Fair driving: Gender and race discrimination in retail car negotiations

Harvard Law Review

Volumn 104, Issue 4, 1991, Pages 817-872

  Ayres, Ian ^a,b  

^a AMERICAN BAR FOUNDATION   (United States)

^b NORTHWESTERN UNIVERSITY   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 0001142078     PISSN: 0017811X     EISSN: None     Source Type: Journal    
DOI: 10.2307/1341506     Document Type: Article

References (212)

1. See Civil Rights Act of 1964, 42 U.S.C. §§ 2000a to 2000h-6 (1988); Fair Housing Act of 1968, id. § 3601.
2. See, e.g., Price Water-house v. Hopkins, 109 S. Ct. 1775 (1989) (considering a gender discrimination claim that an employment decision was based on a mixture of legal and illegal criteria); Watson v. Fort Worth Bank & Trust, 487 U.S. 977 (1988) (plurality opinion) (considering an employment discrimination claim arising under discretionary or "subjective" promotion system); UAW v. Johnson Controls, Inc., 886 F.2d 871 (7th Cir. 1989) (upholding a fetal protection policy against a claim of gender discrimination), cert, granted, 110 S. Ct. 1522 (1990).
3. Academics have debated whether the scope of classes protected under the law is adequate, see, e.g., Note, Redefining Race in Saint Francis College v. Al-Khazraji and Shaare Tefili Congregation v. Cobb: Using Dictionaries Instead of the Thirteenth Amendment, 42 VAND. L. REV. 209, 228-31 (1989), but not whether the scope of markets covered by law is adequate. In other words, analysis has focused on "protected classes," not "protected markets."
4. Gender and race discrimination have been uncovered in a variety of markets. For example, several Washington, D.C., dry cleaners have discriminated against female customers bycharg&f higher prices for women's blouses than for men's shirts. See Matlack, Experts on Call, 21 NAT'L J. 2549, 2549 (1989). Historically, blacks have been discriminated against in the sale of many goods and services. For example, in 1959 black consumers and businesses associated with the NAACP were at times unable to buy such goods as milk, bread, groceries, gas, credit, fertilizer, seed, insecticides, and farm machinery. See AMERICAN FRIENDS SERV. COMM., NAT'L COUNCIL OF CHURCHES OF CHRIST & S. REGIONAL COUNCIL, INTIMIDATION, REPRISAL, AND VIOLENCE IN THE SOUTH'S RACIAL CRISIS (1959) , reprinted in Civil Rights - 1959: Hearings Before the Subcomm. on Constitutional Rights of the Senate Comm. on the Judiciary, 86th Cong., ist Sess. 1573 (1959); see also Perry v. Command Performance, 913 F.2d 99 (3rd Cir. 1990) (involving the refusal by a beauty salon hairdresser to serve a black woman).
5. Gender and race discrimination have been uncovered in a variety of markets. For example, several Washington, D.C., dry cleaners have discriminated against female customers bycharg&f higher prices for women's blouses than for men's shirts. See Matlack, Experts on Call, 21 NAT'L J. 2549, 2549 (1989). Historically, blacks have been discriminated against in the sale of many goods and services. For example, in 1959 black consumers and businesses associated with the NAACP were at times unable to buy such goods as milk, bread, groceries, gas, credit, fertilizer, seed, insecticides, and farm machinery. See AMERICAN FRIENDS SERV. COMM., NAT'L COUNCIL OF CHURCHES OF CHRIST & S. REGIONAL COUNCIL, INTIMIDATION, REPRISAL, AND VIOLENCE IN THE SOUTH'S RACIAL CRISIS (1959) , reprinted in Civil Rights - 1959: Hearings Before the Subcomm. on Constitutional Rights of the Senate Comm. on the Judiciary, 86th Cong., ist Sess. 1573 (1959); see also Perry v. Command Performance, 913 F.2d 99 (3rd Cir. 1990) (involving the refusal by a beauty salon hairdresser to serve a black woman).
6. Gender and race discrimination have been uncovered in a variety of markets. For example, several Washington, D.C., dry cleaners have discriminated against female customers bycharg&f higher prices for women's blouses than for men's shirts. See Matlack, Experts on Call, 21 NAT'L J. 2549, 2549 (1989). Historically, blacks have been discriminated against in the sale of many goods and services. For example, in 1959 black consumers and businesses associated with the NAACP were at times unable to buy such goods as milk, bread, groceries, gas, credit, fertilizer, seed, insecticides, and farm machinery. See AMERICAN FRIENDS SERV. COMM., NAT'L COUNCIL OF CHURCHES OF CHRIST & S. REGIONAL COUNCIL, INTIMIDATION, REPRISAL, AND VIOLENCE IN THE SOUTH'S RACIAL CRISIS (1959) , reprinted in Civil Rights - 1959: Hearings Before the Subcomm. on Constitutional Rights of the Senate Comm. on the Judiciary, 86th Cong., ist Sess. 1573 (1959); see also Perry v. Command Performance, 913 F.2d 99 (3rd Cir. 1990) (involving the refusal by a beauty salon hairdresser to serve a black woman).
7. See BUREAU OF ECONOMIC ANALYSIS, U.S. DEP'T OF COMMERCE, THE NATIONAL INCOME AND PRODUCT ACCOUNTS OF THE UNITED STATES, 1929-82, at 105 (1986) (table 2.3) (showing) annual personal expenditures on cars consistently to be one of the largest categories of expenditures).
8. See BUREAU OF THE CENSUS, U.S. DEP'T OF COMMERCE, STATISTICAL ABSTRACT OF THE UNITED STATES 412 (108th ed. 1988).
9. Since the Fair Housing Act of 1968 outlawed discrimination in the sale and rental of housing, numerous studies have tested whether minorities and whites are treated differently in the housing market. See sources cited infra note 138. The empirical analysis in this Article broadly borrows the methodology of "fair housing" tests. In the classic fair housing test, a black tester and a white tester separately approach a real estate agent or seller and express an interest in the same type of housing. The test of discrimination is simply whether they are treated similarly -are they shown the same houses, in the same neighborhoods, for the same price?
10. This study is the first to focus on whether sellers discriminate on the basis of race or gender when customers bargain similarly. Other studies, in contrast, have focused solely on the existence of race- or gender-based differences in bargaining techniques. See, e.g., Pruitt & Carnevale. Gender Effects in Negotiation: Constituent Surveillance and Contentious Behavior, 22 J. EXPERIMENTAL SOC. PSYCHOLOGY 264 (1986); Sampson & Kardush, Age, Sex, Class, and Race Differences in Response to a Two-Person Non-Zero-Sum Game, 9 J. CONFLICT RESOLUTION 112 (1965). In 1959, Professor Allen F. Jung of the University of Chicago Business School studied whether testers who utilized different bargaining processes obtained different price quotations from identical new car dealers. See Jung, Price Variations Among Automobile Dealers in Chicago, Illinois, 32 J. Bus. 315 (1959). In a five-page article based on the same study, Jung argued that women were not treated significantly differently from men. See Jung, Interviewer Differences Among Automobile Purchasers, 10 APPLIED STATISTICS 93, 96-97 (1961). However, Jung's own explanation of the equal bargaining results belies his interest in carrying out a controlled test: "The natural business acumen of the men and the beauty and charm of the ladies must be considered offsetting factors as far as obtaining lower automobile prices." Id at 96. Jung made no attempt to test for racial discrimination. More recent testing has focused on whether the knowledgeability or aggressiveness of consumers affects the outcome of car negotiation. Guenter Mueller and Withold Galinat's field experiment found that consumers fared better when they initiated the bargaining processes with a "tough" strategy. See Mueller & Galinat, Bargaining Efficiency in Real-Life Buyer-Seller-Interaction: A Field Experiment, in ASPIRATION LEVELS IN BARGAINING AND ECONOMIC DECISION MAKING 80, 85 (R. Tietz ed. 1982). Christina Taylor and Sharon Dawid sent testers to Ford dealerships in Connecticut and New York and found that testers who appeared knowledgeable obtained lower prices than testers who seemed naive. See Taylor & Dawid, Bargaining New Car: The Knowledgeable Versus the Naive Consumer, 59 PSYCHOLOGICAL REPS. 284, 285-86 (1986).
11. This study is the first to focus on whether sellers discriminate on the basis of race or gender when customers bargain similarly. Other studies, in contrast, have focused solely on the existence of race- or gender-based differences in bargaining techniques. See, e.g., Pruitt & Carnevale. Gender Effects in Negotiation: Constituent Surveillance and Contentious Behavior, 22 J. EXPERIMENTAL SOC. PSYCHOLOGY 264 (1986); Sampson & Kardush, Age, Sex, Class, and Race Differences in Response to a Two-Person Non-Zero-Sum Game, 9 J. CONFLICT RESOLUTION 112 (1965). In 1959, Professor Allen F. Jung of the University of Chicago Business School studied whether testers who utilized different bargaining processes obtained different price quotations from identical new car dealers. See Jung, Price Variations Among Automobile Dealers in Chicago, Illinois, 32 J. BUS. 315 (1959). In a five-page article based on the same study, Jung argued that women were not treated significantly differently from men. See Jung, Interviewer Differences Among Automobile Purchasers, 10 APPLIED STATISTICS 93, 96-97 (1961). However, Jung's own explanation of the equal bargaining results belies his interest in carrying out a controlled test: "The natural business acumen of the men and the beauty and charm of the ladies must be considered offsetting factors as far as obtaining lower automobile prices." Id at 96. Jung made no attempt to test for racial discrimination. More recent testing has focused on whether the knowledgeability or aggressiveness of consumers affects the outcome of car negotiation. Guenter Mueller and Withold Galinat's field experiment found that consumers fared better when they initiated the bargaining processes with a "tough" strategy. See Mueller & Galinat, Bargaining Efficiency in Real-Life Buyer-Seller-Interaction: A Field Experiment, in ASPIRATION LEVELS IN BARGAINING AND ECONOMIC DECISION MAKING 80, 85 (R. Tietz ed. 1982). Christina Taylor and Sharon Dawid sent testers to Ford dealerships in Connecticut and New York and found that testers who appeared knowledgeable obtained lower prices than testers who seemed naive. See Taylor & Dawid, Bargaining New Car: The Knowledgeable Versus the Naive Consumer, 59 PSYCHOLOGICAL REPS. 284, 285-86 (1986).
12. This study is the first to focus on whether sellers discriminate on the basis of race or gender when customers bargain similarly. Other studies, in contrast, have focused solely on the existence of race- or gender-based differences in bargaining techniques. See, e.g., Pruitt & Carnevale. Gender Effects in Negotiation: Constituent Surveillance and Contentious Behavior, 22 J. EXPERIMENTAL SOC. PSYCHOLOGY 264 (1986); Sampson & Kardush, Age, Sex, Class, and Race Differences in Response to a Two-Person Non-Zero-Sum Game, 9 J. CONFLICT RESOLUTION 112 (1965). In 1959, Professor Allen F. Jung of the University of Chicago Business School studied whether testers who utilized different bargaining processes obtained different price quotations from identical new car dealers. See Jung, Price Variations Among Automobile Dealers in Chicago, Illinois, 32 J. BUS. 315 (1959). In a five-page article based on the same study, Jung argued that women were not treated significantly differently from men. See Jung, Interviewer Differences Among Automobile Purchasers, 10 APPLIED STATISTICS 93, 96-97 (1961). However, Jung's own explanation of the equal bargaining results belies his interest in carrying out a controlled test: "The natural business acumen of the men and the beauty and charm of the ladies must be considered offsetting factors as far as obtaining lower automobile prices." Id at 96. Jung made no attempt to test for racial discrimination. More recent testing has focused on whether the knowledgeability or aggressiveness of consumers affects the outcome of car negotiation. Guenter Mueller and Withold Galinat's field experiment found that consumers fared better when they initiated the bargaining processes with a "tough" strategy. See Mueller & Galinat, Bargaining Efficiency in Real-Life Buyer-Seller-Interaction: A Field Experiment, in ASPIRATION LEVELS IN BARGAINING AND ECONOMIC DECISION MAKING 80, 85 (R. Tietz ed. 1982). Christina Taylor and Sharon Dawid sent testers to Ford dealerships in Connecticut and New York and found that testers who appeared knowledgeable obtained lower prices than testers who seemed naive. See Taylor & Dawid, Bargaining New Car: The Knowledgeable Versus the Naive Consumer, 59 PSYCHOLOGICAL REPS. 284, 285-86 (1986).
