Racial Disparities In Taxicab Tipping

Transcript

Yale Law School Public Law and Legal Theory Research Paper No. 50 Center for Law, Economics and Public Policy Research Paper No. 276 To Insure Prejudice: Racial Disparities in Taxicab Tipping Ian Ayres, Fredrick E. Vars and Nasser Zakariya This paper can be downloaded without charge from: Social Science Research Network Electronic Paper Collection at: http://papers.ssrn.com/abstract=401201 To Insure Prejudice: Racial Disparities in Taxicab Tipping Ian Ayres, Fredrick E. Vars and Nasser Zakariya* Abstract: We collected data on over 1000 taxicab rides in New Haven, CT in 2001. After controlling for a host of other variables, we find two potential racial disparities in tipping: (1) African-American cab drivers were tipped approximately one-third less than white cab drivers; and (2) African-American passengers tipped approximately one-half the amount of white passengers (African-American passengers are 3.7 times more likely than white passengers to leave no tip). Many studies have documented seller discrimination against consumers, but this study tests and finds that consumers discriminate based on the seller’s race. African-American passengers also participated in the racial discrimination. While African-American passengers generally tipped less, they also tipped black drivers approximately one-third less than they tipped white drivers. The finding that African-American passengers tend to tip less may not be robust to including better controls for passenger social class. But it is still possible to test for the racialized inference that cab drivers (who also could not directly observe passenger income) might make. Regressions suggest that a “rational” statistical discriminator would expect African Americans to tip 56.5% less than white passengers. These findings suggest that government-mandated tipping (via a “tip included” decal) might reduce two different types of disparate treatment. First, mandated tipping would directly reduce the passenger discrimination against black drivers documented in this study. Second, mandated tipping might indirectly reduce the widely-documented tendency of drivers to refuse to pick up black passengers. *Ian Ayres is the Townsend Professor of Law at Yale Law School. Fredrick Vars is an associate at the Chicago law firm of Miller Shakman & Hamilton. Nasser Zakariya is a fellow at the Yale Law School Center for the Study of Corporate Law. Please send comments to: [email protected]. This article is dedicated to Underhill Moore and Suzanne Perry. Underhill Moore took to the streets of New Haven during the 1930s to see whether people observed parking meter regulations. See John Henry Schlegel, American Legal Realism and Empirical Social Science: The Singular Case of Underhill Moore, 29 BUFF. L. REV. 195 (1980). Nearly seventy years later, Perry conducted a pilot study of taxi and pizza delivery tipping that was the inspiration and foundation for the present effort. The authors thank Aditi Bagchi, Caroline Harada, Lee Harris, and Ian Slotin for their heroic efforts as auditors. Jennifer Brown, Emma Coleman, Neil Katyal, and seminar participants at Georgetown Law School provided helpful comments. 1 Table of Contents Introduction......................................................................................................................... 1 I. Race and the History of Tipping................................................................................ 4 II. Description of Data .................................................................................................... 7 III. Results...................................................................................................................... 10 A. Lower Tips for Minority Drivers .......................................................................... 11 B. Lower Tips By Minority Passengers..................................................................... 11 C. Driver and Passenger Racial Intersections............................................................ 13 D. Regression Analysis.............................................................................................. 15 IV. Alternative (Non-Racial) Hypotheses...................................................................... 23 A. Censored Data? ..................................................................................................... 23 B. Lower Tips for Minority Drivers .......................................................................... 25 C. Lower Tips by Minority Passengers ..................................................................... 27 V. Why Are Consumers Discriminating? ..................................................................... 31 VI. Normative Implications ........................................................................................... 34 A. Adding Insult to Injury?........................................................................................ 34 B. Service Compris.................................................................................................... 36 1. Implementation ................................................................................................. 36 2. Reducing Driver Discrimination....................................................................... 37 3. Countervailing Effects ...................................................................................... 46 Conclusion ........................................................................................................................ 