1 There is an obvious connection between this sentence and the previous one. As S. J. Perelman put it, “There's nothing like a good, painstaking survey full of decimal points and guarded generalizations to put a glaze like a Sung vase on your eyeball” (Keep It Crisp 173 (New York: Random House, 1946)). Epstein may cause steam to come out of your ears, but your eyes will certainly not glaze over when reading him there are virtually no decimal points in Forbidden Grounds, and the generalizations are anything but guarded.Google Scholar

2 Glen Cain, “The Economic Analysis of Labor Market Discrimination: A Survey,”in O. Ashenfelter & R. Layard, eds., 1 Handbook of Labor Economics (Amsterdam: Elsevier Scientific, 1986) (“Cain, ‘Economic Analysis’”).CrossRefGoogle Scholar

3 Writing in the late 1960s, Becker noted:Google Scholar

A few of the more extreme nineteenth-century advocates of a competitive market economy believed that eventually its extension and development would eliminate most economic discrimination…. Unfortunately, this has not yet taken place; discrimination exists, and at times even flourishes, in competitive economies, the position of Negroes in the United States being a clear example.Google Scholar

“Discrimination”in International Encyclopedia of the Social Sciences 210 (1968). Had this not been written 25 years ago, it could be taken as a direct slap at Epstein.Google Scholar

4 Epstein scarcely mentions the North, where Jim Crow laws were insignificant. His analysis would seem to require that in the absence of such laws there should be virtually no discrimination. Yet in the two decades between the end of World War II and the passage of the 1964 Civil Rights Act, most Northern states felt discrimination against blacks by private employers was enough of a problem to warrant adoption of fair employment practice laws designed to prohibit it.Google Scholar

5 At several points, Epstein suggests that at least until the passage of Title VII and the Voting Rights Act, local governments used their regulatory or “police” powers ranging from control over sewer hookups to zoning and building code enforcement to reinforce (or substitute for) direct mandates of discrimination. These powers, Epstein asserts, constituted “the vehicle that allowed local governments to trample the ordinary rights of property and contract, which were as valuable to blacks as they were to whites” (at 115).Google Scholar

6 Chapter 8 of the book reads as if Epstein, a long-time defender of the “at will” contract in employment (giving the employer the right to terminate the employee for any reason, or no reason), just couldn't resist the chance to deploy his arguments yet again. For a fuller exposition, see Epstein, , “In Defense of the Contract at Will,” 51 U. Chi. L. Rev. 947 (1984).CrossRefGoogle Scholar

The discussion has relatively little relevance to issues of race and sex discrimination. True, ruling out race and sex as legitimate grounds for dismissal was a first step in eroding the at will contract. But that point could be made in a single sentence, and certainly doesn't require an entire chapter.Google Scholar

Moreover, Epstein's analysis of the at will contract is flawed. For more perceptive treatments of these issues, the interested reader should consult Lazear, Edward, “Job Security Provisions and Employment,” 105 QJ. Econ. 699 (1990); and Levine, David, “Just Cause Employment Policies in the Presence of Worker Adverse Selection,” 9 J. Lab. Econ. 294 (1990). One key argument in favor of the at will contract is that if both workers and employers would rather have a just cause rule, they are free to write it into the employment contract. However, Levine's model demonstrates that if workers know more about their own productivity than employers do, a worker who asks for just cause protection will signal to her employer that she believes she is likely to be fired. To avoid sending such a signal, workers may be forced to accept the at will rule when both workers and employers would prefer just cause contracts (with appropriate side payments). That is, even if workers would be willing to concede more in wages to get a just cause rule than employers would require to give up their power to fire at will, workers' desire to avoid signaling low productivity may prevent this beneficial agreement from being reached.CrossRefGoogle Scholar

7 And why Hobbes, who, after all, believed that we needed a strong state a Leviathan to prevent the war of all against all? Doesn't this seem like a rather strange starting point for a libertarian? I'm grateful to Richard McAdams for this point. His perceptive discussion of Epstein's misreading of Hobbes especially given the latter's emphasis on honor and dignitary harms as causes of social turmoil that necessitate a strong state appears in the San Diego Law Review symposium on Forbidden Grounds, 31 San Diego L. Rev., (1994).Google Scholar