13. This study is the first to focus on whether sellers discriminate on the basis of race or gender when customers bargain similarly. Other studies, in contrast, have focused solely on the existence of race- or gender-based differences in bargaining techniques. See, e.g., Pruitt & Carnevale. Gender Effects in Negotiation: Constituent Surveillance and Contentious Behavior, 22 J. EXPERIMENTAL SOC. PSYCHOLOGY 264 (1986); Sampson & Kardush, Age, Sex, Class, and Race Differences in Response to a Two-Person Non-Zero-Sum Game, 9 J. CONFLICT RESOLUTION 112 (1965). In 1959, Professor Allen F. Jung of the University of Chicago Business School studied whether testers who utilized different bargaining processes obtained different price quotations from identical new car dealers. See Jung, Price Variations Among Automobile Dealers in Chicago, Illinois, 32 J. BUS. 315 (1959). In a five-page article based on the same study, Jung argued that women were not treated significantly differently from men. See Jung, Interviewer Differences Among Automobile Purchasers, 10 APPLIED STATISTICS 93, 96-97 (1961). However, Jung's own explanation of the equal bargaining results belies his interest in carrying out a controlled test: "The natural business acumen of the men and the beauty and charm of the ladies must be considered offsetting factors as far as obtaining lower automobile prices." Id at 96. Jung made no attempt to test for racial discrimination. More recent testing has focused on whether the knowledgeability or aggressiveness of consumers affects the outcome of car negotiation. Guenter Mueller and Withold Galinat's field experiment found that consumers fared better when they initiated the bargaining processes with a "tough" strategy. See Mueller & Galinat, Bargaining Efficiency in Real-Life Buyer-Seller-Interaction: A Field Experiment, in ASPIRATION LEVELS IN BARGAINING AND ECONOMIC DECISION MAKING 80, 85 (R. Tietz ed. 1982). Christina Taylor and Sharon Dawid sent testers to Ford dealerships in Connecticut and New York and found that testers who appeared knowledgeable obtained lower prices than testers who seemed naive. See Taylor & Dawid, Bargaining New Car: The Knowledgeable Versus the Naive Consumer, 59 PSYCHOLOGICAL REPS. 284, 285-86 (1986).
14. This study is the first to focus on whether sellers discriminate on the basis of race or gender when customers bargain similarly. Other studies, in contrast, have focused solely on the existence of race- or gender-based differences in bargaining techniques. See, e.g., Pruitt & Carnevale. Gender Effects in Negotiation: Constituent Surveillance and Contentious Behavior, 22 J. EXPERIMENTAL SOC. PSYCHOLOGY 264 (1986); Sampson & Kardush, Age, Sex, Class, and Race Differences in Response to a Two-Person Non-Zero-Sum Game, 9 J. CONFLICT RESOLUTION 112 (1965). In 1959, Professor Allen F. Jung of the University of Chicago Business School studied whether testers who utilized different bargaining processes obtained different price quotations from identical new car dealers. See Jung, Price Variations Among Automobile Dealers in Chicago, Illinois, 32 J. BUS. 315 (1959). In a five-page article based on the same study, Jung argued that women were not treated significantly differently from men. See Jung, Interviewer Differences Among Automobile Purchasers, 10 APPLIED STATISTICS 93, 96-97 (1961). However, Jung's own explanation of the equal bargaining results belies his interest in carrying out a controlled test: "The natural business acumen of the men and the beauty and charm of the ladies must be considered offsetting factors as far as obtaining lower automobile prices." Id at 96. Jung made no attempt to test for racial discrimination. More recent testing has focused on whether the knowledgeability or aggressiveness of consumers affects the outcome of car negotiation. Guenter Mueller and Withold Galinat's field experiment found that consumers fared better when they initiated the bargaining processes with a "tough" strategy. See Mueller & Galinat, Bargaining Efficiency in Real-Life Buyer-Seller-Interaction: A Field Experiment, in ASPIRATION LEVELS IN BARGAINING AND ECONOMIC DECISION MAKING 80, 85 (R. Tietz ed. 1982). Christina Taylor and Sharon Dawid sent testers to Ford dealerships in Connecticut and New York and found that testers who appeared knowledgeable obtained lower prices than testers who seemed naive. See Taylor & Dawid, Bargaining New Car: The Knowledgeable Versus the Naive Consumer, 59 PSYCHOLOGICAL REPS. 284, 285-86 (1986).
15. This study is the first to focus on whether sellers discriminate on the basis of race or gender when customers bargain similarly. Other studies, in contrast, have focused solely on the existence of race- or gender-based differences in bargaining techniques. See, e.g., Pruitt & Carnevale. Gender Effects in Negotiation: Constituent Surveillance and Contentious Behavior, 22 J. EXPERIMENTAL SOC. PSYCHOLOGY 264 (1986); Sampson & Kardush, Age, Sex, Class, and Race Differences in Response to a Two-Person Non-Zero-Sum Game, 9 J. CONFLICT RESOLUTION 112 (1965). In 1959, Professor Allen F. Jung of the University of Chicago Business School studied whether testers who utilized different bargaining processes obtained different price quotations from identical new car dealers. See Jung, Price Variations Among Automobile Dealers in Chicago, Illinois, 32 J. BUS. 315 (1959). In a five-page article based on the same study, Jung argued that women were not treated significantly differently from men. See Jung, Interviewer Differences Among Automobile Purchasers, 10 APPLIED STATISTICS 93, 96-97 (1961). However, Jung's own explanation of the equal bargaining results belies his interest in carrying out a controlled test: "The natural business acumen of the men and the beauty and charm of the ladies must be considered offsetting factors as far as obtaining lower automobile prices." Id at 96. Jung made no attempt to test for racial discrimination. More recent testing has focused on whether the knowledgeability or aggressiveness of consumers affects the outcome of car negotiation. Guenter Mueller and Withold Galinat's field experiment found that consumers fared better when they initiated the bargaining processes with a "tough" strategy. See Mueller & Galinat, Bargaining Efficiency in Real-Life Buyer-Seller-Interaction: A Field Experiment, in ASPIRATION LEVELS IN BARGAINING AND ECONOMIC DECISION MAKING 80, 85 (R. Tietz ed. 1982). Christina Taylor and Sharon Dawid sent testers to Ford dealerships in Connecticut and New York and found that testers who appeared knowledgeable obtained lower prices than testers who seemed naive. See Taylor & Dawid, Bargaining New Car: The Knowledgeable Versus the Naive Consumer, 59 PSYCHOLOGICAL REPS. 284, 285-86 (1986).
16. Paul Brest has similarly defined race discrimination in terms of "race-dependent decions and conduct." Brest, The Supreme Court, 1975 Term - Foreword: In Defense of the Antidiscrimination Principle, 90 HARV. L. REV. 1, 6 (1976).
17. In the 1988 Term, the Court departed from past precedent and created shifting burdens of persuasion in disparate treatment cases when an employment decision is motivated at least in part by intentional discrimination. See Price Waterhouse v. Hopkins, 109 S. Ct. 1775, 1786 (1989). The Court held that in such "mixed-motives" cases, the plaintiff must first establist that an unlawful motive was a motivating factor, after which the defendant employer bears a burden of persuasion that "it would have made the same decision even if it had not allowed gender to play such a role." Id. at 1788. The year before, in Watson v. Fort Worth Bank & Trust, 487 U.S. 977 (1988), a plurality held that disparate impact analysis was applicable to subjective hiring practices, but required plaintiffs establishing a prima facie case to identify "the specific employment practices that are allegedly responsible for any observed statistical disparities." Id. at 994. Although in both cases a divided Court recognized that subjective and mixed-motives discrimination are actionable, it had difficulty allocating the evidentiary burdens. The Court is not alone in its effort to develop workable standards for novel civil rights claims. More than 150 members of Congress have entered into this debate by sponsoring the Civil Rights Act of 1990, S. 2104, 101st Cong., 2d Sess., 136 CONG. REC. S1018 (daily ed. 1990). Among other things, the Act would have prohibited any employment practice if "race, color, religion, sex or national origin was a motivating factor for [that] employment practice, even though such practice was also motivated by other factors." See id. § 5(a)(1). President Bush successfully rejected Congress's efforts. See Lewis, President's Veto of Rights Measure Survives by 1 Vote, N.Y. Times, Oct. 25, 1990, at A1, col. 3. Academics as well have been increasingly interested in the outer reaches of subjective and mixed-motives discrimination. See e.g., Lawrence, The Id, the Ego, and Equal Protection: Reckoning with Unconscious Racism, 39 STAN. L. REV. 317 (1987) (examining subtle and unconscious forms of racism); Ortiz The Myth of Intent in Equal Protection, 41 STAN. L. REV. 1105 (1989) (same).
18. In the 1988 Term, the Court departed from past precedent and created shifting burdens of persuasion in disparate treatment cases when an employment decision is motivated at least in part by intentional discrimination. See Price Waterhouse v. Hopkins, 109 S. Ct. 1775, 1786 (1989). The Court held that in such "mixed-motives" cases, the plaintiff must first establist that an unlawful motive was a motivating factor, after which the defendant employer bears a burden of persuasion that "it would have made the same decision even if it had not allowed gender to play such a role." Id. at 1788. The year before, in Watson v. Fort Worth Bank & Trust, 487 U.S. 977 (1988), a plurality held that disparate impact analysis was applicable to subjective hiring practices, but required plaintiffs establishing a prima facie case to identify "the specific employment practices that are allegedly responsible for any observed statistical disparities." Id. at 994. Although in both cases a divided Court recognized that subjective and mixed-motives discrimination are actionable, it had difficulty allocating the evidentiary burdens. The Court is not alone in its effort to develop workable standards for novel civil rights claims. More than 150 members of Congress have entered into this debate by sponsoring the Civil Rights Act of 1990, S. 2104, 101st Cong., 2d Sess., 136 CONG. REC. S1018 (daily ed. 1990). Among other things, the Act would have prohibited any employment practice if "race, color, religion, sex or national origin was a motivating factor for [that] employment practice, even though such practice was also motivated by other factors." See id. § 5(a)(1). President Bush successfully rejected Congress's efforts. See Lewis, President's Veto of Rights Measure Survives by 1 Vote, N.Y. Times, Oct. 25, 1990, at A1, col. 3. Academics as well have been increasingly interested in the outer reaches of subjective and mixed-motives discrimination. See e.g., Lawrence, The Id, the Ego, and Equal Protection: Reckoning with Unconscious Racism, 39 STAN. L. REV. 317 (1987) (examining subtle and unconscious forms of racism); Ortiz The Myth of Intent in Equal Protection, 41 STAN. L. REV. 1105 (1989) (same).
19. In the 1988 Term, the Court departed from past precedent and created shifting burdens of persuasion in disparate treatment cases when an employment decision is motivated at least in part by intentional discrimination. See Price Waterhouse v. Hopkins, 109 S. Ct. 1775, 1786 (1989). The Court held that in such "mixed-motives" cases, the plaintiff must first establist that an unlawful motive was a motivating factor, after which the defendant employer bears a burden of persuasion that "it would have made the same decision even if it had not allowed gender to play such a role." Id. at 1788. The year before, in Watson v. Fort Worth Bank & Trust, 487 U.S. 977 (1988), a plurality held that disparate impact analysis was applicable to subjective hiring practices, but required plaintiffs establishing a prima facie case to identify "the specific employment practices that are allegedly responsible for any observed statistical disparities." Id. at 994. Although in both cases a divided Court recognized that subjective and mixed-motives discrimination are actionable, it had difficulty allocating the evidentiary burdens. The Court is not alone in its effort to develop workable standards for novel civil rights claims. More than 150 members of Congress have entered into this debate by sponsoring the Civil Rights Act of 1990, S. 2104, 101st Cong., 2d Sess., 136 CONG. REC. S1018 (daily ed. 1990). Among other things, the Act would have prohibited any employment practice if "race, color, religion, sex or national origin was a motivating factor for [that] employment practice, even though such practice was also motivated by other factors." See id. § 5(a)(1). President Bush successfully rejected Congress's efforts. See Lewis, President's Veto of Rights Measure Survives by 1 Vote, N.Y. Times, Oct. 25, 1990, at A1, col. 3. Academics as well have been increasingly interested in the outer reaches of subjective and mixed-motives discrimination. See e.g., Lawrence, The Id, the Ego, and Equal Protection: Reckoning with Unconscious Racism, 39 STAN. L. REV. 317 (1987) (examining subtle and unconscious forms of racism); Ortiz The Myth of Intent in Equal Protection, 41 STAN. L. REV. 1105 (1989) (same).
20. In the 1988 Term, the Court departed from past precedent and created shifting burdens of persuasion in disparate treatment cases when an employment decision is motivated at least in part by intentional discrimination. See Price Waterhouse v. Hopkins, 109 S. Ct. 1775, 1786 (1989). The Court held that in such "mixed-motives" cases, the plaintiff must first establist that an unlawful motive was a motivating factor, after which the defendant employer bears a burden of persuasion that "it would have made the same decision even if it had not allowed gender to play such a role." Id. at 1788. The year before, in Watson v. Fort Worth Bank & Trust, 487 U.S. 977 (1988), a plurality held that disparate impact analysis was applicable to subjective hiring practices, but required plaintiffs establishing a prima facie case to identify "the specific employment practices that are allegedly responsible for any observed statistical disparities." Id. at 994. Although in both cases a divided Court recognized that subjective and mixed-motives discrimination are actionable, it had difficulty allocating the evidentiary burdens. The Court is not alone in its effort to develop workable standards for novel civil rights claims. More than 150 members of Congress have entered into this debate by sponsoring the Civil Rights Act of 1990, S. 2104, 101st Cong., 2d Sess., 136 CONG. REC. S1018 (daily ed. 1990). Among other things, the Act would have prohibited any employment practice if "race, color, religion, sex or national origin was a motivating factor for [that] employment practice, even though such practice was also motivated by other factors." See id. § 5(a)(1). President Bush successfully rejected Congress's efforts. See Lewis, President's Veto of Rights Measure Survives by 1 Vote, N.Y. Times, Oct. 25, 1990, at A1, col. 3. Academics as well have been increasingly interested in the outer reaches of subjective and mixed-motives discrimination. See e.g., Lawrence, The Id, the Ego, and Equal Protection: Reckoning with Unconscious Racism, 39 STAN. L. REV. 317 (1987) (examining subtle and unconscious forms of racism); Ortiz The Myth of Intent in Equal Protection, 41 STAN. L. REV. 1105 (1989) (same).