48 2 Introduction It has become increasingly common to test whether sellers in retail markets discriminate against buyers.1 But this paper, to our knowledge, is the first to test the other side of the market.2 We test whether retail consumers discriminate against sellers on the basis of the seller’s race. Even though Gary Becker, long ago understood that consumers’ “taste for discrimination” could cause sellers to discriminate against other customers3 – for example, leading restaurant owners to maintain racially segregated lunch counters – no one has tested whether consumers’ taste for discrimination might be directed at a seller’s race itself (or the race of a seller’s employees). This failure to test should not be surprising. Consumer-side race discrimination tests are hard to implement. Consumer refusals to deal are difficult to document.4 In many contexts, it would be infeasible to use an audit procedure where two different-race sellers made identical offers to an individual buyer. And the ability of consumers to discriminate in the terms of contracting is usually severely constrained. Often consumers’ only promise to pay a non-discretionary price. In contrast, tipping is an unusual and natural place to test for consumer-side discrimination, because it is a Peter Siegelman, Race Discrimination in “Everyday” Commercial Transactions: What Do We Know, What Do We Need to Know, and How Can We Find Out, in A NATIONAL REPORT CARD ON DISCRIMINATION IN AMERICA: THE ROLE OF TESTING (Michael Fix & Margery Austin Turner eds., 1998); see also John Yinger, Evidence of Discrimination in Consumer Markets, 12 J. ECON PERSPECTIVES 23 (1998); IAN AYRES, PERVASIVE PREJUDICE? UNCONVENTIONAL EVIDENCE OF RACE AND GENDER DISCRIMINATION (2001). 2 Some studies have indirectly inferred the presence of consumer discrimination. See, e.g., Clark Nardinelli & Curtis Simon, Customer Racial Discrimination in the Market for Memorabilia: The Case of Baseball, 105 QUARTERLY J. ECON. 575, 576 (1990) (“The appeal of sports for the study of discrimination is that it is possible to separate consumer discrimination from the ability to do the work.”); Lawrence Kahn & Peter Sherer, Racial Differences in Professional Basketball Players Compensation, 6 J. LABOR ECON. 40, 42 (1988) (“[A]ll else equal, white representation on a team contributes to home attendance, providing evidence consistent with the idea of consumer discrimination.”). Employment audits are non-retail tests of whether consumers of labor (i.e., employers) discriminate on the basis of seller race. Keith R. Ihlanfeldt, Madelyn V. Young, Intrametropolitan Variation in Wage Rates: The Case of Atlanta Fast-Food Restaurant Workers, 76 REV. ECON. & STAT. 425, 425 (1994) (“Evidence on discrimination suggests that consumer prejudice affects the wages paid to black workers”); see also John Yinger, Measuring Racial Discrimination with Fair Housing Audits: Caught in the Act, 76 AM. ECON. REV. 881, 881 (1986) (“Housing agents cater to the racial prejudice of current or potential white customers.”). 3 GARY S. BECKER, THE ECONOMICS OF DISCRIMINATION (1971). 4 There have been some important sociological studies analyzing consumer preferences for dealing with sellers of particular ethnic and/or racial groups. See Jennifer Lee, From Civil Relations to Racial Conflict: Merchant-Customer Interactions in Urban America, 67 AM. SOC. REV., 77 (2002); ST. CLAIR DRAKE & HORACE R. CAYTON, BLACK METROPOLIS ([1945] 1993); A.E. McCormick & G.C. Kinloch, Interracial Contact in the Customer-Clerk Situation, 126 J. SOC. PSYCH. 551 (1986). More recently, there has been discussion of the rise of FUBU (for us, buy us) consumerism which at heart is a movement of racecontingent consumer choice. Jerre B. Swann, Sr. et al., Trademarks and Marketing, 91 Trademark Rep. 787, 802 (2001) (“FUBU (‘For Us, By Us’) brand clothing became popular in the African-American community in part by tapping into the sense of cultural unity and ‘authenticity’ that wearing the brand fostered.”). But these studies tend to be qualitative, failing to measure the degree of preference or statistical tests of its significance. 1 1 dimension of consumer economic behavior that is both discretionary and potentially observable. We collected data on over 1000 tips to taxicab drivers in New Haven, Connecticut in 2001. After controlling for a host of other variables, we find two potential racial effects: (1) African-American cab drivers were tipped approximately one-third less than white cab drivers; and (2) African-American passengers tipped approximately one-half the amount of white passengers.5 African-American passengers also seemed to participate in the racial discrimination against African-American drivers. While African-American passengers generally tipped less, they also tipped black drivers approximately one-third less than they tipped white drivers. The propensity to “stiff” – by which we mean to leave no tip – was particularly racialized. African-American drivers were 80% more likely to be stiffed than white drivers (28.3 vs. 15.7%). And African-American passengers were almost 4 times more likely than white passengers to leave no tip at all (39.2 vs. 10.6%). Several caveats, however, are in order before accepting these racial interpretations of the data. First, the data were based on cab drivers’ reports. Cab drivers’ racial stereotypes or preconceptions may have led them to systematically under-report black passenger (or over-report white passenger) tips. Second, we could only crudely control for passenger social class. What we attribute to passenger race may be caused at least in part by a tendency of poorer people to tip less. And third, we do not have strong controls for driver quality. Lower tips by African-American passengers might be explained by a general tendency of passengers to give lower tips for poorer service, coupled with drivers providing inferior service to African-American passengers. And if the white drivers in our study had cleaner cars or for some other reason provided better service, that could explain the race-of-driver disparity. We will respond to each of these alternative