8 Sociologist Christopher Jencks makes the case for an empirical component in public policy judgments very well:Google Scholar

Some ideological differences are simply a matter of conflicting values.… In cases of this kind, making the same value judgment in every situation, regardless of context, seems to make sense. But most ideological arguments depend on facts as well as values…. When facts matter, applying the same principles to every situation leads to foolishness…. We know that many people respond to economic incentives. All else equal, therefore, sensible people expect that raising AFDC benefits will encourage both out-of-wedlock births and family breakups. But that general conclusion is of no use in a debate over welfare policy. In such a debate the important question is how many additional fatherless families we will have if we raise AFDC by, say, $100 a month. If the number is very small, we can afford to ignore the problem. If it is very large, even hard-core liberals will want to consider other approaches to helping such families. We cannot resolve quantitative questions of this kind by invoking general principles.Google Scholar

Rethinking Social Policy 12 (Cambridge, Mass.: Harvard University Press, 1992).CrossRefGoogle Scholar

9 Personal communication.Google Scholar

10 Writing in 1962, before the passage of Title VII, Milton Friedman offered an even starker view of this structural similarity:Google Scholar

[Antidiscrimination] legislation involves the acceptance of a principle that proponents would find abhorrent in almost every other application. If it is appropriate for the state to say that individuals may not discriminate in employment because of color or race or religion, then it is equally appropriate for the state, provided a majority can be found to vote that way, to say that individuals must discriminate in employment on the basis of color, race or religion. The Hitler Nuremberg laws and the laws in Southern states imposing special disabilities upon Negroes are both examples of laws similar in principle to [state Fair Employment Practice Commissions].Google Scholar

Capitalism and Freedom 113 (Chicago: University of Chicago Press, 1962) (“Friedman, Capitalism and Freedom”).Google Scholar

11 Much of what I wanted to say on this topic turns out to have been covered more cogently by Robert Cooter, “Market Affirmative Action” (forthcoming in the San Diego Law Review symposium on Forbidden Grounds) (“Cooter, ‘Market Affirmative Action’”).Google Scholar

12 Economies of scale may make it possible for a single seller to produce all the output demanded by the market at a lower cost than any group of more than one seller. This kind of “natural monopoly” (as is allegedly the case in certain kinds of public utilities or railroads) is one setting in which economics suggests that governmental regulation of some kind may be appropriate to keep prices at a lower level than a monopolist would charge.Google Scholar

Public goods are distinguished from normal private goods in that the former are “nonrival” (when the government protects me from foreign attack, it does not reduce your consumption of national defense) and “nonexcludable” (there is no way to prevent me from benefiting from national defense expenditures, so I have no incentive to pay for them).Google Scholar

Finally, externalities occur when one person's behavior influences another's utility in ways that are not reflected in market transactions. The classic example is smoke from a factory polluting the air downwind that is breathed by others. If the air is treated as a free resource, factory owners see (virtually) no cost to dumping unlimited quantities of smoke into it. Regulation of some sort is often suggested as a means of curbing externalities of this kind.Google Scholar

13 “Market Affirmative Action” at 5.Google Scholar

14 “Is Title VII Efficient?” 134 U. Pa. L. Rev. 1411 (1986). Becker's model of employer animus which is the most widely used economic model of discrimination suggests that discriminators end up paying for indulging in their preferences to not associate with groups they don't like. Because they hire high-cost whites instead of lower-cost black workers, they earn lower profits than a nondiscriminator would. This means that nondiscriminatory employers will have a cost advantage over their discriminatory competitors. And as long as there are discriminatory employers around, production will not be efficient, because it will be possible for the nondiscriminatory employers to buy the assets of the discriminatory firms, end discrimination, and make a profit in doing so by lowering costs. Donohue's key insight is that discrimination is a disequilibrium phenomenon because it is inefficient; if Title VII can help speed up the transition to the efficient long-run equilibrium by getting discriminatory employers out of the market faster than they would otherwise choose to exit, then the law can increase efficiency.Google Scholar

15 “Advocacy versus Analysis in Assessing Employment Discrimination Law,” 44 Stan. L. Rev. 1583 (1992).CrossRefGoogle Scholar

