio of RNs to inpatient days do not have a significant effect on for-profit hospital behavior. The for-profit

Hospital Competition on Nonprofit and For-profit Hospital Medicaid Share

289

Table 3. Regression Results: Dependent Variable = Percent of Hospital Medicaid Inpatient Days (MAID) Coefficient Estimate

Description Intercept Pm

Medicaid reimbursement rate.

POV

Percent MSA population in poverty.

EL

Percent state population Medicaid eligible.

POP

Census population of MSA

EMER TEACH

Does hospital operate an emergency room? (dummy = 1 for yes) Is the hospital a teaching hospital? (dummy = 1 for yes)

GOVBD

Government-owned hospital beds per capita.

HHI

Herfindahl Hirschman Index.

HMO

HMO penetration rate.

FPBD

For-profit hospital beds per capita.

MKTOCC

Market bed occupancy rate.

RN

RN FTE/Total hospital inpatient days.

OBS

IZES

Does hospital provide services for all complicated cases in obstetric unit? (dummy = 1 for yes) Does hospital provide hospital-based MRI services? (dummy = 1 for yes) Does the hospital provide open-heart surgery? (dummy = 1 for yes) Hospital beds set up and staffed

D

D = 1 if nonprofit.

MRI OH

DFPBD DRN

Number of Hospitals: 1,091 (899 NFP, 192 FP) Adjusted R-squared: 0.13 F Value: 9.67 *5% Significance *Statistically significant at the 1 percent level. **Statistically significant at the 5 percent level.

0.036 (0.951) –0.039 (–1.435) 0.461* (6.994) 0.361* (2.823) 0.000000004 (0.264) –0.030* (–2.889) 0.023 (1.678) –13.09* (–3.287) –0.00000001 (–0.004) 0.054 (1.279) 10.72 (1.191) 0.017 (0.437) –0.245 (–0.272) 0.046* (4.362) –0.004 (–0.542) –0.027* (–3.396) –0.00004** (–1.755) 0.083* (5.508) –24.8* (–2.228) –5.42* (–4.258)

**10% Significance

290

SHERI L. EISERT

hospital is not as sensitive to some of the factors which have a negative influence on nonprofit Medicaid share, although for-profit hospitals already provide a smaller share of inpatient services to Medicaid patients. The presence of for-profit hospitals can be interpreted as a measure of the degree of nonprofit hospital market competition, therefore the insignificant result. The results also indicate there may be similarities between nonprofit and forprofit behavior, although insignificant dummy coefficients should be interpreted with caution. The market poverty rate and the percent state population Medicaid eligible have a similar significant positive influence on the Medicaid shares for both organizational forms, as expected. Also, the presence of government-owned hospitals reduces the Medicaid share for both nonprofit and for-profits and the degree of this effect is similar. The other demand factor, whether a hospital operates an emergency room, maintains the unexpected result, for both nonprofit and for-profit hospitals. Hospitals that operate emergency rooms serve a smaller share of Medicaid patients than hospitals that do not operate emergency rooms. Emergency rooms may be an indicator of high hospital quality rather than a Medicaid demand variable. Another explanation may be that the provision of emergency room services is a substitute for the provision of Medicaid services since the 1969 IRS ruling considers the hospital operation of an emergency room as one way of providing a benefit to the community. The price competition variables did not significantly influence either the nonprofit or for-profit hospitals Medicaid share. The nonprofit hospital’s decision to accept Medicaid patients is only influenced by one of the market competition variables, the presence of for-profit hospitals in the market. The less competition there is among nonprofit hospitals (greater for-profit hospital presence), the less incentive there is for these hospitals to serve Medicaid patients in order to receive funding from philanthropy. Therefore, the results indicate that conventional measures of price competition do not influence a hospital’s decision to accept Medicaid patients. For-profit hospitals already provide care to a smaller share of Medicaid patients than nonprofit hospitals and the level of market price competition does not affect this share. Three of the hospital quality variables have a similar significant effect on nonprofit and for-profit hospital behavior. Hospitals that perform open-heart surgery and hospitals of larger bed size serve a smaller share of Medicaid patients than hospitals that do not perform open-heart surgery and hospitals of smaller bed size. Higher quality hospitals, whether nonprofit of for-profit, serve a smaller share of Medicaid patients. The higher the level of obstetric care, the larger a hospital’s Medicaid share, whether nonprofit or for-profit. This

Hospital Competition on Nonprofit and For-profit Hospital Medicaid Share

291

indicates that a high level of obstetric care is more closely related to Medicaid demand than an indicator of hospital quality.