21. See 42 U.S.C. § 1981 (1988) (prohibiting racial discrimination in contracting); id. § 1982 (prohibiting racial discrimination in the sale of personal property).
22. 14 Stat. 27 (1866).
23. State civil and human rights statutes also fail explicitly to protect either women or blacks from discrimination in the sale of consumer goods and services. See, e.g., Human Rights Act, ILL. REV. STAT. ch. 68, paras, 1-101 to 5A-102 (1989) (prohibiting discrimination in employment, real estate transactions, financial credit, public accommodations, and higher education). Some state statutes might prohibit such discrimination by construing the retail sale of goods or services to be a "public accommodation." See IDAHO CODE § 18-7301(2) (1987) (guaranteeing "full enjoyment" of any public accommodation); id. § 18-7302(c) (defining "full enjoyment" to include tne right to purchase any "article of personal property offered or sold on, or by, any establishment to the public").
24. 14 Stat. 27, reenacted by Enforcement Act of 1870, § 18, 16 Stat. 144 (codified at 42 U.S.C. §§ 1981, 1982 (1988)).
25. Jones v. Alfred H. Mayer Co., 392 U.S. 409, 443 (1968).
26. The increasingly accepted conception of "relational contract," I. MACNEIL, THE NEW SOCIAL CONTRACT 10 (1980), runs counter to the notion that all unprotected markets are discrete exchanges and therefore immune to animus-based discrimination. Indeed, it is difficult, at a theoretical level, to see why retail purchases of personal property involve less personal contact than many public accommodations. Although the uniform pricing of many consumer goods eliminates the possibility of price discrimination, the ongoing relational nature of exchange may allow gender-based or racial animus to be reflected along other dimensions of product or service quality See infra note 149. Moreover, studies in markets with more discrete exchanges indicate that women and minorities can be disadvantaged when there is gender- or race-based product differentiation. See sources cited supra note 4.
27. For example, the necessity of bargaining dramatically increases the search costs of buying a good and may therefore engender transactional inefficiencies as well. The social utility of bargaining is generally addressed in Ayres & Miller, "I'llo Sell It to You at Cost": Legal Methods to Promote Retail Markup Disclosure, 84 NW. U.L. REV. 1047, 1062-70 (1990).
28. The testers did not inform the salespeople that they were participating in a test. This lack of disclosure raises significant ethical concerns - as the salesperson's time is spent without chance of a sale. The study has several features designed to mitigate the problem of wasting the salespersons' time during the negotiation process. Most important, the testers visited the dealerships during the least busy times of the week (from the hours of 9-12 and 1-5 Monday to Friday). During these times few people shop for cars, and there are often several salespeople without customers to serve. In addition, testers were instructed that if all the salespeople of a dealership were busy, they should return to the dealership at another time. In only one of raort than 180 visits did the testers have to discontinue the test because of crowding. Steps were also taken to minimize the time that the testers spent with the salespeople. The test itself was designed to be completed in 10 to 15 minutes and the testers were instructed to spend no more than an hour at a dealership. The Federal Judicial Center Advisory Committee on Experimentation in the Law has proposed guidelines for limiting the use of deception in legal experimentation. The committee concluded that "[d]eception requires (1) that the concealment itself be indispensable to the validity of experimental results, and (2) that the burden of justification for the practice concealed not merely be met, but met by a clear and convincing margin." ADVISORY COMM. ON EXPERIMENTATION IN THE LAW, FED. JUDICIAL CENTER, EXPERIMENTATION IN THE LAW 46 (1981). The first requirement is easily met: asking salespeople if they could be tested for race and gender discrimination would certainly change their behavior. Whether the study meets the secondary burden of justification is a closer question. As reported below, blacks in this study were often forced to pay two to three times the markup of white males. If this amount of discrimination holds for all sales in the United States, blacks annually would pay $150 million more for new cars than do white males. The benefits from documenting such potentially significant discrimination seem to meet the burden of justification. The tests have been given approval by the Human Subject Research Committees of both the American Bar Foundation and Northwestern University. See Letter from Northwestern University Institutional Review Board (June 12, 1989); oral approval from American Bar Foundation (May 1988). Deceptive tests of new car sales have been conducted by other researchers. See supra note 8. In other fields, social scientists have feigned to be, among other things, cancer patients in hospitals and potential buyers in shoe stores. See, e.g., Schaps, Cost, Dependency, and Helping, 21 J. PERSONALITY & SOC. PSYCHOLOGY 74 (1972) (involving accomplices posing as shoe store customers). The Supreme Court itself has condoned similar deception by giving fair housing testers standing. See, e.g., Havens Realty Corp. v. Coleman, 455 U.S. 363 (1982).
29. The testers did not inform the salespeople that they were participating in a test. This lack of disclosure raises significant ethical concerns - as the salesperson's time is spent without chance of a sale. The study has several features designed to mitigate the problem of wasting the salespersons' time during the negotiation process. Most important, the testers visited the dealerships during the least busy times of the week (from the hours of 9-12 and 1-5 Monday to Friday). During these times few people shop for cars, and there are often several salespeople without customers to serve. In addition, testers were instructed that if all the salespeople of a dealership were busy, they should return to the dealership at another time. In only one of raort than 180 visits did the testers have to discontinue the test because of crowding. Steps were also taken to minimize the time that the testers spent with the salespeople. The test itself was designed to be completed in 10 to 15 minutes and the testers were instructed to spend no more than an hour at a dealership. The Federal Judicial Center Advisory Committee on Experimentation in the Law has proposed guidelines for limiting the use of deception in legal experimentation. The committee concluded that "[d]eception requires (1) that the concealment itself be indispensable to the validity of experimental results, and (2) that the burden of justification for the practice concealed not merely be met, but met by a clear and convincing margin." ADVISORY COMM. ON EXPERIMENTATION IN THE LAW, FED. JUDICIAL CENTER, EXPERIMENTATION IN THE LAW 46 (1981). The first requirement is easily met: asking salespeople if they could be tested for race and gender discrimination would certainly change their behavior. Whether the study meets the secondary burden of justification is a closer question. As reported below, blacks in this study were often forced to pay two to three times the markup of white males. If this amount of discrimination holds for all sales in the United States, blacks annually would pay $150 million more for new cars than do white males. The benefits from documenting such potentially significant discrimination seem to meet the burden of justification. The tests have been given approval by the Human Subject Research Committees of both the American Bar Foundation and Northwestern University. See Letter from Northwestern University Institutional Review Board (June 12, 1989); oral approval from American Bar Foundation (May 1988). Deceptive tests of new car sales have been conducted by other researchers. See supra note 8. In other fields, social scientists have feigned to be, among other things, cancer patients in hospitals and potential buyers in shoe stores. See, e.g., Schaps, Cost, Dependency, and Helping, 21 J. PERSONALITY & SOC. PSYCHOLOGY 74 (1972) (involving accomplices posing as shoe store customers). The Supreme Court itself has condoned similar deception by giving fair housing testers standing. See, e.g., Havens Realty Corp. v. Coleman, 455 U.S. 363 (1982).
30. The study randomly determined the order in which members of a tester pair bargained at a dealership so that the white male tester was at times the first tester and at times the second tester to bargain at a dealership. The paired testers usually visited the dealership on the same day (and at least within four days of each other). The dealerships were randomly selected from the population of Chicago dealerships selling the particular cars, and the tester pairs were randomly assigned to the dealerships.
31. This distributional context removes collaboration and problem solving as measures of effective bargaining. See R. FISHER & W. URY, GETTING TO YES 73-79 (1981) (noting examples of problems that arise in negotiation). For example, the bargainers could not structure the timing of payment to enhance the gains from trade. In many real-world bargaining contexts, collaborative or "win-win" solutions do not exist. See White, The Pros and Cons of "Getting to Yes" (Book Review), 34 J. LEGAL EDUC. 115, 116 (1984). The bargaining instead resembles the classic "split-the-dollar" game in which two contestants can split a dollar if they can agree on how to divide it between them. See E. RASMUSEN, GAMES AND INFORMATION 227-29 (1989).
32. This distributional context removes collaboration and problem solving as measures of effective bargaining. See R. FISHER & W. URY, GETTING TO YES 73-79 (1981) (noting examples of problems that arise in negotiation). For example, the bargainers could not structure the timing of payment to enhance the gains from trade. In many real-world bargaining contexts, collaborative or "win-win" solutions do not exist. See White, The Pros and Cons of "Getting to Yes" (Book Review), 34 J. LEGAL EDUC. 115, 116 (1984). The bargaining instead resembles the classic "split-the-dollar" game in which two contestants can split a dollar if they can agree on how to divide it between them. See E. RASMUSEN, GAMES AND INFORMATION 227-29 (1989).
33. This distributional context removes collaboration and problem solving as measures of effective bargaining. See R. FISHER & W. URY, GETTING TO YES 73-79 (1981) (noting examples of problems that arise in negotiation). For example, the bargainers could not structure the timing of payment to enhance the gains from trade. In many real-world bargaining contexts, collaborative or "win-win" solutions do not exist. See White, The Pros and Cons of "Getting to Yes" (Book Review), 34 J. LEGAL EDUC. 115, 116 (1984). The bargaining instead resembles the classic "split-the-dollar" game in which two contestants can split a dollar if they can agree on how to divide it between them. See E. RASMUSEN, GAMES AND INFORMATION 227-29 (1989).
34. If the salesperson showed the tester more than one car, the script instructed the tester: "within two or three minutes focus your attention on the car with the lowest sticker price. This will be the car that you will then bargain over. You should indicate this by saying: 'I'm interested in buying this car.'" Tester Script 4 (Nov. 8, 1990) (on file at the Harvard Law School Library).
35. Testers were instructed to respond to questions such as "will you need help with a loan?" by saving, "No, I can provide my own financing." Id. at 11.
36. If the salesperson failed to quote an initial price, testers would ask, "How much would the car cost me to buy it [sic] today, including taxes and other fees?" Id. at 5.
37. Because sellers will seldom sell below their marginal cost, the marginal cost counteroffer established an initial position that approximated the seller's reservation price (the minimum amount for which seller should sell to make any profit). Estimates of dealer cost were obtained from Consumer Reports Auto Price Service (Mar. 2, 1990) (computer printout) and EDMUND'S 1989 NEW CAR PRICES (Nov. 1989).
38. Because sellers will seldom sell below their marginal cost, the marginal cost counteroffer established an initial position that approximated the seller's reservation price (the minimum amount for which seller should sell to make any profit). Estimates of dealer cost were obtained from Consumer Reports Auto Price Service (Mar. 2, 1990) (computer printout) and EDMUND'S 1989 NEW CAR PRICES (Nov. 1989).
39. It should be noted that the testers did not make legally binding counteroffers. The testers were carefully trained not to sign anything so that they would be protected by the statute of frauds. See U.C.C. § 2-201 (1987) (invalidating oral contracts for more than $500). Moreover, the testers did not make actual counteroffers but merely invited additional offers by saying: "Would you sell me this car today for $. . . ?" Tester Script, supra note 21, at 6.
40. Consider, for example, a seller who initially offers to sell a car for $13,000. The tester counters at $10,000 (an estimate of the car's marginal cost). If the salesperson lowered the initial offer to $12,000, the tester would wait five minutes and split the difference by offering $11,000 [(12,000 + 10,000) ÷ 2].
41. In addition to the types of factors described above, the script also controlled ancillary aspects of the bargaining. For example, testers parked their cars away from the dealerships walked onto the dealership lots, and waited in the center of the showroom to be approached by a salesperson. Significantly, the script allowed the testers to be steered to different cars and different salespeople. Forcing the second tester to seek out the same car or the same salesperson as the first tester would have introduced non-uniformity in the testers' bargaining strategies Moreover, the study was designed to test for disparate treatment using the car dealership as the unit of analysis. Allowing testers to bargain with different salespeople afforded a test of whethfr dealerships engage in more sophisticated forms of discrimination by steering classes of testers to particular kinds of cars or particular kinds of salespeople. See infra pp. 833-34. For ethical reasons, the testers did not tape-record the bargaining sessions. Because tht individual testers were the only observers of the field bargaining sessions, there are two potential types of experimental error in the results. First, the testers may have failed accurately to observe and describe their own behavior; second, the testers may have failed accurately to observe and describe the behavior of the salesperson. The training and initial tester observation were used as prophylactics to minimize both types of errors. As with fair housing studies, the testers were aware that the research was intended to determine whether race and gender disparate treatment exists. This methodology introduces the possibility that the testers' expectations or motives affected the results. However, subsequent "double blind" testing, conducted so that neither the sellers nor the buyers/testers knew that the study tested for race or gender discrimination, yielded similar results. See infra note 36. Those teters in the subsequent study were told only that the study concerned how sellers negotiate car sales.