16 A loose definition of statistical discrimination might be the use of an observable variable such as race or gender as a proxy for a relevant but unobservable variable like productivity, with which it is believed to be correlated.Google Scholar

17 See, e.g., Schwab, Stewart, “Is Statistical Discrimination Efficient 76 Am. Econ. Rev. 228 (1986); and Lundberg, Shelley & Startz, Richard, “Private Discrimination and Social Intervention in Competitive Labor Markets,” 73 Am. Econ. Rev. 340 (1983). In the Lundberg-Startz model, e.g., employers use stereotypes in making employment decisions and treat all minority applicants as low-productivity workers. Knowing they will be judged by the “average” or stereotype for their group and not their own educational achievement, minority workers correctly reason that additional education will not do them much good, and thus invest in less education. A law forbidding decisions on the basis of stereotypes, as Title VII arguably does, could thus increase personal investment in education and lead to a better outcome for both workers and employers.Google Scholar

18 This is a major theme of Cooter's article which provides specific examples in the context of employment discrimination and of Steven Rhoads's book The Economist's View of the World (Cambridge: Cambridge University Press, 1985).Google Scholar

19 In fact, Kenneth Arrow's famous impossibility theorem demonstrates that under reasonable conditions, there is no way to make such choices. See Arrow, Social Choice and Individual Values (New York: Wiley, 1951).Google Scholar

20 Amartya Sen, Collective. Choice and Social Welfare 200 (San Francisco: Holden-Day, 1970).Google Scholar

21 This is a loose paraphrase of the theory developed by Gary Becker in The Economics of Discrimination (Chicago: University of Chicago Press, 1957; 2d ed. 1971), the work that first used economic theory to address race discrimination. The key insight of Becker's theory is that employers must bear at least some of the costs of their discriminatory preferences when they substitute higher-cost white workers for equally productive and cheaper blacks. Kenneth Arrow succinctly summarized the problem with Becker's model, however: “it predicts the absence of the phenomenon it was designed to explain.” Arrow, “Some Mathematical Models of Race in the Labor Market,”in A. H. Pascal, ed., Racial Discrimination in Economic Life 192 (Lexington, Mass.: Lexington Books, 1972).Google Scholar

22 The most compelling piece of evidence on the economics of segregation in southern labor markets is the work of Heckman, James & Payner, Brook, “Determining the Impact of Federal Antidiscrimination Policy on the Economic Status of Blacks: A Study of South Carolina,” 79 Am. Econ. Rev. 138 (1989). Epstein suggests that the existence of a law barring blacks from certain kinds of employment in textiles explains these results. However, no such law existed in other southern states, yet the same patterns were observed elsewhere.Google Scholar

Epstein cites some evidence of business resistance to mandatory segregation laws (especially for railways and streetcars). But this evidence is itself questionable. Barbara Y. Welke, “‘All The Women Are White; All the Blacks Are Men,’ Or Are They? Law and Segregation on Common Carriers, 1855-1914” at 87-90 (American Bar Foundation Working Paper 9215, 1992), finds evidence that many common carriers did support mandatory segregation. What's more, Epstein does not offer good evidence that employers were opposed to segregation, as required by his theory, or that labor market discrimination was only sustained by government intervention.Google Scholar

23 Economists James Heckman and J. Hoult Verkerke claim that the “informal … Southern code was enforced primarily through social and economic pressure with the threat of private violence should less severe sanctions fail.”“Racial Disparity and Employment Discrimination Law: An Economic Perspective,” 8 Yale L. & Pol'y Rev. 276, 278 (1990), quoted in Epstein at 97.Google Scholar

24 The Highest Stage of White Supremacy: The Origins of Segregation in South Africa and the American South (Cambridge: Cambridge University Press, 1982).Google Scholar

25 He does cite a South Carolina law that required segregation in certain jobs in textile plants. But readers of this journal will not need to be told that there is a big difference between law on the books and law in action, and Epstein never moves from the former to even the most primitive version of the latter.Google Scholar

26 Alexis de Tocqueville, Democracy in America 357-59 (New York: Alfred A. Knopf, 1963). Tocqueville was of course writing about a period 70 or 80 years before the time Epstein discusses, but his insight that it was “custom,” not law, that was the root of the race problem remains apposite.Google Scholar