hypotheses below by bringing to bear additional pieces of evidence. Audit testing of the participating drivers provides limited evidence that the drivers were providing equal service to passengers of different races. And recent surveys of consumers suggest that even after controlling for consumer income and quality of service, African Americans tend to tip substantially less than whites. In the end, we believe the study provides strong prima facie evidence of the two highlighted racial effects – that is, of passenger disparate treatment against African-American drivers and a lower African-American propensity to tip generally. But this study is far from the final word. In the main body of the Article, we will discuss the tipping behavior with regard to other passenger and driver races. The results with regard to Hispanic passengers are similar to those of African-American passengers. But both for the sake of brevity and because no other races were as well represented in our data, we limit the discussion of the results in the introduction to just whites and blacks. 5 2 If we tentatively accept the finding of a race-of-driver disparity, a natural question to ask is “Why?” The data are not well suited to answer this question, but they do contain some clues. First, the higher propensity of passengers to stiff black drivers seems more consistent with conscious decision making than the less visible tendency of non-stiffing passengers to tip black drivers a lower percentage. Because driver allocation was more or less random, it does not appear that some passengers simply do not tip anybody – rather, there seem to be passengers who are more likely to decide not to tip AfricanAmerican drivers. In contrast, another portion of the disparity may resonate more with unconscious disparate treatment. Passengers tend to round up their tips (to the nearest dollar above their target level) more often when tipping white drivers than when tipping black drivers (32.3% vs. 24.6%).6 When confronted with a last-second decision (based on the final fare) about whether to round up or round down, even passengers who believe they are hard-wired 15% tippers may in practice unconsciously allow the driver’s race to impact their rounding decision.7 When we decompose the overall racial disparity in tips received, we find that racial disparities in stiffing and rounding account respectively for about 27% and 36% of the overall disparity. It is less clear that we should accept the evidence that African-American passengers tend to tip less. But it turns out that this seemingly racial result may nonetheless have important policy implications. While passenger poverty instead of race may really be driving our finding that African-American passengers tend to tip less, it is important to emphasize that, like us, cab drivers also cannot directly observe passenger wealth. But they can observe passenger race and a variety of non-racial measures (such as pickup location, dress, etc.). While our limited data do not allow us to conclusively address whether poverty is the “true” cause of the result, our data do allow us to estimate what kind of statistical inferences a cab driver would make about the size of the likely tip given the observable characteristics of the passengers. Our “statistical” discrimination regressions suggest that “rational” drivers might expect to earn a 56.5% lower tip from an African-American passenger than from a white passenger (after controlling for a host of non-racial observable characteristics). Overall, in our data a driver should expect about 13.8% lower revenue when stopping to pick up an African-American passenger (relative to a white passenger). This result has policy relevance because such driver inferences may play a role in the well-documented refusal to deal with minority passengers.8 The data suggest that at least a portion of driver-side discrimination may be caused, not by animus or by (rational or irrational) statistical inferences about crime, but instead by inferences about how much These percentages are taken over all fares, so that 24.6 percent of all black driver fares are rounded up to the nearest dollar above the passenger’s target level. 7 This finding parallels the results for the Implicit Association Tests (“IATs”) which analogously suggest that unconscious racial influences affect timed sorting decisions. Anthony G. Greenwald et al., Measuring Individual Difference in Implicit Cognition: The Implicit Association Test, 74 J. PERSONALITY & SOC. PSYCH. 1464 (1998); AYRES, supra note 1, at 184. 8 See infra at Part VI.B.2. 6 3 passengers of different races are likely to tip. Indeed, we will show that this revenue effect is orders of magnitude greater than any rational inferences that might be made about the propensities of passengers of different races to rob cab drivers. Our two core racial findings are relevant to a single normative implication. These findings suggest that government-mandated tipping (via a “tip included” decal prominently posted in cabs) might reduce two different types of disparate racial treatment. First, mandated tipping would directly reduce the passenger discrimination against black drivers documented in this study. Second, mandated tipping might indirectly reduce the tendency of drivers to refuse to pick up black passengers – at least to the extent that this driver discrimination is caused by statistical inferences about differences in tipping. There are, however, at least two reasons to pause before eliminating discretionary tipping. First, although research suggests that tips are not strongly correlated with quality of service, tipping (at least in theory) may induce better service. Second, poorer individuals, whose rides are currently subsidized by passengers who do tip, will be less able to afford the increased fare under a mandatory tipping regime. (To the