42. Testers were sensitized to issues of body language and nonverbal cues. For example, they were told to avoid eye contact and not to cross their arms.
43. The script provided an all-purpose default or residual response for questions not otherwise anticipated. For example, if the salesperson asked the tester a detailed question about the tester's career or personal background, the tester was instructed to respond: "I don't mean to be rude, but I'm kind of pressed for time, and would rather just talk [about] buying a car" Tester Script, supra note 21, at 12.
44. This aspect of the script can be analogized to the party game in which one person is told to leave the room so that the group can make up a story about him or her. When the person returns to the room, he or she asks yes or no questions in order to construct the story. The trick to the game is that the group never constructs a story, but simply decides to answer "yes" to any question ending in a vowel (and "no" to any question ending in a consonant). The questioner in the party game thus effectively constructs the story, thereby revealing what's on his or her own mind.
45. The additional testing involved 36 testers (eight black females, five black males, six white females, and 17 white males) conducting over 400 tests. See infra note 36. This larger study indicates that individual bargaining effects cannot explain the disparate treatment: different black female testers, for example, were treated similarly -that is, sellers consistently treated them worse than they did white males. See infra note 38.
46. Statistical significance measures whether discrimination could result from random chance. To say, for example, that the average final offers sellers made to black and white testers are statistically different at a five percent significance level means that the differences would only be produced randomly five percent of the time (one out of 20 times). If a sample size is large enough, even small absolute differences in price (of, say, five dollars) will be statistically sigificant. Cf. 29 C.F.R. § 1607.4(D) (1990) (generally defining adverse impact for purposes of EEOC finding of employment discrimination as existing only when disparities are both statis-tically significant and proportionally large).
47. The lowest price offered by a dealer could come either when the dealer attempted to accept a tester's final offer or when a dealer refused to lower his last offer. See supra p. 824.
48. The cost estimate, obtained from Consumer Reports and Edmund's, see supra note 24, is one of marginal cost, in that the dealer's fixed and overhead costs are not included. These cost estimates ignore "hold backs," "incentives," and other types of manufacturer refunds that reduce the dealer's net marginal cost. Domestic car manufacturers traditionally (and foreign car manufacturers recently) have periodically refunded approximately three percent of the dealer's original cost as a so-called "hold back." See R. SUTTON, DON'T GET TAKEN EVERY TIME 23 (1986). In addition, manufacturers at times will institute "dealer incentives" - additional refunds to dealers for sales. See, e.g., Weekly Incentive Survey, AUTOMOTIVE NEWS, Mar. 5, 1990. at 38. Because the exact size of these hold backs and incentives is not public knowledge, the tost estimates were not discounted to reflect these amounts.
49. The cost estimate, obtained from Consumer Reports and Edmund's, see supra note 24, is one of marginal cost, in that the dealer's fixed and overhead costs are not included. These cost estimates ignore "hold backs," "incentives," and other types of manufacturer refunds that reduce the dealer's net marginal cost. Domestic car manufacturers traditionally (and foreign car manufacturers recently) have periodically refunded approximately three percent of the dealer's original cost as a so-called "hold back." See R. SUTTON, DON'T GET TAKEN EVERY TIME 23 (1986). In addition, manufacturers at times will institute "dealer incentives" - additional refunds to dealers for sales. See, e.g., Weekly Incentive Survey, AUTOMOTIVE NEWS, Mar. 5, 1990. at 38. Because the exact size of these hold backs and incentives is not public knowledge, the tost estimates were not discounted to reflect these amounts.
50. The data from some early negotiations were discarded because testers' initial counteroffers overestimated the dealers' marginal cost (because of recently announced rebates), so that dealers were accepting the inflated counteroffers.
51. Over 400 additional tests, similarly conducted, have just been completed in the Chicago area. Thirty-six testers (five black males, eight black females, six white females, and 17 white males) bargained over nine different car models. After controlling for the type of car and the type of bargaining strategy, see infra note 42, the final average offers sellers made to white females were $211 higher than those made to white males; final offers to black females Wire $397 higher than to white males; and final offers to black males were $1022 higher than to white males. (These differences were each statistically significant at least at the five percent level.) Again, the findings confirm the major conclusion of this study: black consumers are offered significantly higher prices than white males. As in the original test, sellers consistently offered white male testers the lowest prices and white female testers the second lowest prices. In the additional testing, however, black males fared worse than black females. This suggests that individual characteristics of the testers may have influenced the results. But see supra note 31 and infra note 38. The black male tester in the initial experiment, for example, was himself a former car salesperson and is currently a law student. It is possible that the lower offers he received in the initial experiment were by-products of his overly aggressive deviations from the script. This possibly idiosyncratic aggressiveness should, however, have biased the initial study against finding discrimination and thus further strengthens the credibility of the results. Ancillary tests of discrimination also buttress the results reported in this Article. Three Thus seems that dealers implement one of two strategies against female customers - either quoting them large markup prices or very low "low-ball" prices.
52. If the dealer offer to white males is used as the bench-mark, the "gender effect" - white female offer minus white male offer - was approximately $150. Similarly, the "race effect" - "black male offer minus white male offer - was approximately $400. Sellers asked black women, however, to pay nearly $900 more than white men, a figure 50% greater than the combined isolated race and gender effects. Because the discrimination against black females cannot be nonarbitrarily allocated to race or gender discrimination, this superadditivity makes it impossible to say whether blacks as a class did worse than women as a class.
53. The proposition is also supported by the results of the additional tests discussed above See supra note 36. Analyzing the effects of individual testers bargaining for specific cars demonstrated that only 10 individual testers out of 126 tester/model combinations were statistically different from the group average.
54. The seller's rate of concession is the amount by which a seller lowers the previous offer.
55. In calculating these averages, it was important to disaggregate bargaining of different lengths because a seller's offer in any particular round might depend on whether the seller intended to bargain into further rounds (if necessary). It is important to note the possibility that averaging together the round-by-round offers of different types of testers may disguise heterogeneous seller strategies. For example, if sellers had two ways of bargaining with white males, averaging the two different strategies would mask this heterogeneity.
56. As described above, the tester elicits an initial offer from the salesperson and then makes an initial counteroffer at an estimate of dealer cost. After the dealer's second offer the tester splits the difference for the first time. The "split the difference" strategy cannot perpetuate initial offer discrimination in the second round, because the testers counter at marginal cost regardless of the dealer's initial offer.
57. For example, second-offer discrimination by sellers forced black female testers to raise their initial counteroffers by an average of $866, whereas white male testers raised their initial counteroffers by only $218. Because black females had already conceded such a large amount, sellers may have been more likely to advance disparate offers in subsequent rounds. To investigate further whether a "split the difference" strategy perpetuated initial-offer discrimination, some testers in the additional testing used instead a "fixed concession" bargaining strategy. This group of testers began by making an initial offer equal to dealer cost. In subsequent rounds they increased their offer by 10% of the difference between the dealer's last offer and the car's sticker price. Because these fixed concessions were independent of the magnitude of initial-offer discrimination, this strategy should have been more resistant to the perpetuation effect. The additional testing found, however, that fixed concessions did not significantly reduce the perpetuation of initial-offer discrimination. Black and women testers received higher final offers regardless of whether or not they used a split-the-difference strategy.
58. There is a difference between claiming that the script does not establish disparate treatment and claiming that it does not establish a "relevant" form of disparate treatment. If few consumers employ a split-the-difference strategy, critics could plausibly argue that the results have little relevance in establishing that race and gender discrimination takes place in real-world negotiation After all, if testers uniformly stuck their tongues out, findings of disparate treatment would lack relevancy. But at least for those consumers who do split the difference, the results reveal a relevant form of disparate treatment.
59. In the sale of housing, for example, sellers generally discriminate in order to discourage blacks from purchasing. See R. HELPER, RACIAL POLICIES AND PRACTICES OF REAL ESTATE BROKERS 42-46 (1969). Refusals to bargain and the steering of black consumers are the classic methods of achieving this end. Even in the sale of housing, however, there are numerous cases detailing discrimination with the intent to sell at a higher price, and such discrimination was explicitly outlawed by the Fair Housing Act, 42 U.S.C. § 3604(b) (1988). See, e.g., United States v. Pelzer Realty Co., 484 F.2d 438, 442-43 (5th Cir. 1973) (finding illegal a realtor's requirement that black home-buyers either bring the realtor additional business or pay higher prices). See generally R. SCHWEMM, HOUSING DISCRIMINATION LAW 155-56 (1983) (summarizing the requirements of § 3604(b)'s prohibition on discriminatory terms).
60. In the sale of housing, for example, sellers generally discriminate in order to discourage blacks from purchasing. See R. HELPER, RACIAL POLICIES AND PRACTICES OF REAL ESTATE BROKERS 42-46 (1969). Refusals to bargain and the steering of black consumers are the classic methods of achieving this end. Even in the sale of housing, however, there are numerous cases detailing discrimination with the intent to sell at a higher price, and such discrimination was explicitly outlawed by the Fair Housing Act, 42 U.S.C. § 3604(b) (1988). See, e.g., United States v. Pelzer Realty Co., 484 F.2d 438, 442-43 (5th Cir. 1973) (finding illegal a realtor's requirement that black home-buyers either bring the realtor additional business or pay higher prices). See generally R. SCHWEMM, HOUSING DISCRIMINATION LAW 155-56 (1983) (summarizing the requirements of § 3604(b)'s prohibition on discriminatory terms).
61. In the sale of housing, for example, sellers generally discriminate in order to discourage blacks from purchasing. See R. HELPER, RACIAL POLICIES AND PRACTICES OF REAL ESTATE BROKERS 42-46 (1969). Refusals to bargain and the steering of black consumers are the classic methods of achieving this end. Even in the sale of housing, however, there are numerous cases detailing discrimination with the intent to sell at a higher price, and such discrimination was explicitly outlawed by the Fair Housing Act, 42 U.S.C. § 3604(b) (1988). See, e.g., United States v. Pelzer Realty Co., 484 F.2d 438, 442-43 (5th Cir. 1973) (finding illegal a realtor's requirement that black home-buyers either bring the realtor additional business or pay higher prices). See generally R. SCHWEMM, HOUSING DISCRIMINATION LAW 155-56 (1983) (summarizing the requirements of § 3604(b)'s prohibition on discriminatory terms).
62. Critics might argue that the black and female testers would not have received a higher price if, at the end of the test, they had given the dealership a "take it or leave it" price. But why should black and female consumers expend additional effort to gain a lower price? It may be that black and female testers could also have received the white male price if they had executed twenty push-ups during the course of bargaining. If so, that white male testers did not have to execute the push-ups to receive the better price would clearly constitute discrimination.
63. The relatively brief period of time spent per round with black female testers (11.2 minutes) may indicate that salespeople were not bargaining seriously with them. See infra note 69 (discussing qualifying a customer as a prerequisite for "serious" bargaining).
64. Using ordered statistics, one can estimate the expected gains that different testers would experience by searching for the minimum price at additional dealerships. The more prices vary from dealer to dealer, the more likely it becomes that a search will turn up better offers. See G. STIGLER, THE ORGANIZATION OF INDUSTRY 173-75 (1968).
65. See infra p. 848.
66. Consumers rationally value information concerning a seller's costs in "thin markets," in which the infrequency of transactions makes the competitive price hard to determine. See Ayres & Miller, supra note 17, at 1059-60.
67. Multivariate regression analysis produces quantitative estimates of the effects of different factors on the same variable. For an excellent introduction to regression analysis, see Fisher, Multiple Regression in Legal Proceedings, 80 COLUM. L. REV. 702 (1980). See also Fisher, Statisticians, Econometricians, and Adversary Proceedings, 81 J. AM. STATISTICAL A. 277 (1986) (discussing the problems that arise and standards of behavior that should be employed when statisticians act as witnesses); Rubinfeld & Steiner, Quantitative Methods in Antitrust Litigation, 46 LAW & CONTEMP. PROBS. 69 (1983) (examining the use of statistics and econometrics in the study and practice of antitrust law).