V. DISCUSSION Many studies have shown nonprofit hospitals are not providing sufficient amounts of charity care to justify their tax-exempt status. Some studies have even found that for-profit hospitals provide more charity care than nonprofit hospitals. This study, which takes the definition of charity care a step beyond the interpretation of free care, concludes that nonprofit hospitals are providing care to a larger share of Medicaid patients than for-profits, indicating that they are influenced by the community benefit standard. Nonprofit hospitals may accept a larger share of Medicaid patients than for-profit hospitals but for-profit hospitals still serve Medicaid patients. Medicaid patients can be profitable on the margin even when the reimbursement rate is less-than-average-cost. Even though nonprofit hospitals may be adhering to a community benefit standard by providing more of their services to Medicaid patients than forprofit hospitals, nonprofit hospitals may look more like for-profit hospitals when considering other factors that are expected to have an effect on a hospital’s decision to accept Medicaid patients. The higher the quality of nonprofit hospital inputs, the smaller the nonprofit hospital Medicaid share, which is similar to for-profit hospital behavior. Nonprofit hospitals slough off Medicaid patients to government-owned hospitals when there is an opportunity to do so, just like for-profit hospitals. Although, the nonprofit hospital may reduce its Medicaid share in the presence of government-owned hospitals, this does not necessarily imply that nonprofit hospitals reduce their benefit to the community. Nonprofit hospitals may substitute out of Medicaid services into other types of benefits to the community in the presence of government-owned hospitals Nonprofit hospitals also reduce their share of Medicaid patients the greater the presence of for-profit hospitals in the market. Nonprofit hospitals have less incentive to provide below-cost reimbursed care when there are few nonprofit hospitals in the market competing for philanthropic funding. The loss of the philanthropic dollar may have more influence on providing a benefit to the community than the threat of revocation of tax-exempt status. According to this study, market price competition does not influence a hospital’s decision to accept Medicaid patients, whether nonprofit or for-profit. There are other factors that hospitals consider when accepting Medicaid patients.

292

SHERI L. EISERT

NOTES 1. Newhouse (1970), and Frank & Salkever (1991). 2. Preston (1988). 3. Easley & O’Hara (1982). 4. Pope (1989). 5. According to data collected by the Prospective Payment Assessment Commission the average hospital payment as percentage of hospital costs was 79 percent in 1991, the relevant time frame of this study ProPac (1993). Only four states experienced reimbursement rates of more than 100% of average costs. The lowest reimbursement rate was in Illinois at 56% of average hospital costs. 6. For a more extensive discussion on the legal evolution of the interpretation of hospital community benefit, see Fox & Schaffer (1991), Hall & Colombo (1991) and Hyman (1991). 7. GAO (May 1990). 8. Fishman, P. & Mariger, R. (1992). 9. Pattison, R. & Katz, H. (1983). 10. Watt, M. et al. (1986). 11. Robinson & Luft (1985) and Pope (1989). 12. Zwanziger & Melnick (1988), Melnick et al. (1992), Dranove & Satterwaite (1992) Dranove et al. (1993). 13. Although, with the recent increase in Medicaid managed care contracts and variation in state Medicaid risk adjusted premiums, market competition is beginning to play a significant role in the pricing of government insurance. 14. Potter & Longest (1994). 15. Fax & Schaffer (1991). 16. Utah County v. Intermountain Health Care, 709 P2d 265,268 (Utah, 1985) and School District of the City of Erie v. Hamot Medical Center, 602 A2d 407 (Pa. Cmwlth., 1992). 17. Calloway Community Hospital Association v. Ronald Craighead, 759 SW2d 253 (MO. App., 1988) and Downtown Hospital Association v. Tennessee State Board of Equalization, 760 SW2d 954 (Tenn.App., 1988). 18. See Frank & Salkever (1991) and Schlesinger et al. (1997) for a discussion describing how nonprofit hospitals may either provide fewer or more services to the uninsured in markets with greater competition. 19. Hospital input quality is distinguished from hospital output quality since measures of hospital output quality, such as mortality and morbidity, are not necessarily directly tied to the behavior of the hospital. For instance, mortality and morbidity rates are also related to the demographic characteristics of the population. 20. When the demand curve shifts to the right, the marginal revenue will also shift to the right, for an increase in hospital quality whether it is dependent or independent of output. The difference lies in the fact that when the increase in quality is dependent on output, the marginal cost-curve shifts up and left. This causes the further reduction in the number of government inpatient days relative to when there is an increase in hospital quality independent of output.