extent minority individuals tend to be less wealthy, this shift would have a disparate racial impact.) The remainder body of the paper is divided into six parts. Part I briefly reviews the role of race in the history of tipping in the United States. Part II describes the data collected for this study. Part III presents the core results – highlighting both the racial and nonracial determinants of cab driver tipping. Part IV considers alternative, non-racial hypotheses. Part V explores what might be causing the customer discrimination. And finally, Part VI discusses normative implications of the findings. I. Race and the History of Tipping Tipping is a substantial component of our economy. More than thirty service professions are regularly tipped.9 Restaurant tips alone in the United States have been estimated at $26 billion a year.10 The tipping norm is now broadly accepted both as a matter of equity – to increase the wages of workers in the service industry – and as a matter of efficiency – to increase the quality of service.11 People tend not to know what percentage of income Michael Lynn et al., Consumer Tipping: A Cross-Country Study, 20 J. CONSUMER RES. 478 (1993). Ofer H. Azar, The Social Norm of Tipping: A Review (unpublished manuscript, 2003), available at http://www.papers.ssrn.com. 11 Uri Ben Zion & Edi Karni, Tip Payments and the Quality of Service, in O.C. Ashenfelter & W.E. Oates, ESSAYS IN LABOR MARKET ANALYSIS 37 (1977), explicitly modeled a repeated interaction between a customer who chooses how much to tip and a service agent who chooses how much effort to provide to show how a marginal reward for effort could induce the service agent to provide more than the minimal effort level. 10 9 4 their parents give to charity, but they know their parents’ tipping percentage.12 Indeed, parents often explicitly tell their children how much they should tip in various settings.13 But what is less well known is that the social practice of tipping was much more controversial 100 years ago. Critics referred to the practice as “un-American” and incompatible with democracy.14 Former Yale Law Professor, William Howard Taft was the “patron saint of the anti-tip crusade”15 and Ralph Waldo Emerson roundly condemned the practice: “I sometimes succumb and give the dollar, yet it is a wicked dollar which by and by I shall have the manhood to withhold.”16 Tipping was attacked as bribery and as “training school for graft.”17 In the early twentieth century 7 states and the District of Columbia passed “anti-tipping” statutes that to varying degrees outlawed the practice.18 Today many patrons and workers in the service industry look upon the tipping practice, not only as non-stigmatizing, but indeed as a worker’s entitlement for work well done. At the turn of the last century, in contrast, tipping was often viewed as a marker of degradation. Both the giving and the receiving of tips were perceived as an acceptance of the recipient’s inferiority.19 In The 12 BARRY NALEBUFF & IAN AYRES, WHY NOT? SIMPLE METHODS OF EVERYDAY INGENUITY (forthcoming 2003). Researchers have explored a variety of server strategies that can enhance restaurant tipping: In one well-known 1984 experiment, researchers found that a waitress who touched her customers, whether male or female, on the hand or shoulder when asking if the meal was all right, raised her tips to 14 per cent, from 11 per cent. . . . So does crouching at the table when taking an order or, if the server is a women, putting a smiley face on the bill. For male waiters, the smiley face cuts tips. William Grimes, The Tip: A Reward, But for Whom?, N.Y. TIMES, at A16 (Feb. 24, 1999). 13 Id. 14 Segrave, supra note 11, at 5-6 (“What, may I ask, is more un-American than tipping? It doesn’t belong in American society; it doesn’t belong in a democracy. It is the product of lands where for centuries there has been a servile class.”). Scott, supra note 11, at 43 (“In the American democracy to be servile is incompatible with citizenship. Every tip given in the United States is a blow at our experiment in democracy.”) 15 Taft An Anti-Tipper, N.Y. TIMES 2 (June 20, 1908) (his barber observed “Never a tip did he give. I understand that he thinks he has paid for the work when he gives the regular price and I guess he is right.”). 16 Regulating Tips, 45 SCRIBNER’S MAGAZINE 252 (Feb. 1909). 17 In 1920 William Rufus Scott launched the Commercial Bribery and Tipping Review. Scott argued that tipping was not only a form of bribery whereby one customer sought unfair advantage over another but a breeding ground for social corruption more generally. Scott wondered whether “a messenger who thought the public owed him gratuities would develop into a man with sound morals.” Segrave, supra note 11, at 43: There is a direct connection between corruption in elections and the custom of tipping. The man who lives upon tips will not see the dishonesty of selling his vote. Scott, supra note 10, 148. Stephanie Cook, A History of 'Handing it Over', CHRISTIAN SCIENCE MONITOR (Oct. 23, 2000) (“[M]any Americans loathed the custom, branding it un-American and undemocratic. The Anti-Tipping Society of America, an alliance of 100,000 traveling salesmen, managed to have tipping abolished in seven states from 1905 to 1919.”); see also KERRY SEGRAVE, TIPPING 29 (1998). 19 The Oxford English Dictionary in 1933 defined tip as “a small present of money given to an inferior.” 18 OXFORD ENGLISH DICTIONARY 134 (1933). “[Tipping] makes the daily income of the worker dependent upon his subservience to the passing humor of the customer. It promotes fawning and sycophancy, and kills dignity and independence.” SEGRAVE, supra 18, at 35. 