68. Multivariate regression analysis produces quantitative estimates of the effects of different factors on the same variable. For an excellent introduction to regression analysis, see Fisher, Multiple Regression in Legal Proceedings, 80
69. Multivariate regression analysis produces quantitative estimates of the effects of different factors on the same variable. For an excellent introduction to regression analysis, see Fisher, Multiple Regression in Legal Proceedings, 80 COLUM. L. REV. 702 (1980). See also Fisher, Statisticians, Econometricians, and Adversary Proceedings, 81 J. AM. STATISTICAL A. 277 (1986) (discussing the problems that arise and standards of behavior that should be employed when statisticians act as witnesses); Rubinfeld & Steiner, Quantitative Methods in Antitrust Litigation, 46 LAW & CONTEMP. PROBS. 69 (1983) (examining the use of statistics and econometrics in the study and practice of antitrust law).
70. Variables that take on values only of zero or one are often referred to as "dummy" variables. The coefficients associated with a dummy variable indicate the average difference associated with the dummy characteristic (for example, tester type), controlling for other multivariate effects. The variables TWM and SWM have been suppressed to avoid linear dependence in the data matrix. Interaction terms including these variables were also suppressed. intercept term of the regression equation expresses the average offer that a white male ceived from a white male salesperson. The race and gender dummies reflect deviation from this bench-mark result contingent on the race and gender of tester and salesperson.
71. Information on neighborhood characteristics was taken from CITY or CHICAGO, CHICAGO STATISTICAL ABSTRACT (1984).
72. Formally, the linear regression would also include error terms that are assumed to be independent and normally distributed. Because the same testers visited more than one dealership it might be possible alternatively to model the error structure and test for individual tester effects. See supra note 36.
73. 1 (294·54 + 219.51 = $514.05). These average markups differ from those reported above, see Table 2, supra p. 832, because a smaller sample size of 119 was used. The smaller sample size was chosen to restrict the data to paired sets of observations. The means for this smaller sample reveal more discrimination against white women and black women but less discrimination against black men than in the larger sample. The discrimination against black male testers is still statistically significant, but the amounts of discrimination are not completely robust to changes in sample size.
74. The intercept coefficient expresses the seller white male dummy variable (SWM) and its effect. There were no black female sellers in the sample.
75. Several of the dealership characteristics are highly multicollinear. For example, rental and housing value variables are closely correlated. The failure of individual characteristics to be statistically significant may be an artifact of this multicollinearity. An alternate way to deal with the problem of multicollinearity would be to test the joint hypothesis that the set of dealership characteristic variables are statistically significant.
76. The dummy coefficients in constrained and unconstrained regressions are significantly different. Model Three indicates that a white female buying from a white male should pay $1015 (778 + 237), whereas Model Two indicates only $799 (603 + 196). Conversely, Model Three indicates that a white female buying from a black male seller should pay $949 (778 + 237 + 194 - 260), whereas Model Two would predict she would pay $898 (603 + 196 + 99).
77. See, e.g., J. RUBIN & B. BROWN, THE SOCIAL PSYCHOLOGY OF BARGAINING AND NEGOTIATION 163 (1975).
78. See supra pp. 833-34.
79. For a tentative explanation, see infra p. 847 & note 84.
80. See G. BECKER, THE ECONOMICS OF DISCRIMINATION 14-15, 39-54 (1957).
81. For a discussion of refusals to bargain in the sale of houses, see supra note 44.
82. In addition, consumer animus might cause salespeople to encourage blacks to leave the dealership quickly to cater to bigoted white customers. The testing indicates, however, that such refusals were rare.
83. Cf. C. GILIGAN, IN A DIFFERENT VOICE (1982) (discussing the differences between male and female moral reasoning and expression). The possibility of role-based prejudice calls into question the appropriateness of using a uniform buyer strategy to test for discrimination. Different consumer types may have "different paths" to the same price. As a theoretical matter, comparing the treatment of uniform testers might therefore overstate the overcharge for most black consumers who may adopt different but equally efficient bargaining strategies. Partly for this reason, it would be useful to combine the results of this study with uncontrolled studies of actual sales. In future research, Peter Siegelman and I plan to survey recent car purchasers to test this "different paths" hypothesis. Even if this hypothesis were correct, however, forcing black or female consumers to conform to a particular societal stereotype represents a form of discrimination.
84. The commission structure of car sales might mitigate the effects of employer animus. Non-bigoted salespeople would not want to sacrifice their commissions just to satisfy an employer's bigoted preferences. Conversely, profit-maximizing employers, eager for their employees to generate sales, would have incentives to prevent employees from indulging their bigotry.
85. See Phelps, The Statistical Theory of Racism and Sexism, 62 AM. ECON. REV. 659, 659 (1972) (arguing that individuals may discriminate based on their previous statistical expence with a group - such as blacks or women - rather than judge people on an individual basis).
86. Richard Posner makes this distinction when he suggests that either revenue-based or cost-based statistical discrimination may cause disparate pay for male and female employees. See Posner, An Economic Analysis of Sex Discrimination Laws, 56 U. CHI. L. REV. 1311, 1319-20 (1989).
87. In the study, however, this basis for disparate treatment was removed, because the testers volunteered that they could finance their own purchases, thereby relieving the dealers of any default risk. If the dealers did not believe the testers' representations that they could finance the purchases, however, statistical discrimination based on inferred default risk might exist. If salespeople distrust a tester's stated interest in "purchasing a car today," they may be unwilling to enter into serious negotiations. The inferences sellers draw from customer representations may vary with the context. For example, federal agents have charged Maryland car dealerships with knowingly aiding the drug trade by accepting cash payments for cars -indicating that at least one dealership took buyers who claimed they could "buy a car today" very seriously. See Duke, U.S. Seizes 48 Cars in P-G. Raid, Wash. Post, July 22, 1989, at A1, col. 2.
88. Similarly, if warranty service were unprofitable, statistical discrimination would dictate that dealerships prefer those consumer groups that made fewer warranty claims. Dealers may also discriminate between consumer groups because of statistical inferences that one group is more likely to make repeated purchases or more likely to refer other consumers to the same dealership. Inferences about the profits from such ancillary sales of other cars or services might alternatively be characterized as cost-based statistical discrimination because the disparate treatment stemming from those inferences will not be eliminated by dealership competition in the new car market.
89. Revenue-based price discrimination persists in other consumer markets as well. For example, universities force undergraduate and graduate students to assist financial aid offices in estimating their ability to pay tuition. See Ayres, Colleges in Collusion, NEW REPUBLIC, Oct. 16, 1989, at 19, 19.
90. Revenue-based statistical discrimination is a form of what Pigou call "third-degree" price discrimination. A. PIGOU, THE ECONOMICS OF WELFARE 240-56 (1920). Under regimes of third-degree price discrimination, the seller divides "customers into two or more independent groups, each of which has its own continuous demand function reflecting quantities sold to that group at alternative prices." F. SCHERER, INDUSTRIAL MARKET STRUCTURE AND ECONOMIC PERFORMANCE 316 (2d ed. 1980) (footnote omitted).
91. Revenue-based statistical discrimination is a form of what Pigou call "third-degree" price discrimination. A. PIGOU, THE ECONOMICS OF WELFARE 240-56 (1920). Under regimes of third-degree price discrimination, the seller divides "customers into two or more independent groups, each of which has its own continuous demand function reflecting quantities sold to that group at alternative prices." F. SCHERER, INDUSTRIAL MARKET STRUCTURE AND ECONOMIC PERFORMANCE 316 (2d ed. 1980) (footnote omitted).
92. A consumer's willingness to pay cannot be larger than his or her ability to pay.
93. The distinction between market demand and firm-specific demand is common to law-and-economic analysis. See, e.g., Landes & Posner, Market Power in Antitrust Cases, 94 HARV. L. REV. 937, 947-52 (1981) (discussing the relationship between market demand, firm-specific demand, and market power). For individual consumers, this is the distinction between a market reservation price and a firm-specific reservation price.
94. The process of bargaining may reveal information to dealers about a particular consumer that facilitates revenue-based discrimination. For example, if a consumer's distaste for bargain"ig is correlated with his or her willingness to pay, the process of bargaining could allow lowvaluing consumers to "signal" their lower reservation prices by bargaining longer. In the language of game theory, this signal can be "credible" because the high-valuing consumers face, by assumption, larger costs of signalling. See E. RASMUSEN, supra note 20, at 205; Ayres, Paying Games with the Law, 42 STAN. L. REV. 1291, 1304-05 (1990). Such signalling does not, however, require sellers to treat one class of buyers differently from another. A seller could choose a uniform concession rate and then let the different-valuing consumers separate them-selves by the way in which they bargain.
95. Game-theory models of bargaining suggest that a seller will charge higher prices to groups with higher bargaining costs. See E. RASMUSEN, supra note 20, at 234; Sutton, Non-Cooperative Bargaining Theory: An Introduction, 53 REV. ECON. STUD. 709, 711 (1986).
96. See supra note 22.
97. See supra note 70 and accompanying text (discussing warranty service or repeat or referred sales).
98. George Stigler has considered whether price dispersion in cars could be attributed to costbased differences in the provision of service. He concluded: "it would be metaphysical, and fruitless, to assert that all dispersion is due to heterogeneity." G. STIGLER, supra note 47, at 172.
99. For example, dealers should give better prices to customers who are more likely to refer other customers to the dealership. If white males are more likely to buy their next car or send their friends to buy from the dealership, profit-maximizing dealers should give them a lower price. In recent testing, see supra note 36, an effort was made to mitigate discrimination based on "post-sale servicing" and "referral." Testers affirmatively told salespersons that they were moving out of state (to California) within a month. However, having more than one tester make this representation at a single dealership increased the likelihood that dealers would suspect a test and so was discontinued.
100. In one extreme case a salesperson angrily told a black tester, "You should walk fast to your car because blacks aren't welcome here." Female testers were repeatedly referred to as "honey," "girl," "cutie," etc. Additionally, salespeople said things like, "You are a pretty girl, so I'll give you a great deal," or made explicit sexual references such as, "We can't drop our pants until it's paid for. " Many of these sorts of comments were made following the conclusion of bargaining, after the salesperson had failed to make a sale.
101. Nine hundred dollars compensation for bigotry seems implausible both because animuscompensating wage differentials in the (more rational) employment context are generally smaller and because slightly less bigoted dealers should be willing to charge less. This $900 represents a total "compensation" for animus that is split between the salesperson and the dealership owner. Consider, for example, a scenario in which the salesperson's compensation is 900C per hour, where "c" (o < c < 1) is the salesperson's commission rate. If the disparate treatment is exclusively predicated upon salesperson animus, the animus compensation for the salesperson is determined by this commission rate. Yet even for values of c as low as 0.10, it still seems unlikely that the salesperson would demand $90 per hour more for selling to (associating with) a black female than a white male.
102. See supra pp. 840-41.
103. Theories of intraracial bigotry do not explain why white male testers would receive their worst deals from white salesmen. That is, those theories do not explain why animus would affect relations within the dominant class. However, the bigotry of white consumers might be one reason black male testers were quoted higher prices by black salesmen. If white consumers are reluctant to buy from black salesmen, to "make their profit quota" black salesmen may be forced to try to earn higher profits from sales to blacks. An economic system of discrimination way thus enlist the victimized class to participate in the discriminatory behavior.
104. See Table 7, supra p. 839.
105. See supra pp. 831-33. Although salespeople might enjoy wasting the time of women or Macks, it is implausible that consumer preferences could have caused salespeople to bargain longer with black males.
106. The multivariate regressions explained only about one-third of the variance in final offer profits. See Table 7, supra p. 839. Although these results are standard for cross-sectional agressions, see, e.g., W. LANDES & R. POSNER, THE ECONOMIC STRUCTURE OF TORT LAW 222 (1987) (reporting that regressions explain only two percent of dependent-variable variance), the unexplained variance indicates either that salespeople use a partially random - what game theorists call "mixed" - strategy when bargaining or that relevant variables have been omitted from the regression.
107. See supra pp. 840-41.
108. Alternatively, salespeople of the consumer's race and gender may be able to play on the consumer's mistaken belief that salespeople of the consumer's race and gender are more trustworthy.
109. Consumers who "fall in love" with a particular car signal the dealership that it has a monopoly with regard to their business and can charge accordingly. Consumer advocates stress the importance of being able to conceal interest in a particular car. See R. BUTTON, supra note 34, at 84-87. One dealer, interviewed informally, espoused a desire to close his showroom in the evening, if his competitors would follow suit. Although forcing consumers to purchase at inconvenient times would seem to reduce the demand for cars, the dealer felt that restricting showroom hours would also reduce the amount of search that buyers undertake. Thus, the dealer believed that although he might not get as many people in his showroom, he would have less competition for those who did arrive. The Federal Trade Commission in fact has recently ruled that "an agreement among car dealers in the Detroit metropolitan area to close dealer showrooms on Saturdays and on three weekday evenings is an unlawful restraint of trade." Detroit Auto Dealers Ass'n, No. 9189 (F.T.C. 1989) (1989 FTC LEXIS 10, *84).
110. Statistical analysis was completed by adding dummy variables for instances of the contingent responses to the earlier regressions (Models Two and Three). See Table 7, supra P- 839.
111. The test-drive coefficient was statistically significant at the two percent level, and the car ownership coefficient was statistically significant at the eight percent level. The R-squared for the revised regression was 44% with 93 degrees of freedom.