27 Hortense Powdermaker, After Freedom: A Cultural Study in the Deep South 44-45 (New York: Atheneum, 1939; 2d ed. 1967).Google Scholar

28 In an ironic passage, Epstein attacks two economists, James Heckman and J. Hoult Verkerke, for being too anthropological in arguing for the importance of extralegal forces in maintaining Jim Crow.Google Scholar

In order to sustain the basic position that cultural and social norms are sufficient to sustain Jim Crow, it becomes necessary to ask whether Jim Crow would have survived if southern whites had voluntarily relinquished their control over the ballot, the police force, the courts, and other instruments of state domination…. The tighter the social cartel against blacks under Jim Crow, the larger the returns to new [nondiscriminatory] entry in the marketplace, notwithstanding a social consensus in the opposite direction. In speaking of the same problem in connection with litigation in Japan, J. Mark Ram-seyer and Minoru Nakazato have demonstrated in the Japanese context that these community norms are fragile in the sense that a single outsider who has little to fear from ostracism can undermine them by going to court…. Japan is regarded as a closed society, perhaps even more so than the Old South. If the single marginal entrant can change the social balance of power in the one case, then it can do so in the other.(At 97)Google Scholar

What's remarkable here is the sense that the fragility of culture in the face of economic incentives is simply a principle that, once established, can be applied to the Old South without the need for supporting evidence of any kind.Google Scholar

Moreover, Heckman and Verkerke do not say that cultural and social norms were sufficient by themselves to sustain Jim Crow, merely that they played an important role in sustaining segregation and that the law was insufficient by itself. Google Scholar

29 In addition, the pattern of segregation began before the law was passed and continued long after the law was struck down as unconstitutional, which further argues against the importance of the law as a bulwark of segregation. See Donohue, 44 Stan. L. Rev. at 1583-95 (cited in note 15).Google Scholar

30 This accords well with the views of the distinguished Southern historian Joel Williamson, who writes:Google Scholar

Violence and the great threat of violence was one way in which [whites] sought to lower the self-esteem of blacks and thus render them more controllable on the way to their demise. We could never catalog and describe, nor even count all of these instruments of discrimination…. Two of [the] tools used to reduce, and hence, to manage blacks were disenfranchisement and segregation. These tools were often the subject of laws and legalities…. Perhaps … because these laws and legalities were so amenable to the kind of verbal reduction traditionally favored by scholars…. they have been so much emphasized by students of race relations. [They are, however] only a small and not totally representative portion of reality.Google Scholar

The Crucible of Race: Black-White Relations in the American South since Emancipation 224 (Oxford: Oxford University Press, 1984) (“Williamson, Crucible of Race”).Google Scholar

31 Even Milton Friedman the eminence grise (eminence bald?) of the Chicago School of Economics, whose Capitalism and Freedom (cited in note 30) is a paean to libertarianism notes in a different context: “To a considerable extent … in the [Indian] caste system, … restrictions [on employment opportunities] were enforced by general social customs rather than explicitly by government.”Id. at 138.Google Scholar

32 Williamson, Crucible of Race 247.Google Scholar

33 Cain, “Economic Analysis” (cited in note 2).Google Scholar

34 One is a study of discrimination in the pricing of baseball cards (Nardinelli, Clark & Simon, Curtis, “Customer Racial Discrimination in the Market for Memorabilia: The Case of Baseball,” 105 Q.J. Econ. 575 (1990)); the other examines disparities in the prices quoted to test buyers of new cars (Ayres, Ian, “Fair Driving: Gender and Race Discrimination in Retail Car Negotiations,” 104 Harv. L. Rev. 817 (1991)).CrossRefGoogle Scholar

35 Margery Austin Turner, Michael Fix, & Raymond J. Struyk, Opportunities Denied, Opportunities Diminshed: Discrimination in Hiring (Washington: Urban Institute Press, 1991).Google Scholar

36 See, e.g., Havens Realty Corp. v. Coleman, 445 U.S. 363 (1982) (granting standing to fair housing testers to sue for damages).Google Scholar