18 5 Itching Palm, a 1916 manifesto against the practice, William R. Scott said that tipping is “[the] willingness to be servile for a consideration.” This degradation conception of tipping was intimately tied to race. Kerry Segrave has taken the lead in excavating this history: [A] Gunton’s Magazine article in 1896 . . . remarked that in the United States “we have been comparatively free from this offensive semi-mendicancy.” However, tipping was a growing custom among a certain class of laborers such as domestic servants, coachmen, barbers, waiters, and railroad porters: “It will be observed that these occupations are nearly all filled by foreigners and negroes who for the most part have been reared under the patronizing and semi-feudal influences of paternal or ante-wage condition.” Centuries of slavery had left blacks in menial jobs while Europeans were menial workers due to the “aristocratic, patronizing conditions of Europe.”20 For some, the practice of tipping was intimately connected to the perceived inferiority of African Americans. In 1902, for example, a Southern journalist named John Speed remarked: I have never known any but Negro servants. Negroes take tips, of course; one expects that of them -- it is a token of their inferiority. But to give money to a white man was embarrassing to me. I felt defiled by his debasement and servility. Indeed, I do not know how any native-born American could consent to take a tip. Tips go with servility, and no man who is a voter in his country by birthright is in the least justified in being in service.21 The modern tipping norm was incubated in a history rife with explicit racism -- as can be seen in the public prominence given to a seemingly insignificant vignette, again reported by Segrave: In 1907 Senator Tillman of South Carolina tipped black porter George Hollister 25 cents as he departed Omaha, Nebraska's Paxton Hotel. Tillman was well known for maintaining that he never “tips a nigger.” Familiar with the senator's views, Hollister joked that he would have the quarter made into a watch charm. So newsworthy was the event that the New York Times editorialized that Tillman's previous position was “not meanness. We must assume that the Senator abstains from tipping as a matter of principle.” Tipping blacks was not unusual then, and people on long journeys, said the editor, frequently were forced “to convert themselves into fountains playing quarters upon the circumambient Africans.”22 The practice of tipping – far from being perceived as a way of increasing the pay of service workers -- was frequently seen as an employer strategy for exploiting its workers, 20 21 SEGRAVE, supra note 12, at 6. John Gilmer Speed, Tips and Commissions, 69 LIPPINCOTT’S MAGAZINE 748 (June 1902). 22 SEGRAVE, supra note 12, at 11. 6 particularly black workers. The Pullman Company in particular was repeatedly singled out for fostering the tipping norm for its all black workforce as a way of economizing on its wage bill. In 1914 when the Railroad Commission of California asked why the company only hired blacks from the south, an executive explained, “the southern Negro is more pleasing to the traveling public. He is more adapted to wait on people with a smile.”23 The St. Louis Republic newspaper concluded: It was the Pullman Company which fastened the tipping habit on the American people and they used the negro as the instrument to do it . . .24 Pullman made public the fact that its African-American porters were poorly paid so the public would pay them instead. When the Pullman porters organized into the Brotherhood of Sleeping Car Porters in 1925, one of the first things they did was to petition the Interstate Commerce Commission asking for an order prohibiting tips. The petition in part said: Only Negroes, many of them ex-slaves, were employed as porters. This caused the work to be looked on as menial and servile and led to the giving and taking of gratuities.25 That porters would ask the ICC to prohibit a form of their compensation is remarkable. True, a tipping prohibition would put pressure on the Pullman Company to pay higher wages. But it is harder to imagine why the increase in wages would more than offset the loss of tipping income.26 To our minds, the petition might have reflected a non-economic motive. Even if the prohibition would not increase their net incomes, the black porter’s union might have wanted it nonetheless – possibly to increase dignitary dimensions of their employment. They too may have seen the receiving of tips as “a token of their inferiority” and wanted to move away from an equilibrium where they had to scrape and bow for their living. This brief detour does not begin to serve as a full history of tipping practices or its intersections with race. But it may somewhat destabilize and problematize current norms about the inherent desert of service workers to tips. We will pick up this history again both when we discuss the tendency of minority passengers to tip less than whites and when we discuss our proposal to discourage tipping. II. Description of Data In April and May 2001, we collected tipping data from 1066 surveys completed by 12 different New Haven medallion taxicab drivers (six black men, four white men, two 23 24 WILLIAM R. SCOTT, THE ITCHING PALM: A STUDY OF THE HABIT OF TIPPING IN AMERICA 105-07 (1916). Id. at 111-12. 25 Porters Assail Tipping, N.Y. TIMES, § 10, at 3 (Nov. 24, 1927). 