Hospital Competition on Nonprofit and For-profit Hospital Medicaid Share

293

21. During 1991, for-profit hospitals did not provide services in Washington D. C. nor in the following nine states: Connecticut, Delaware, Michigan, Minnesota, North Dakota, Rhode Island, South Dakota, Vermont and Wisconsin. 22. Medicaid payments as a percentage of costs were calculated from a ProPAC analysis of Annual Survey data from the American Hospital Association published in ProPAC’s June 1993 ‘Report to the Congress’. 23. The empirical model was estimated with state fixed effects, using dummy variables for regions. Neither of these specifications changed the empirical results. 24. The correlation between the percent of state population below poverty and the percent state population Medicaid eligible is 0.45 and is significantly different from zero at the p = 0.001 level. 25. These four measures of competition were tested for multicollinearity. The correlation coefficients ranged from 0.05 to 0.30 in absolute value. Therefore, multicollinearity between these four variables is not a significant problem. These variables can be interpreted as four different approaches to measuring competition. 26. Estimated percent of Medicaid births for 1991 from Frost et al., ‘State Implementation of the Medicaid Eligibility Expansions for Pregnant Women’, Alan Guttmacher Institute, New York, 1993, as cited in GAO Report HEHS–94–152BR on Medicaid Prenatal Care.

REFERENCES Dranove, D., & Satterwaite, M. (1992). Monopolistic Competition When Price and Quality are Difficult to Observe. RAND Journal of Economics, 23, 518–534. Dranove, D., Shanley, M., & White, W. D. (1993). Price and Concentration in Hospital Markets: The Switch From Patient Driven to Payer Driven Competition. Journal of Law and Economics, 36, 179–204. Easley, D., & O’Hara, M. (1982). The Economic Role of the Nonprofit Firm. The Bell Journal of Economics, 13, 531–538. Fishman, P., & Mariger, R. (1992). Do Nonprofits Provide More Subsidized Health Care than ForProfit Hospitals? Working paper. Fox, D., & Schaffer, D. (1991). Tax Administration as Health Policy: Hospital, the Internal Revenue Service, and the Courts. Journal of Health Policy, Politics and Law, 16, 251–279. Frank, R., & Salkever, D. (1991). The Supply of Charity Services by Nonprofit Hospitals: Motives and Market Structure. RAND Journal of Economics, 22, 430–445. General Accounting Office (1994). Medicaid Prenatal Care. HEHS–94–152BR. General Accounting Office (1990). Nonprofit Hospitals: Better Standards Needed for Tax Exemption. GAO/HRD–90–84. Hall, M. A., & Colombo, J. (1991). The Charitable Status of Nonprofit Hospitals: Toward a Donative Theory of Tax Exemption. Washington Law Review, 66, 307–411. Hyman, D. A. (1991). The Conundrum of Charitability: Reassessing Tax Exemption for Hospitals. American Journal of Law and Medicine, 16, 327–380. Melnick, G., Zwanziger, J., Bamezai, A., & Pattison, R. (1992). The Effects of Market Structure and Bargaining Position on Hospital Prices. Journal of Health Economics, 11 217–233. Newhouse, J. (1970). Toward a Theory of Nonprofit Institutions: An Economic Model of a Hospital. American Economic Review, 60, 64–74.

294

SHERI L. EISERT

Pattison, R., & Katz, H. (1983). Investor-Owned and Not-For-Profit Hospitals: A Comparison Based on California Data. The New England Journal of Medicine, 309, 347–353. Pope, G. (1989). Hospital Nonprice Competition and Medicare Reimbursement Policy. Journal of Health Economics, 8, 147–172. Potter, M., & Longest, B. (1994). The Divergence of Federal and State Pollicies on the Charitable Tax Exemption of Nonprofit Hospitals. Journal of Health Policy, Politics and Law, 19, 393. Preston, A. (1988). The Nonprofit Firm: A Potential Solution to Inherent Market Failures. Economic Inquiry, 26, 493–506. Prospective Payment Assessment Commission (ProPAC) (1992 and 1992). Medicare and the American Health Care System: Report to the Congress. Robinson, J., & Luft, H. (1985). The Impact of Hospital Market Structure on Patient Volume, Average Length of Stay, and the Cost of Care. Journal of Health Economics, 4, 333–356. Schlesinger, M., Dorwart, R., Hoover, C., & Epstein, S. (1997). Competition, Ownership, and Access to Hospital Services: Evidence from Psychiatric Hospitals. Medical Care, 35 974–992. Thorpe, K., & Phelps, C. (1991). The Social Role of Not-For-Profit Organizations: Hospital Provision of Charity Care. Economic Inquiry, 29, 472–484. Watt, M., Derzon, R., Renn, S., Schramm, C., Hahn, J., & Pillari, G. (1986). The Comparative Economic Performance of Investor-Owned Chain and Not-For-Profit Hospitals. The New England Journal of Medicine, 314, 89–97. Zwanziger, J., & Melnick, J. (1988). The Effects of Hospital Competition and the Medicare PPS Program on Hospital Cost Behavior in California. Journal of Health Economics, 7, 301–320.