37 There are statistical and conceptual problems with aggregating the data, but they are presented in this form for expository convenience. For a more detailed analysis, see James Heckman & Peter Siegelman, “The Urban Institute's Audit Studies: Their Methods and Findings,” in Michael Fix & Raymond Struyk, eds., Clear and Convincing Evidence: The Measurement of Discrimination in America (Washington: Urban Institute Press, 1992) (“Heckman & Siegelman, ‘Urban Institute's Audit Studies’”).Google Scholar

38 Along with its similar study of discrimination against Hispanics Harry Cross et al., Employer Hiring Practices: Differential Treatment of Hispanic and Anglo Job Seekers (Urban Institute Report 90-4; Washington: Urban Institute Press, 1990) the Urban Institute's work represents the first serious attempts to use paired audits to investigate hiring discrimination. For a critical but appreciative assessment of the results, see Heckman & Siegelman, “Urban Institute's Audit Studies.”Google Scholar

39 Could Epstein have confused the Urban Institute a nonprofit, nonpartisan, public policy research institution along the lines of the Brookings Institute with the Urban League, the civil rights advocacy group? This might account for some of his suspicion of the study's findings.Google Scholar

40 Heckman & Siegelman, “Urban Institute's Audit Studies,” provide some further references on this point.Google Scholar

41 Gardner Lindzey & Elliot Aronson, 2 The Handbook of Social Psychology 66 (2d ed. Reading, Mass.: Addison-Wesley, 1975). For a more extensive treatment, see Robert Rosenthal,Experimenter Effects in Behavioral Research (2d ed. New York: Irvington Publishers, 1976).Google Scholar

42 It should be pointed out that the UI testing protocols suggest that strenuous efforts were made to keep the testers as neutral and objective as possible.Google Scholar

Furthmore, Epstein ignores all of the baises against finding discrimination in the UI study. The antidiscriminatory bias of relying on newspaper advertisements as a sampling frame was mentioned earlier. To take another example, suppose that firms use probationary hiring periods and on-the-job evaluation as substitutes for preemployment screening. (Such personnel policies seem to be quite widespread, although little is known about their extent. See, e.g., Erica Groshen & Eng Send Loh, “What Do We Know about Probationary Periods?” (Federal Reserve Bank of Cleveland, 1993).) If it exists, discrimination will then show up in differential firing rates rather than in differences in offers made to identical testers. Since the UI studies do not actually place testers on the job and examine their subsequent treatment by employers, they will be powerless to detect this kind of discrimination, as Heckman & Siegelman, “Urban Institute's Audit Studies,” point out.Google Scholar

43 What relative wages “should we expect to observe for blacks and whites, equally qualified, given the [Urban Institute] data,” Epstein asks, rhetorically (at 58). “The answer is, there will not be much difference if any at all” because “while testers apply for jobs in matched pairs, real job seekers look for those employers who will offer them the most favorable deals. Black workers will gravitate to firms that afford them a preference or to those that are indifferent in matters of race.”Google Scholar

44 Substitute “discrimination” for “complying with Title VII” in the following paragraph and you immediately see the asymmetry in Epstein's views. “Employers (and employees) may adopt strategies to mitigate the costs of complying with Title VII, and this is all to the good. But mitigation does not restore the market to the status quo ante of the unregulated world. Mitigation costs money [more precisely, resources of some kind], so after the appropriate steps are taken,” the employer is worse off than he would have been if the need to mitigate had never arisen (at 181). This is nothing more than a definition of opportunity cost, but it appears that only the costs imposed by Title VII, and not those caused by discrimination, are cognizable in Epstein's worldview.Google Scholar

45 Emphasis added; Heckman & Siegelman, “Urban Institute's Audit Studies,” at 219 (cited in note 37). In other words, black workers will be more likely to accept any given wage offered by an employer than will whites, because blacks know they will have a hard time finding another offer due to hiring discrimination.Google Scholar

46 “Price Theory” is the Chicago School's term for 1950s/60s vintage microeconomic theory as exemplified in George Stigler's text, The Theory of Price (New York: Macmillan, 1946). It has been largely superseded in the economics profession by the new economic theory of asymmetric information and strategic rivalry (see, e.g., Eric Rasmusen, Games and Information (Oxford: Basil Blackwell, 1989)), which often yields diametrically opposite conclusions.Google Scholar