112. See G. STIGLER, supra note 47, at 175-76.
113. See BUREAU OF THE CENSUS, U.S. DEP'T OF COMMERCE, STATISTICAL ABSTRACT OF THE UNITED STATES 409 (no ed. 1990).
114. See Conversation with Cliff Winston, Senior Research Fellow, Brookings Institution (Nov. 26, 1990); Mannering & Winston, Brand Loyalty and the Decline of Automobile Firms, table E (Nov. 1990) (unpublished manuscript) (presenting the results of a regression regarding duration between car purchases). For a general discussion of consumer demand, see C. WINSTON & ASSOCIATES, BLIND INTERSECTION? POLICY AND THE AUTOMOBILE INDUSTRY (1987), discussing now government policies for evening the competitiveness of the U.S. automobile manufacturers have affected consumer demand; and Mannering & Winston, A Dynamic Empirical Analysis of Household Vehicle Ownership and Utilization, 16 RAND J. ECON. 215 (1985), which found that U.S. households have maintained preferences for American cars and have retained strong brand loyalties.
115. See Conversation with Cliff Winston, Senior Research Fellow, Brookings Institution (Nov. 26, 1990); Mannering & Winston, Brand Loyalty and the Decline of Automobile Firms, table E (Nov. 1990) (unpublished manuscript) (presenting the results of a regression regarding duration between car purchases). For a general discussion of consumer demand, see C. WINSTON & ASSOCIATES, BLIND INTERSECTION? POLICY AND THE AUTOMOBILE INDUSTRY (1987), discussing now government policies for evening the competitiveness of the U.S. automobile manufacturers have affected consumer demand; and Mannering & Winston, A Dynamic Empirical Analysis of Household Vehicle Ownership and Utilization, 16 RAND J. ECON. 215 (1985), which found that U.S. households have maintained preferences for American cars and have retained strong brand loyalties.
116. See infra p. 856.
117. Steve Salop and Joseph Stiglitz have shown that this form of price dispersion may turn on how costly it is for consumer groups to gather information. See Salop & Stiglitz, Bargains and Ripoffs: A Model of Monopolistically Competitive Price Dispersion, 44 REV. ECON. STUD. 493, 493-95 (1977).
118. Similarly, a strong belief that white males will walk away from high offers might not be tested in equilibrium because salespeople would not want to chance starting the negotiation with a high offer. Deborah Rhode has argued analogously in the employment context: "Once jobs become 'typed' as male or female, socialization processes tend to perpetuate those labels." Rhode, Occupational Inequality, 1988 DUKE L.J. 1207, 1219. It will be difficult to differentiate animus-based discrimination from statistical discrimination when sellers' inferences are systematically mistaken. When sellers' inferences are erroneous, cost- or revenue-based relationships with race or gender will not appear, and research may need to depend on attitudinal studies to distinguish animus from irrationality as the cause of disparate treatment. Indeed, the line between animus and irrationality may blur as animus or fear may at some level cause the irrational inferences. Irrational stereotyping may analytically serve as a bridge between rational statistical discrimination and animus. Empirically, it may prove impossible to distinguish animus-based conduct from irrational stereotyping: both types of conduct are inconsistent with profit maximization.
119. Sellers might need more than a single negotiation session to modify gender- and race-specific bargaining strategies developed over long periods of time. Therefore, that the data do not show a decrease in the magnitude of discrimination between the initial and final offers cannot by itself establish seller irrationality or stereotyping.
120. The following table illustrates the average differences between initial and final offers made to black males and white and black females. The dollar figures in each column represent the amount in excess of the corresponding figures for white males: table presented In contrast to the white female results, the offers to black testers are consistent with rational statistical discrimination, because the amount of discrimination decreased from the initial to the final offer. As sellers make rational inferences from larger samples of information (acquired through the process of further bargaining), there should be a reduction in variance in later rounds of bargaining as well as a reduction in the amount of disparity between final offers to white males and final offers to the other groups. This is similar to Steven Jay Gould's explanation for the decline of the .400 hitter in baseball. See S. GOULD, THE FLAMINGO'S SMILE 218-21 (1985).
121. See supra note 100.
122. See, e.g., G. BECKER, supra note 62, at 38, 70-71 (arguing that competition drives inefficient, bigoted producers from the market). For an enlightening exchange on whether civil rights laws tend to enhance or retard these competitive forces, see Donohue, Is Title VII Efficient?, 134 U. PA. L. REV. 1411 (1986); Posner, The Efficiency and the Efficacy of Title VII 136 U. PA. L. REV. 513 (1987); and Donohue, Further Thoughts on Employment Discrimination Legislation: A Reply to Judge Posner, 136 U. PA. L. REV. 523 (1987) [hereinafter Donohue, Further Thoughts].
123. See, e.g., G. BECKER, supra note 62, at 38, 70-71 (arguing that competition drives inefficient, bigoted producers from the market). For an enlightening exchange on whether civil rights laws tend to enhance or retard these competitive forces, see Donohue, Is Title VII Efficient?, 134 U. PA. L. REV. 1411 (1986); Posner, The Efficiency and the Efficacy of Title VII 136 U. PA. L. REV. 513 (1987); and Donohue, Further Thoughts on Employment Discrimination Legislation: A Reply to Judge Posner, 136 U. PA. L. REV. 523 (1987) [hereinafter Donohue, Further Thoughts].
124. See, e.g., G. BECKER, supra note 62, at 38, 70-71 (arguing that competition drives inefficient, bigoted producers from the market). For an enlightening exchange on whether civil rights laws tend to enhance or retard these competitive forces, see Donohue, Is Title VII Efficient?, 134 U. PA. L. REV. 1411 (1986); Posner, The Efficiency and the Efficacy of Title VII 136 U. PA. L. REV. 513 (1987); and Donohue, Further Thoughts on Employment Discrimination Legislation: A Reply to Judge Posner, 136 U. PA. L. REV. 523 (1987) [hereinafter Donohue, Further Thoughts].
125. Competition should generally cause sellers to charge a uniform price equal to their cost. Price dispersion and supra-competitive pricing has been observed as a persistent phenomenon in some markets with multiple sellers. See Kelman, Trashing, 36 STAN. L. REV. 293, 316-17 (1984) (noting that similar Palo Alto gas stations charged disparate prices for full service and that similar Washington, D.C., photographers charged disparate prices for passport photos).
126. Even if car dealers found it to be more "expensive" on average to enter into transactions with black customers (because, for example, of a higher default risk), it might be socially inefficient to allow dealers to discriminate in that cost-based manner. Cost-based statistical "Crimination imposes a "tax" on all black consumers regardless of their actual individual characteristics and might discourage blacks from efficiently investing in credit-worthiness. See Donohue, Further Thoughts, supra note 102, at 533-34; Schwab, Is Statistical Discrimination Efficient?, 76 AM. ECON. REV. 228, 233 (1986).
127. Even if car dealers found it to be more "expensive" on average to enter into transactions with black customers (because, for example, of a higher default risk), it might be socially inefficient to allow dealers to discriminate in that cost-based manner. Cost-based statistical "Crimination imposes a "tax" on all black consumers regardless of their actual individual characteristics and might discourage blacks from efficiently investing in credit-worthiness. See Donohue, Further Thoughts, supra note 102, at 533-34; Schwab, Is Statistical Discrimination Efficient?, 76 AM. ECON. REV. 228, 233 (1986).
128. Consumers might also prefer going to a dealership with "stated prices" because they could avoid the costs of bargaining for a car.
129. See, e.g., EDMUND'S 1989 NEW CAR PRICES, supra note 24, at 188-89 (reprinting an advertisement).
130. If homogeneous products are sold without price dispersion, a seller's profits are distributed equally across goods sold. Price dispersion implies, however, that some products are sold for higher profit. Price dispersion causes a seller's profits to be disproportionately concentrated in its high mark-up sales.
131. This evidence comes from confidential conversations with car dealers and salespeople.
132. The term "sucker" does not imply that high-markup buyers are irrational or even uninformed. As argued above, see supra p. 849, a willingness to pay a high markup may be rational given high search costs or a high aversion to bargaining. Echoing the explicit sexual languages they used to address female customers, some salesmen use the sexual term "lay-downs" to refer to women who are willing to pay the sticker or near sticker price. See Brown, Sexism in the Showroom?, Wash. Post, Feb. 12, 1989, at H1, col. 1.
133. This is especially true if the additional stated-price sales are drawn from the manufacturer's other dealers. In general, manufacturers should prefer the dealer sales process that extracts the largest amount of consumer surplus, because manufacturers should be able to capture those dealership profits through higher wholesale car prices and higher franchise fees. Although it would seem that manufacturers charging a fixed price to dealers would want to encourage ealers to sell a high volume of cars, manufacturers may find that they can reap higher overall profits by charging higher prices to dealers under a relatively low volume bargaining regime tan under a higher volume stated-price regime.
134. The reaction of rival dealerships could also reduce the profitability of advertising a stated rice. Dealerships that "will not be undersold" could simply match their rivals' advertised price. "he first dealership to advertise lower prices might be providing a public service, but if consumers merely use its advertisements to receive matching offers at other dealerships, the first dealrship's advertising strategy might not generate a sufficient increase in sales volume to be profitable for that dealership.
135. Although statistical theories of discrimination are often couched in terms of inferences bout group means, inferences can also be made about aspects of the probability distribution for a group that can rationally affect behavior. Pedestrians' aversion to young male drivers, for »ample, need not be based on a belief that the average male youth drives recklessly; the recklessness of merely five percent of such drivers would make pedestrians rationally leery of all. This "search for suckers" analysis is similar to models in which seller behavior turns on the proportion of comparison shoppers in the population. See, e.g., Schwartz & Wilde, Intervening in Markets on the Basis of Imperfect Information: A Legal and Economic Analysis, 127 U. PA. L. REV. 630, 647-51 (1979).
136. The sensitivity of sellers to the characteristics of a small percentage of high-markup consumers is analogous to the results of game-theory models. See, e.g., Kreps, Milgrom, Roberts Wilson, Rational Cooperation in the Finitely Repeated Prisoners' Dilemma, 27 J. ECON. THEORY 245, 245-47 (1982) (explaining that a firm might be rationally sensitive to a low probability that rivals will adopt an irrationally competitive stance). As Eric Rasmusen has summarized: The beauty of [this] model is that it requires only a little incomplete information: a small probability [of irrationality]. It is not unreasonable to suppose that a world which contains Neo-Ricardians and McGovernites contains a few mildly irrational tit-for-tat players, and such behavior is especially plausible among consumers, who are subject to less evolutionary pressure than firms. E. RASMUSEN, supra note 20, at 118.
137. See Consumer Fed'n of Am., U.S. Consumer Knowledge: The Results of a Nationwide Test 8 (1990) (on file at the Harvard Law School Library). Similarly, during interviews conducted in confidential litigation research unrelated to this study, prospective jurors were asked whether "most people pay sticker price for their cars." Twenty percent of those surveyed responded "yes."
138. Telephone interview with Stephen Brobeck, Executive Director, Consumer Federation of America (Oct. 15, 1990). The study also revealed that women were 7.2% more likely to answer questions about automobile purchases incorrectly. See Consumer Fed'n of Am., supra note 115, at 9. A belief among black consumers that the sticker price is not negotiable would not be erroneous if black consumers have found that dealers are actually unwilling to bargain with them. The tests conducted in this study, however, indicate that dealers were willing to make at least some price concessions to black testers.
139. The pattern of discrimination uncovered in this Article also creates an opportunity for a different kind of competitive response: entrepreneurs might profitably offer to negotiate on behalf of consumers who might otherwise be forced to pay high markups. This study suggests, for example, that a white male consumer would have a competitive advantage in attaining a lower offer. Several types of transaction costs, however, inhibit the viability of such negotiation services. Consumers may not know the extent to which they may be subject to high markups. Entrepreneurs attempting to market their negotiation services may have a hard time credibly communicating both that a problem exists and that their service provides a solution. After all, a consumer using the service would have difficulty verifying that he or she received a competive price.
140. The process of bargaining does not necessitate the disparate treatment of bargainers. A seller could make uniform initial offers followed by uniform concessions - and let the consumers sort themselves at different accepted prices. Sellers could still adopt tough bargaining strategies by refusing to make concessions some proportion of the time. Avoiding disparate treatment would merely require that sellers adopt what game theorists call a uniform "mixed strategy." See E. RASMUSEN, supra note 20, at 69. A mixed strategy for bargaining would specify the Probability that a seller would make a particular type of concession. Under this scenario, the Process of bargaining would still have a disparate impact on different consumers - less savvy customers or those averse to bargaining would accept the higher initial offers - but there would not be disparate treatment.
141. As stated earlier, §§ 1981 and 1982 do not address discrimination based on gender. See supra p. 821.
142. 42 U.S.C. §§ 1981, 1982 (1988). Although not identical, §§ 1981 and 1982 both originate in § 1 of the Civil Rights Act of 1866, ch. 31, 14 Stat. 27, 27. See General Bldg. Contractors Ass'n v. Pennsylvania, 458 U.S. 375, 384 (1982). More important, courts generally treat the two sections identically, at least with respect to the requirements for establishing a claim of intentional discrimination. See Choudhury v. Polytechnic Inst. of New York, 735 F.2d 38, 43 n.5 (2d Cir. 1984). This Article treats the two sections as interchangeable.