26 One possibility is that making transparent the low net income of the porters would induce government to increase their pay. 7 “other minority”27 men). Like most other localities, New Haven regulates both the number of taxies on the road and the price the taxis can charge.28 The taxis are common carriers who have a duty to “service all” customers.29 There are about 140 medallions in New Haven, predominantly owned by the MetroCab Company. With only a few exceptions, the cabs in our study were leased from MetroCab on a fixed cost basis – meaning that the drivers were the residual claimants of all revenues including tips.30 The drivers were instructed to complete the surveys immediately after dropping off their passengers and were paid one dollar per survey. The amount tipped was calculated as the difference between the amount due (“the fare”) and the total amount paid by the passenger to the driver. Drivers reported information on passenger and driver profiles, including sex, race, age, passenger dress (as a proxy for wealth), and driver experience. Drivers were also asked to indicate if they knew the passengers, if the passengers were regular clients, if conversation took place between them, if the pick-up was in response to a call and if the passenger paid cash. Other data included pick-up and drop-off neighborhoods, travel times, day of week, time of day, temperature, and weather. Table 1 reports summary statistics for this survey dataset. In this “other minority” racial category, one of the cab drivers self-reported his race as being “Arab (Franco-Moroccan)” and the other reported his race as being “Asian (Indian).” The racial composition of New Haven cab drivers (who are predominantly white or black) differs markedly from that of New York or Washington D.C. J. M. Shenoy, African Americans Flag Down NY Cabbies (Nov. 6, 1999), available at www.rediff.com/news/1999/nov/06us1.htm (“About 40 per cent of New York’s 25,000 drivers of yellow and livery cabs are from the Indian subcontinent.”). 28 See Lee A. Harris, Taxicab Regulation: The Ever-Elusive Freedom to Contract for a Ride 11 (unpublished manuscript 2001): [T]he main constraints in the taxicab industry can be schematized, loosely, into two categories— entry controls and price restrictions. First, regulation in the taxicab industry often takes the form of entry constraints, via a medallion system, like the New York model. [Second,] regulations take the form of price controls. In New York, London, and Washington DC, the amount charged for a given trip is determined by local regulatory agencies. In Indianapolis, by contrast, only a maximum amount is prescribed by law. The artificial restriction of supply has caused New York medallions to be worth well in excess of $150,000. Marcus Cole has suggested that the medallion system reduces competition and allows drivers more leeway in refusing to pickup minority passengers. Marcus Cole, Medallion Monopoly Drives Taxicab Racism, 9 Liberty & Law 4 (Feb. 2000). 29 In many jurisdictions, the “service all” rule is enforced by prohibiting taxicab drivers from asking the destination of passengers until they have entered the vehicle. Id. at 22. In New Haven, dispatchers inquire as to destination, but do not pass that information along to the driver. Some drivers in the study, however, circumvent the dispatchers by distributing personal business cards with cell phone numbers. During our testing process, one such driver declined to pick up a tester after inquiring as to drop-off location. 30 Drivers tend to be paid either via “a ‘fixed cost’ system, where drivers are paid an amount over a certain payment to the taxicab firm, or (to a far lesser extent) a commission system, where drivers are paid a percentage of the profits.” Id. at 5 n.16. 27 8 Table 1. Summary Statistics Continuous Variables Tip ($) Tip as % of Fare Stiffing Rate Travel Time (min) Travel Distance (mil) Average Speed (mil/hr) Amount Due ($) Amount Paid ($) Temperature (Fº) Passenger Age* Driver Age Driver Exp (wks) Obs 1059 1059 1059 952 943 937 1064 1059 1044 1039 1016 1066 Mean 1.22 0.16 0.24 9.94 4.59 27.12 9.26 10.48 52.25 32.77 39.78 62.69 Std. Dev 2.25 0.27 0.43 11.89 7.73 20.38 11.51 12.86 10.38 13.45 7.95 53.47 Min -5 -0.14 0 1.5 0.2 3.20 2.5 3 28 5 24 2 Max 27.5 5.22 1 200 90 180 150 170 80 85 51 192 Indicator Variables** Obs Percent*** Mean Tip Mean Tip% Passenger Sex: Female 506 47.47 0.97 15.5% Male 443 41.56 1.46 16.6% Passenger Race: Asian 95 9.06 1.02 16.2% Black 319 30.41 0.60 9.2% Hispanic 138 13.16 0.81 12.0% Other 14 1.33 0.84 10.7% White 483 46.04 1.82 21.6% Driver Race: Black 517 48.5 1.02 12.6% Other 99 9.29 0.76 12.4% White 450 42.21 1.54 20.3% Passenger Dress: Below Average 146 13.7 0.79 12.8% Average 858 80.49 1.16 15.1% Above Average 41 3.85 1.74 21.8% Respond to Call 648 68.9 1.24 15.9% Luggage 196 18.39 1.61 15.3% Regular Customer 188 17.64 1.87 25.0% Acquaintance 254 23.83 1.85 24.6% Conversation 720 67.54 1.37 17.3% Rain or Snow 27 2.53 2.28 37.8% Cash 935 98.1 1.21 16.5% *For multiple passenger rides, drivers were instructed to record information only for the passenger who paid the fare **Among the variables not listed here are indicators for neighborhood and addresses, days of the week, times of day, fare types (on the dollar, 25¢ 50¢ and 75¢ above the dollar), individual drivers, and categorical variables derived from continous variables ***Variable percents may not add to 100 if variable is undefined for any observations Overall, the average tip was $1.22, and the average tip as a percentage of fare was 15.8%. 23.8% of the passengers left no tip (the mean “stiffing rate”). The data contain substantial numbers of both male and female passengers and is also well balanced with 9 regard to black and white driver observations (N = 517 and 450, respectively),31 which aids greatly in developing statistically reliable tests of whether consumers discriminate in tipping. Unlike New York or Washington cab drivers who sometimes obtain a substantial portion of their fares from passengers who hail them from the street, New Haven cab drivers obtain fares predominantly by responding to a radio call or by waiting at a designated stand (for example, at the New Haven train station or airport). More than two thirds of our observations come from responses to a dispatcher’s call and the remainder are mostly from drivers waiting their turn at cab stands. Because both the dispatcher and cab stand calls are distributed on a queued basis and