47 Donohue, John & Siegelman, Peter, “The Changing Nature of Employment Discrimination Litigation,” 43 Stan. L. Rev. 983 (1991) (disparate impact cases account for only a small share of all employment discrimination claims).Google Scholar

48 The distinction between disparate treatment and disparate impact (facially neutral policies that have an unfavorable effect on a protected class) is discussed below.Google Scholar

49 “Although the costs of trying disparate treatment cases are substantial,” Epstein claims without any serious evidence on balance I have little doubt that these costs, had they been made explicit in the early debates, would have been regarded as acceptable by the Congress that passed the 1964 act. As a practical compromise, moreover, a limitation of Title VII race cases to individual disparate treatment cases would honor the original intentions of the statute while removing most of its heavy costs. These cases are not especially controversial today. (At 181)Google Scholar

50 Donohue & Siegelman, 43 Stan. L. Rev. The distorted perceptions caused by the focus on cases with published opinions is documented in Siegelman, Peter & Donohue, John, “Studying the Iceberg from Its Tip? An Analysis of Differences between Published and Unpublished Federal Employment Discrimination Cases,” 24 Law & Soc'y Rev. 1133 (1991). See also Eisenberg, Theodore & Schwab, Stewart, “What Shapes Perceptions of the Federal Court System,” 56 U. Chi. L. Rev. 501 (1989).CrossRefGoogle Scholar

51 The formal outline for how to prove a disparate impact case (articulated in McDonnell Douglas v. Green, 411 U.S. 792 (1973)) provides for a three-part standard: first the plaintiff must prove that (s)he is a member of a class of persons protected under Title VII and that she was fired or not hired from a job for which she was qualified; second, the defendant is given a chance to justify his behavior as motivated by factors other than race; finally, the plaintiff gets a chance to show that the employer's stated explanation is simply a pretext to cover up his use of race. About this burden shifting exercise, Epstein has this to say: “I doubt that anyone could come up with a simpler set of rules that would capture the full set of relevant considerations under the statute in a clearer and more logical form than was done in McDonnell Douglas and its progeny” (at 168).Google Scholar

52 For an early effort that is cognizant of both the costs and benefits of the law, and arrives at a much more balanced view than Epstein's, see Owen Fiss's now classic “A Theory of Fair Employment Laws,” 38 U. Chi. L. Rev. 235 (1971).CrossRefGoogle Scholar

53 Cain, “Economic Analysis” at 710 (cited in note 2).Google Scholar

54 Donohue, 44 Stan. L. Rev. (cited in note 15).Google Scholar

55 H. Rep. no 9014, 88th. Cong. 1st sess., 1963, at 149 (views of additional Republican sponsors), quoted in Epstein at 162.Google Scholar

56 “I think the law becomes a secondary element. I mean, if the law wasn't there, we'd still be doing what we're doing because we think it's the right thing to do and our policies and procedures are based on what we feel are sound personnel practices.” Anonymous corporate personnel officer, describing EEO law, as quoted in Edelman, Lauren, Erlanger, Howard S. & Lande, John, “Internal Dispute Resolution: The Transformation of Civil Rights in the Workplace,” 27 Law & Soc'y Rev. 497, 515 (1993). See also Dobbin, Frank, Edelman, Lauren, Meyer, John W. & Scott, W. Richard, “Equal Opportunity Law and the Construction of Internal Labor Markets,” 99 Am. J. Soc. 396 (1993).CrossRefGoogle Scholar

57 Fiss, 38 U. Chi. L. Rev., argues in my view convincingly-that Title VII was never meant to be a color-blind statute; instead, he suggests, it should be thought of as a legitimate attempt to extend a (rather limited set of) benefits to black citizens.Google Scholar

58 McDonald v. Santa Fe Trail Transp. Co., 427 U.S. 273 (1976). See also Blumrosen, Alfred, “Strangers No More: All Workers are Entitled to ‘Just Cause’ Protection under Title VII,” 2 Indus. Rel. L.J. 519 (1978), and Rutherglen, George, “Abolition in a Different Voice,” 78 Va. L. Rev. 1463 (1992) (review of Forbidden Grounds). Whether the body of judicial decisions and administrative enforcement policies has actually treated discrimination against blacks and whites equally seriously is beside the point here, since the argument is one about what the law says, not how it has been enforced.Google Scholar