143. 392 U.S. 409 (1968).
144. See id. at 437.
145. Section 1982's prohibition on discrimination in the sale or purchase of property "includ[es] discrimination in modes of negotiation." Newbern v. Lake Lorelei, Inc., 308 F. Supp. 407, 416 (S.D. Ohio 1968). Moreover, although § 1982 is not often used in the context of the sale or purchase of personal property, the text of the statute covers such sales or purchases. See, e.g., Scott v. Eversole Mortuary, 522 F.2d 1110, 1113 (gth Cir. 1975) (finding that § 1982 covers the "attempted purchase of caskets"). Section 1981's prohibition on discrimination in the making of contracts is more extensive, covering "offer[s] to make a contract only on discriminatory terms.' Patterson v. McLean Credit Union, 109 S. Ct. 2363, 2372 (1989) (emphasis added). Pattenon's lasting significance, of course, lies in its holding that § 1981 does not prohibit discrimination in post-formation relations. See id. at 2373. Thus it seems that fair driving plaintiffs could sue under either § 1981 or § 1982. Defendants in the present context might try to escape §1981 or §1982 liability by distinguishing preliminary offers meant as the starting point for negotiations from later, more authoritative offers. This argument is untenable, however, because earlier offers help determine subsequent offers (and in turn the ultimate terms of the contract). Salespeople must believe that their actions in the initial rounds of bargaining affect the final contract price; they would hardly be willing to take the time to bargain for several rounds if they thought otherwise.
146. See, e.g., Patterson, 109 S. Ct. at 2377; Goodman v. Lukens Steel Co., 482 U.S. 656, 664, 669 (1987); General Bldg. Contractors, 458 U.S. at 389, 391; Phillips v. Hunter Trails Community Ass'n, 685 F.2d 184, 187 (7th Cir. 1982). Disparate treatment claims are distinguishable from disparate impact claims under title VII. Accordingly, the Supreme Court in General Building Contractors rejected disparate impact claims under §1981 as uncognizable. See 458 U.S. at 390-91. Disparate impact claims cover facially neutral (employer) practices that have discriminatory effects. They require no showing of intent to establish a prima facie case. See Griggs v. Duke Power Co., 401 U.S. 424, 430 (1971).
147. International Bhd. of Teamsters v. United States, 431 U.S. 324, 335 n. 15 (1977) (emphasis added).
148. In a recent title VII case, a plurality of the Supreme Court found that a prima facie case of disparate treatment is made out whenever "gender played a motivating part in an employment decision" and defined this standard to "mean that, if we asked the employer at the moment of the decision what its reasons were and if we received a truthful response, one of those reasons would be that the applicant or employee was a woman." Price Waterhouse v. Hopkins, 109 S. Ct. 1775, 1790 (1989) (plurality opinion). More recently, Judge Posner addressed the state of mind requirement in a fair housing disparate treatment case. See Village of Bellwood v. Dwivedi, 895 F.2d 1521 (7th Cir. 1990), Judge Posner defined disparate treatment as "treating a person differently because of his race; it implies consciousness of race, and a purpose to use race as a decision-making tool." Id. at 1529-30; see also id. (defining conditions when a person would be "guilty of intentional discrimination, or what is the same thing, of disparate treatment").
149. This standard of intent covers not only differential treatment caused by a defendant's animus toward blacks, but also the conscious use of race as a proxy to further some other legitimate goal. Price Waterhouse, for example, explicitly rejected the use of gender as a proxy for legitimate employment traits, such as aggressiveness: "an employer who acts on the basis of a belief that a woman cannot be aggressive . . . has acted on the basis of gender." 109 S. Ct. at 1790-91.
150. If the actual black testers from this study were to bring suit, they would face the additional hurdle of establishing standing under §§ 1981 and 1982 because they lacked a bona fide intent to purchase. In the fair housing context, however, the Supreme Court has stated that testers need not intend to buy to have standing to bring civil suits under title VIII. See Havens Realty Corp. v. Coleman, 455 U.S. 363, 373-75 (1982). The Eleventh and Third Circuits have extended this standing doctrine to fair housing claims brought under § 1982. See Watts v. Boyd Properties, Inc., 758 F.2d 1482, 1485 (11th Cir. 1985); Meyers v. Pennypack Woods Home Ownership Ass'n, 559 F.2d 894, 898 (3rd Cir. 1977). At least in the Eleventh and Third Circuits, fair driving testers would have a strong argument for standing to sue under § 1982.
151. An example of such rare, direct evidence would be an internal memorandum from the defendant-dealer to his manager explaining that he considered the prospective buyer's race before offering an initial price.
152. 411 U.S. 792 (1973). The McDonnell Douglas allocation of the burdens of proof has been applied in subsequent title VII cases. See, e.g., Texas Dep't of Community Affairs v. Burdine, 450 U.S. 248, 252-56 (1981); see also International Bhd. of Teamsters v. United States, 431 U.S. 324, 335 n.i5 (1977) (reiterating the underlying rationale of the McDonnell Douglas scheme that "[p]roof of discriminatory motive is critical, although it can in some situations be inferred from the mere fact of differences in treatment"). The Supreme Court recently distinguished this type of title VII case from that in which there is direct evidence of subjective discriminatory intent, but there is also direct evidence of other, more legitimate motives (the "mixed motives" case). See Price Waterhouse, 109 S. Ct. at 1787-88. Justice O'Connor explicitly distinguished between the two types of cases in her concurrence. See id. at 1801 (O'Connor, J., concurring in the judgment). The allocation of the burdens of proof are significantly different in the two types of cases. The Supreme Court has ruled that "the McDonnell Douglas/Burdine scheme of proof should apply in § 1981 cases." Patterson v. McClean Credit Corp., 109 S. Ct. 2363, 2378 (1989). For examples of lower federal courts' applying the McDonnell Douglas/Burdine standard to both § 1981 and § 1982 claims, see Phiffer v. Proud Parrot Motor Hotel, Inc., 648 F.2d 548, 551 (9th Cir. 1980) (§ 1982 claim); and Chauhan v. M. Alfieri Co., 707 F. Supp. 162, 165-66 (D.N.J. 1988) (§ 1981 claim), rev'd, 897 F.2d 123 (3d Cir. 1990).
153. As applied to the fair driving context, the plaintiff would have to show that she belongs to a racial minority; that she was as ready, willing, and able to buy a car as similarly situated whites; that despite this fact, the defendant offered her higher prices than those offered similarly situated whites; and that the defendant continued to offer lower prices to similarly situated whites. See McDonnell Douglas, 411 U.S. at 802. Courts have allowed plaintiffs flexibility in demonstrating disparate treatment. See, e.g., United States Postal Serv. Bd. of Governors v. Aikens, 460 U.S. 711, 714 n.3 (1983); International Bhd. of Teamsters v. United States, 431 U.S. 324, 358 n.44 (1977); McDonnell Douglas, 411 U.S. at 802 n.13. For example, in fair housing cases, testers need not seek actual purchase. See Havens Realty, 455 U.S. at 373-75
154. McDonnell Douglas, 411 U.S. at 802. Texas Department of Community Affairs v. Burdine, 450 U.S. 248 (1981), makes clear, however, that the defendant only bears a burden of production. See id. at 252-56.
155. McDonnell Douglas, 411 U.S. at 804.
156. See, e.g., Smith v. Anchor Bldg. Corp., 536 F.2d 231, 233 (8th Cir. 1976) (findings prima facie inference of discrimination "where a black rental applicant meets the objective requirements of a landlord, and the rental would likely have been consummated were he or she a white applicant"); Newbern v. Lake Lorelei, Inc., 308 F. Supp. 407, 417 (S.D. Ohio 1968) (finding a prima facie inference of discrimination "[w]here a Negro offeror . . . meets the objective requirements of a seller-developer . . . and shows further that a substantial number of lots have been sold and none of the buyers is a Negro").
157. See supra pp. 823-24. When a black man or woman suspects that he or she has been the victim of housing discrimination, he or she typically complains to a local fair housing organization. The organization may then dispatch a white tester to observe whether the seller treats him or her differently. Fair housing organizations conduct "audits" in which pairs of trained testers target a particular area or seller. The advance preparation for such audits makes them relatively controlled, but fair housing organizations seldom train and control their testers is rigorously as did this study. Interview with Ellen Shogan, Executive Director, Fair Housing Council of Greater Washington (Nov. 26, 1990).
158. One might argue that more instances of discrimination are necessary in the car buying context because differences in seller behavior are more subjective than the outright refusals to deal or steering that is prevalent in fair housing cases. On reflection, however, this argument cannot be sustained. Quoting two different prices is just as objective a measure of disparate treatment as disparate willingness to deal.
159. See, e.g., Metro Fair Hous. Servs. v. Morrowood Garden Apartments, 576 F. Supp. 1090, 1093 (N.D. Ga. 1983) ("Where a white tester is given substantially different information from that given an otherwise similar black tester, an inference that race was a factor can be drawn."), rev'd sub nom. Watts v. Boyd Properties, Inc., 758 F.2d 1482 (11th Cir. 1985). Of course, a genuine fair driving plaintiff -one not part of a controlled test -will have a harder time showing that he or she was treated differently from similarly situated whites.
160. See, e.g., Galster, More Than Skin Deep: The Effect of Housing Discrimination on the Extent and Pattern of Racial Residential Segregation in the United States, in HOUSING DESEGREGATION AND FEDERAL POLICY 119 (J. Goering ed. 1986); Massey & Denton, Trends in the Residential Segregation of Blacks, Hispanics, and Asians: 1970-80, 52 AM. SOC. REV. 802 (1987); Yinger, Measuring Racial Discrimination with Fair Housing Audits: Caught in the Act, 76 AM. ECON. REV. 881 (1986).
161. See, e.g., Galster, More Than Skin Deep: The Effect of Housing Discrimination on the Extent and Pattern of Racial Residential Segregation in the United States, in HOUSING DESEGREGATION AND FEDERAL POLICY 119 (J. Goering ed. 1986); Massey & Denton, Trends in the Residential Segregation of Blacks, Hispanics, and Asians: 1970-80, 52 AM. SOC. REV. 802 (1987); Yinger, Measuring Racial Discrimination with Fair Housing Audits: Caught in the Act, 76 AM. ECON. REV. 881 (1986).
162. See, e.g., Galster, More Than Skin Deep: The Effect of Housing Discrimination on the Extent and Pattern of Racial Residential Segregation in the United States, in HOUSING DESEGREGATION AND FEDERAL POLICY 119 (J. Goering ed. 1986); Massey & Denton, Trends in the Residential Segregation of Blacks, Hispanics, and Asians: 1970-80, 52 AM. SOC. REV. 802 (1987); Yinger, Measuring Racial Discrimination with Fair Housing Audits: Caught in the Act, 76 AM. ECON. REV. 881 (1986).
163. Assuming §§ 1981 and 1982 could be expanded to include women, this attitude might be particularly prevalent in cases in which the victims of dealership discrimination are women. In my discussions with other academics, I have noted greater resistance to the idea of gender than of race discrimination in car sales. When I used racial examples in presenting my results, my interlocutors more often accepted the conclusion of discrimination; when I used gender examples, my interlocutors more often challenged whether the testers were uniform.
164. The Supreme Court has clearly indicated that the defendant bears only a burden of production, not persuasion, in rebutting the plaintiff's prima facie case. See Texas Dep't of Community Affairs v. Burdine, 450 U.S. 248, 255 (1981).
165. The defendant might also directly rebut the plaintiff's evidence of disparate treatment by putting forth evidence that the test was not controlled in all relevant respects and that sellers gave different offers not because of race but because of some other uncontrolled characteristic on which testers differed. Sellers might persuasively argue, for example, that the higher offers received by black women in this study do not violate § 1981: because black women were paired with white males, sellers could argue that they were discriminating on the basis of the unprotected characteristic of gender and not the protected characteristic of race.
166. Village of Bellwood v. Dwivedi, 895 F.2d 1521, 1531 (7th Cir. 1990). Or, to quote another court in a slightly different context: "it is now understood that under § 1982 . . . there cannot in this country be markets or profits based on the color of a man's skin." Contract Buyers League v. F & F Inv., 300 F. Supp. 210, 216 (N.D. 111. 1969), off'd sub nom. Baker v. F & F Inv., 420 F.zd 1191 (7th Cir.), cert, denied, 400 U.S. 821 (1970).
167. 825 F.2d 458 (D.C. Cir. 1987), rev'd on other grounds, 490 U.S. 228 (1989).
168. Id. at 468-69. Paul Brest has similarly rejected reading a mens rea requirement into a standard for discrimination: "Race-dependent decisions need not be race-conscious, but may reflect unconscious racially selective indifference. Such indifference violates the antidiscrimination principle when its effect is to deny benefits to minority persons, or impose burdens on them, which would not be denied or imposed if they were white." Brest, supra note 9, at 14 (footnote omitted).