because New Haven drivers do not have as much discretion (as hailed cab drivers) in turning down fares, the structure of service provision tends to randomize driver-customer pairings. The tendency toward randomization increases the power of our test of consumer discrimination. But we emphasize that this randomization effect is not perfect. Some fares are generated by direct (cellphone) requests. Moreover, Table 2 shows that passenger races were not in fact randomly distributed across driver races: Table 2. Driver and Passenger Race Frequency Passenger Race Driver Race White Black Hispanic Asian White 50.7% 25.6% 13.0% 9.4% Black 44.2% 35.1% 12.5% 7.1% Other 34.3% 28.3% 17.2% 17.2% Number of Fares 483 319 138 95 Pearson Test of Independence: chi2(8) = 25.9082 Pr = 0.001 Other 1.3% 1.0% 3.0% Number of Fares 446 504 99 1049 14 Black drivers were more likely than white drivers to have black passengers (35.1 vs. 25.6%) and white drivers were more likely than black drivers to have white passengers (50.7 vs. 44.2%). But drivers of each race were still exposed to substantial numbers of black, Hispanic and white customers (which again aids in testing for the existence of customer discrimination).32 III. Results While we will ultimately rely on multivariate regression analysis, our central results are suggested by simply calculating the race-specific means for the tip amount. 31 This did not occur by chance – drivers were recruited with the goal of achieving balance between black and white drivers. Accordingly, the data are not necessarily representative of the racial mix of cab drivers in New Haven. In this regard, it should be noted that more than half of drivers approached (while queued at the train station) declined to participate in the study. 32 There was a substantially higher percentage of African Americans and Asians among our passengers (30.4% and 9.1% respectively) than is found in the Greater New Haven population (11.2% and 2.4% respectively) – and there was a substantially lower percentage of whites (46% in our sample vs. 74.8% in the general population). See Census Table DT_DEC_2000_SF3_U_DATA1 (2003). 10 A. Lower Tips for Minority Drivers Table 3 summarizes the central evidence of consumer discrimination: Table 3. Average Tips & Tipping Percentage by Driver Race Race Disparity Ratios*: Driver Race Avg. Tip Avg. Tip % tip tip% White $1.54 20.3% Black $1.02 12.6% 0.66 0.62 Other $0.76 12.4% 0.50 0.61 *Disparity here is defined as the given statistic divided by the white statistic White drivers are tipped substantially more than black (or other) drivers whether measured either in terms of the average tip amount or the average tip percentage. White drivers were tipped 61% more than black drivers (20.3 vs. 12.6%) and 64% more than our “Other Minority” drivers (20.3 vs. 12.4%). Put simply, passengers systematically tipped white drivers substantially more than non-white drivers. Disparate treatment in tipping can also be seen in the disparate propensities to stiff drivers of different races. Table 4. Stiffing Rate by Driver Race Race Driver Stiffing Disparity Race Rate Ratios*: White 15.7% Black 28.3% 1.80 Other 36.4% 2.31 *Disparity here is defined as the given statistic divided by the white statistic Table 4 shows that the rate at which white drivers were stiffed (15.7%) was far less than that of non-white drivers (28.3 and 36.4% for black and “Other Minority” drivers, respectively). Black drivers were 80% more likely to be stiffed than white drivers (and our “Other Minority” drivers were 131% more likely). B. Lower Tips By Minority Passengers The second racial effect that can be seen in the aggregate data concerns the propensity of different racial groups to tip different amounts. 11 Table 5. Average Tips & Tipping Percent by Passenger Race Passenger Race Disparity Ratios*: Race Avg. Tip Avg. Tip % tip tip% White $1.82 21.6% Black $0.60 9.2% 0.33 0.42 Hispanic $0.81 12.0% 0.44 0.55 Asian $1.02 16.2% 0.56 0.75 Other $0.84 10.7% 0.46 0.49 *Disparity here is defined as the given statistic divided by the white statistic Table 5, for example, shows that the average tipping percentage of African-American passengers was only 42% of the average tipping percent of white passengers (9.2 vs. 21.6%). The tipping percentage of 138 Hispanic passengers was only slightly more than half of the white passenger tipping percentage (12.0 vs. 21.6%). And Asian passengers tipped only 75% of the white passenger tipping percentage (16.2 vs. 21.6%). And, as before, the racial disparity in the average tip amount or tip percentage can also be seen in the different propensities of passengers to stiff drivers. Table 6. Stiffing Rate by Passenger Race Race Disparity Passenger Stiffing Ratios* Race Rate White 10.6% Black 39.2% 3.69 Hispanic 34.3% 3.23 Asian 15.8% 1.49 Other 35.7% 3.36 *Disparity here is defined as the given statistic divided by the white statistic Table 6 shows that African-American passengers are 3.7 times more likely to stiff (and Hispanic passengers are 3.2 times more likely) than white passengers (39.2% and 34.3% stiffing rates for blacks and Hispanics respectively vs. only a 10.6% stiffing rate for white passengers). It is important to emphasize, however, that the racial disparities uncovered in Tables 5 and 6 (as well as the racial disparities uncovered in Tables 3 and 4) are just first cuts at the data. These tables do not control for other variables – such as the socioeconomic status of the passenger or the service quality of the driver – that might be driving the result. These issues and others will be explicitly discussed below in considering alternative hypotheses (see infra Part IV). But, as emphasized in the introduction, these passenger race results may have policy relevance even without controlling for any non-racial factors. An irrational statistical 12 discriminator who sees only passenger