59 John Donohue's review of Epstein, 44 Stan. L. Rev., provides an excellent example of what this kind of analysis looks like.Google Scholar

60 401 U.S. 424 (1971). This is a crude version of the facts in the Supreme Court decision that enunciated the disparate impact doctrine. Epstein fails to note the irony that it was the Burger court, hardly noted for its boldness, liberalism, or judicial activism that was responsible for this decision indeed, the 8-0 opinion was written by Chief Justice Warren Burger himself.Google Scholar

61 For example, Epstein (at 188) quotes Sen. Hubert Humphrey, a leading sponsor of the legislation, in congressional debate: “I would like to make an offer…. If the Senator can find in Title VII … any language which provided that an employer will have to hire on the basis of percentage or quota related to color … or race I will start eating the pages one after another, because it is not in there.” Since the phrase “disparate impact” didn't exist in 1964, Epstein correctly notes that the debates couldn't crystallize around these exact words; the question whether discrimination could be proved solely by bottom-line statistical results with no reliance on intent was thus never explicitly addressed.Google Scholar

62 Sec. 703(h) of title VII.Google Scholar

63 Much the same argument is made by Michael Gold (“Griggs' Folly: An Essay on the Theory, Problems, and Origin of the Adverse Impact Definition of Employment Discrimination and a Recommendation for Reform,” 7 Indus. Rel. L.J. 429 (1985)) whom Epstein doesn't cite.Google Scholar

64 Congress indicated its support for the disparate impact theory in the legislative history accompanying its 1972 amendments to Title VII. And the 1991 Civil Rights Act specifically overturned several Supreme Court opinions that cut into the scope of the disparate impact doctrine (Wards Cove v. Antonio, 490 U.S. 642 (1989)) and limited the use of the 1866 Civil Rights Act (Patterson v. McLean Credit Union, 491 U.S. 164 (1989)). In these decisions and others (e.g., General Elec. Co. v. Gilbert, 429 U.S. 125 (1976), holding that discrimination on the basis of pregnancy did not violate Title VII, subsequently overturned by the Pregnancy Discrimination Act of 1978) the Supreme Court actually encouraged Congress to do its job explicitly confront the issue of civil rights and decide what kind of laws it really wanted. See Ian Ayres & Robert Gertner, “Filling Gaps in Incomplete Contracts: An Economic Theory of Default Rules,” 99 Yale L.J. 87, 129 (1991), for a “penalty default rule” theory of statutory interpretation.Google Scholar

65 See, e.g., Lundberg & Startz, 73 Am. Econ. Rev. (cited in note 17).Google Scholar

66 Any labor economist would recognize this as a problem of “endogeneity bias.” Cain offers a measured discussion of this problem, in nontechnical language, in “The Uses and Limits of Statistical Analysis in Measuring Discrimination,”in Emily Hoffman, ed., Essays on the Economics of Discrimination (Kalamazoo, Mich.: W. E. Upjohn Institute, 1991); he concludes that historical context is critical, and statistical analysis alone cannot completely resolve the question of how much discrimination there is.Google Scholar

67 E.g., Albemarle Paper Co. v. Moody, 422 U.S. 405 (1975); Evans v. City of Evanston, 881 F.2d 382 (7th Cir. 1989).Google Scholar

68 This point was first made by a National Research Council report on testing (John Hartigan & Alexandra Wigdor, eds., Fairness in Employment Testing (Washington: National Academy Press, 1989)). There may be some gains to be had from matching firms and workers, but these are found to be quite small in the only empirical study of which I am aware. See James Heckman & Guilherme Sedlacek, “Self-Selection and the Distribution of Hourly Wage Rates,” 8 J. Lab. Econ. S329 (1990).Google Scholar

69 Connecticut v. Teal, 457 U.S. 440 (1981).Google Scholar

70 If one believes that the test had no predictive power at all, then the fact that Ms. Teal scored lower than some other black applicant whose place she took is irrelevant, and the decision look more reasonable.Google Scholar

71 Epstein's previous book was Takings, an attack on governmental regulation of land use.Google Scholar