169. But see Dwivedi, 895 F.2d at 1532 (stating that "since few of the defendants' customers were white, the defendants had little experience with white customers and may therefore have treated the white testers differently out of ignorance rather than design").
170. See, e.g., Griggs v. Duke Power Co., 401 U.S. 424, 432 (1971); cases cited supra note 124.
171. An example of such a facially neutral practice would be dealers' giving discounts to any customer, black or white, who lived in areas designated as low-maintenance-cost zones. To the extent that poorer neighborhoods tend to have worse roads that wear down cars, blacks would be disproportionately denied discounts. Although the practices revealed in this study are anything but facially neutral, fair driving plaintiffs may have more difficulty establishing disparate treatment than disparate impact. In the title VII context, the Supreme Court has extended disparate impact analysis to subjective hiring decisions because plaintiffs face similar difficulties in establishing "subconscious stereotypes and prejudices." Watson v. Fort Worth Bank & Trust, 487 U.S. 977, 990 (1988) (plurality opinion). The Court explained: [E]ven if one assumed that [subjective] discrimination can be adequately policed through disparate treatment analysis, the problem of subconscious stereotypes and prejudices would remain. . . . If an employer's undisciplined system of subjective decisionmaking has precisely the same effects as a system pervaded by impermissible intentional discrimination, it is difficult to see why Title VIIs proscription against discriminatory actions should not apply. Id. at 990-91.
172. Disadvantaged consumers need bench-mark information in order to realize that they have been harmed. See Felstiner, Abel & Sarat, The Emergence and Transformation of Disputes; Naming, Blaming, Claiming . . ., 15 LAW & SOC'Y REV. 631, 633-35 (1980-81). In the fair housing context, by contrast, black consumers can more directly infer disparate treatment when advertised apartments are suddenly unavailable.
173. To illustrate, a tailor may find it difficult to advertise disparate suit prices for whites and blacks, but he or she may be able to produce a poorer product for black customers or to delay the suit's delivery. Of course, because cars are mass-produced and homogenous, nonprice discrimination in the context of car sales would more likely manifest itself in terms of service.
174. Reinier Kraakman has argued analogously that there is no reason to believe that discounts from fundamental value only occur in those few securities (open-ended mutual funds) for which a bench-mark comparison exists. See Kraakman, Taking Discounts Seriously: The Implications of "Discounted" Share Prices as an Acquisition Motive, 88 COLUM. L. REV. 891, 902 (1988).
175. Another potential target might be race or gender discrimination in intercorporate transactions. Many corporate transactions are individually negotiated because of their idiosyncratic nature. A corporation arguably violates § 1981 if it bargains differently with the agents of other corporations based on the agents' gender or race. Race or gender discrimination against the agents of a corporation may give those agents a cause of action against that corporate "person." Even if such disparate treatment is actionable under current constructions of the civil rights laws, the idiosyncratic nature of corporate transactions makes it difficult for the victims of disparate treatment to infer discrimination.
176. 109 S. Ct. 2363 (1989).
177. Id. at 2374.
178. See id. at 2376.
179. 15 U.S.C. § 45(a)(1) (1988).
180. See, e.g., MASS. GEN. L. ch. 93A (1988).
181. See S. OPPENHEIM, G. WESTON, P. MAGGS & R. SCHECHTER, UNFAIR TRADE PRACTICES AND CONSUMER PROTECTION 701-04 (1983).
182. See, e.g., Mariano, Housing Bias Settlement Sets Record, Wash. Post, Apr. 13, 1990, at B1, col. 2 (describing a settlement by a development company that offered white testers apartments but told black testers that "no apartments were available").
183. The degree to which such statements are accepted as a normal aspect of car sales is reflected in cases involving salesperson "puffing." Puffing, or the making of excessive representations by salespeople during their attempts to sell cars, is often treated by courts as inactionable. See RESTATEMENT (SECOND) OF TORTS § 402B comment g (1965) (stating that "puffing," such as the statement that "an automobile is the best on the market for the price," is not a misrepresentation by the seller).
184. See, e.g., Kalwajtys v. FTC, 237 F.2d 654, 656 (7th Cir. 1956), cert, denied, 352 U.S. 1025 (1957); see also Schmidt & Burns, Proof or Consequences: False Advertising and the Doctrine of Commercial Speech, 56 U. CIN. L. REV. 1273, 1276 (1988) (noting that the Postal Service and the FTC prohibit implicit misrepresentation).
185. Legislative or judicial rules could go beyond such a default rule to establish immutable rules against disparate treatment in bargaining on the basis of gender or race. See Ayres & Gertner, Filling Gaps in Incomplete Contracts: An Economic Theory of Default Rules, 99 YALE L.J. 87, 88-89 (1989). Immutable rules may seem superior, but they are unnecessary because few sellers would contract around a "no-discrimination" default rule by expressly reserving the right to discriminate. If lawmakers instead choose a "discrimination-allowed" default rule for implicit representations, it is possible that some sellers would contract around this presumptive rule in their advertisements by holding themselves out as equal-opportunity sellers. Indeed, this process may be at play in dealership advertisements that proclaim that a particular dealership will not mistreat female customers. See, e.g., Advertisement for Silvei ke Dodge & Leasing, BOSTON WOMAN, Winter 1990, at 7a ("Where you don't have to bring a man along to be treated like a customer."). At a minimum, sellers that opt for such explicit representations should be held liable under existing consumer protection laws for any disparate treatment that contradicts the advertisement's representation.
186. Legislative or judicial rules could go beyond such a default rule to establish immutable rules against disparate treatment in bargaining on the basis of gender or race. See Ayres & Gertner, Filling Gaps in Incomplete Contracts: An Economic Theory of Default Rules, 99 YALE L.J. 87, 88-89 (1989). Immutable rules may seem superior, but they are unnecessary because few sellers would contract around a "no-discrimination" default rule by expressly reserving the right to discriminate. If lawmakers instead choose a "discrimination-allowed" default rule for implicit representations, it is possible that some sellers would contract around this presumptive rule in their advertisements by holding themselves out as equal-opportunity sellers. Indeed, this process may be at play in dealership advertisements that proclaim that a particular dealership will not mistreat female customers. See, e.g., Advertisement for Silvei ke Dodge & Leasing, BOSTON WOMAN, Winter 1990, at 7a ("Where you don't have to bring a man along to be treated like a customer."). At a minimum, sellers that opt for such explicit representations should be held liable under existing consumer protection laws for any disparate treatment that contradicts the advertisement's representation.
187. Contractual theorists have traditionally argued that contractual gaps should be filled with provisions for which the parties would have bargained. See Ayres & Gertner, supra note 161, at 89-00.
188. 109 S. Ct. 2363 (1989).
189. Id. at 2376.
190. Proof of discrimination would most likely be attributable to individual dealerships unless plaintiffs can find "horizontal" collusion among dealerships or "vertical" pressure from manufacturers to discriminate. Disparate treatment by employees would be attributable to the firm under traditional notions of agency. See, e.g., Miller v. Bank of Am., 600 F.2d 211, 213 (9th Cir. 1979).
191. After all, airlines would not stop selling to businesspeople if price discrimination on the basis of inferences about willingness to pay were prohibited.
192. Under the common law, unconscionable contracts are unenforceable. See, e.g., Williams v. Walker-Thomas Furniture Co., 350 F.2d 445 (D.C. Cir. 1965); U.C.C. § 2-302 (1988). Academics have distinguished between procedural and substantive unconscionability. See, e.g., Epstein, A Critical Reappraisal, 18 J.L. & ECON. 293, 301 (1975).
193. 52 Misc. 2d 26, 27, 274 N.Y.S.2d 757, 759 (Dist. Ct. 1966), rev'd on other grounds, 54 Misc. 2d 119, 281 N.Y.S.2d 964 (Sup. Ct. App. Term 1967).
194. 105 N.H. 435, 439, 201 A.2d 886, 889 (1964).
195. See, e.g., S. MACAULEY, J. KIDWELL, W. WHITFORD & M. GALANTER, CONTRACTS: LAW IN ACTION 613 (forthcoming 1991).
196. See supra p. 857.
197. Currently, similar regulations mandate that common carriers contract at filed rates. All common carriers subject to the Interstate Commerce Act, 24 Stat. 379 (1887) (codified as amended in scattered sections of 49 U.S.C.), are required to sell at the rates filed with the Interstate Commerce Commission. See Palmer & DeGiulio, Terminal Operations and Multimodal Carriage, 64 TUL. L. REV. 281, 312-13 (1989). The Robinson-Patman Anti-Discrimination Act, 15 U.S.C. § 13 (1988), similarly restrains price dispersion among common carriers.
198. Currently, similar regulations mandate that common carriers contract at filed rates. All common carriers subject to the Interstate Commerce Act, 24 Stat. 379 (1887) (codified as amended in scattered sections of 49 U.S.C.), are required to sell at the rates filed with the Interstate Commerce Commission. See Palmer & DeGiulio, Terminal Operations and Multimodal Carriage, 64 TUL. L. REV. 281, 312-13 (1989). The Robinson-Patman Anti-Discrimination Act, 15 U.S.C. § 13 (1988), similarly restrains price dispersion among common carriers.
199. Currently, similar regulations mandate that common carriers contract at filed rates. All common carriers subject to the Interstate Commerce Act, 24 Stat. 379 (1887) (codified as amended in scattered sections of 49 U.S.C.), are required to sell at the rates filed with the Interstate Commerce Commission. See Palmer & DeGiulio, Terminal Operations and Multimodal Carriage, 64 TUL. L. REV. 281, 312-13 (1989). The Robinson-Patman Anti-Discrimination Act, 15 U.S.C. § 13 (1988), similarly restrains price dispersion among common carriers.
200. Information about other aspects of how car prices are distributed might also reduce price dispersion. Knowing, for example, the lowest price paid by a consumer (or the prices paid by the lowest 10% of consumers) would alert consumers to any overcharging.
201. See Ayres & Miller, supra note 17, at 1063-64.
202. Id. at 1048. Such a regulatory system would significantly reduce the costs of search for all consumers: [B]uying a new car would be easier and more equitable in a world where retailers revealed their true costs. Consumers armed with information about the retailer's markup would not need to search at as many dealerships -for the simple reason that consumers would have a much better idea when they were getting a good deal. Markup information can thus serve as a dramatic substitute for consumer search. Id.
203. Regulating markup disclosure, after all, would be nothing more that a government mandated Edmund's service that would more systematically give all consumers the informaion that many already discover.
204. See Wiley, Antitrust and Core Theory, 54 U. CHI. L. REV. 556, 558 (1987).
205. Similar arguments might be resurrected at the manufacturing level. Although retailers do not incur large fixed costs in selling cars, manufacturers' fixed costs are substantial. Manufacturers may need to extract the profits from retail price dispersion in order to break even. Manufacturers may be able to extract these dealership high-markup profits through lump sum
206. Most notably, they might attempt to make up for the loss of concentrated high profits with lower trade-in prices or higher interest fees.
207. A. SMITH, AN INQUIRY INTO THE NATURE AND CAUSES OF THE WEALTH OF NATIONS 25 (R. Campbell & A. Skinner eds. 1981) (ist ed. 1776).
208. Cognoscenti will recognize such bargaining as an application of the Coase theorem. See Coase, The Problem of Social Cost, 3 J.L. & ECON. 1 (1960). However, even Coase recognized that transacting is not costless and that bargaining would not always occur. See Donohue, Diverting the Coasean River: Incentive Schemes to Reduce Unemployment Spells, 99 YALE L.J. 549, 549 n.2 (1989); Ellickson, The Case for Coase and Against "Coaseanism," 99 YALE L.J. 611, 612-13 (1989).
209. Cognoscenti will recognize such bargaining as an application of the Coase theorem. See Coase, The Problem of Social Cost, 3 J.L. & ECON. 1 (1960). However, even Coase recognized that transacting is not costless and that bargaining would not always occur. See Donohue, Diverting the Coasean River: Incentive Schemes to Reduce Unemployment Spells, 99 YALE L.J. 549, 549 n.2 (1989); Ellickson, The Case for Coase and Against "Coaseanism," 99 YALE L.J. 611, 612-13 (1989).
210. Cognoscenti will recognize such bargaining as an application of the Coase theorem. See Coase, The Problem of Social Cost, 3 J.L. & ECON. 1 (1960). However, even Coase recognized that transacting is not costless and that bargaining would not always occur. See Donohue, Diverting the Coasean River: Incentive Schemes to Reduce Unemployment Spells, 99 YALE L.J. 549, 549 n.2 (1989); Ellickson, The Case for Coase and Against "Coaseanism," 99 YALE L.J. 611, 612-13 (1989).
211. The vast majority of goods cannot be bargained for: retailers compete for consumers trough a "stated price" that they can change from day-to-day but over which they will not bargain.
212. Interview with a Chicago car dealer (May 1989).