race would be apt to infer that Black or Hispanic passengers tip only half as much as whites – and this inference might lead this kind of driver to discriminate against these passengers. We will return to this issue more formally in Part VI. C. Driver and Passenger Racial Intersections A natural question to ask is whether minority passengers participate in tipping black drivers less than white drivers. Table 7 suggests that they do. Table 7. Average Tipping Percent by Passenger and Driver Race Passenger Race Disparity Race Driver Race Avg. Tip % Observations Ratios* White 26.7% 224 White Black 17.9% 222 0.67 Other 13.2% 34 0.49 White 11.0% 112 Black Black 7.4% 176 0.67 Other 13.1% 28 1.19 White 17.5% 57 Hispanic Black 7.1% 63 0.41 Other 11.3% 17 0.65 White 16.1% 42 Asian Black 18.1% 36 1.12 Other 12.3% 17 0.77 White 14.8% 6 Other Black 11.0% 5 0.74 Other 1.8% 3 0.12 * Disparity is defined as the given (Black or Other) statistic divided by the white statistic Indeed, Table 7 shows that measured on a percentage basis the driver race disparity was nearly identical for black and white passengers. White passengers tipped white drivers 49% more than black drivers (26.7 v. 17.9%), while black passengers tipped white drivers 48% more than black drivers (11.0 v. 7.4%). To be clear, black passengers generally tipped less than white passengers, but black passengers tipped black drivers even less. Thus, at least with regard to the two largest racial groups in our sample,33 there seem to be largely independent passenger and driver race effects.34 33 Observations from Hispanic passengers produced the largest driver disparity – with the tipping percentage for white drivers being 146% higher than the tipping percentage given to black drivers (17.5 vs. 7.1%). Asian passengers were the only group to tip African-American drivers a higher percentage than whites (18.1 and 16.1% to black and white drivers, respectively). 34 Table 8 reports the analogous numbers concerning the average amount tipped: 13 The tendency of minority passengers to join in the larger pattern of passenger discrimination against minority drivers can also be seen in an analysis of stiffing rates. Table 9. Stiffing Rate by Passenger and Driver Race Race Disparity Driver Race Rate Observations Ratios* White 6.3% 224 White Black 12.2% 222 1.95 Other 29.4% 34 4.71 White 33.0% 112 Black Black 43.2% 176 1.31 Other 39.3% 28 1.19 White 22.8% 57 Hispanic Black 42.9% 63 1.88 Other 41.2% 17 1.81 White 9.5% 42 Asian Black 13.9% 36 1.46 Other 35.3% 17 3.71 White 33.3% 6 Other Black 20.0% 5 0.60 Other 66.7% 3 2.00 *Disparity here is defined as the given (Black or Other) statistic divided by the white statistic Table 9 shows that white passengers are almost twice as likely to stiff black drivers as white drivers (12.2 vs. 6.3%). And again, the behavior of minority passengers qualitatively mirrors this disparity. From Table 6, we already know that AfricanAmerican passengers are more likely to stiff drivers than white passengers, but Table 9 shows that African-American passengers are almost a third more likely to stiff black Table 8. Average Tips by Passenger and Driver Race Passenger Race Disparity Race Driver Race Avg. Tip Observations Ratios* White $2.21 224 White Black $1.55 222 0.70 Other $0.97 34 0.44 White $0.72 112 Black Black $0.49 176 0.68 Other $0.75 28 1.04 White $1.11 57 Hispanic Black $0.59 63 0.53 Other $0.57 17 0.51 White $0.84 42 Asian Black $1.43 36 1.70 Other $0.63 17 0.75 White $1.04 6 Other Black $0.90 5 0.86 Other $0.33 3 0.32 * Disparity is defined as the given (Black or Other) statistic divided by the white statistic The results of Tables 7 and 8 are qualitatively the same. Again, black and Hispanic passengers, like whites, tipped black cab drivers less. 14 drivers as white drivers (43.2 v. 33.0%) and Hispanic passengers are 88% more likely to stiff black drivers than white drivers (42.9 v. 22.8%). D. Regression Analysis The disparities reported in the previous tables, however, are provided only for heuristic purposes. Without more, we would not know whether the results would be statistically significant, or whether the racial results would hold up once we controlled for a host of non-racial factors that might influence the amount that passengers tip. This section corrects for these deficiencies by offering a series of regressions testing whether the foregoing racial effects persist in a more nuanced analysis. Table 10 reports the result of four nested regressions that relate the tipping percentage to the passenger and driver race as well as an increasing number of non-racial right-hand side variables. 15 Table 10. Regressions with Tipping Percentage as Dependent Variable Constant** Racial Effects: Driver Black Driver Other Passenger Black Passenger Hispanic Passenger Asian Passenger Other Other Variables:*** Passenger Female Passenger Age Below Average Dress Above Average Dress Weeks Driving Cab Survey Experience Driver Age Conversation (1=yes) Repeat Passenger (1=yes) Acquaintance (1=yes) Multiple Passengers (1=yes) Dispatched Pick-up (1=yes) Amount Due Amount Due Squared Fare 25¢ Fare 50¢ Fare 75¢ Cash (1=yes) Travel Time Travel Distance Average Speed (Travel Time/Travel Distance) Night (Between 7PM and 7AM; 1=yes) Late (Between 11PM and 5 AM; 1=yes) Temperature Rain/Snow (1=yes) Luggage (1=yes) Pick-up Nghbd with Below Average 911 Calls ‡ ‡ ‡ ‡ † 1 0.252 -0.067 -0.067 -0.117 -0.093 -0.053 -0.107 Mean 32.77 Std. Dev 13.45 2 0.379 -0.095 -0.166 -0.093 -0.061 -0.044 -0.058 -0.004 0.041 -0.034 0.032 0.000 -0.015 -0.035 0.029 0.039 0.110 0.038 0.010 -0.108 0.062 0.037 0.051 0.001 0.002 -0.026 0.050 -0.020 0.017 0.012 -0.012 0.206 0.020 3 0.289 -0.092 -0.158 -0.090 -0.065 -0.048 -0.051 -0.004 0.042 -0.035 0.033 0.000 -0.009 -0.040 0.023 0.047 0.112 0.037 0.017 -0.105 0.062 0.036 0.053 0.000 0.007 -0.023 0.045 -0.015 0.025 0.010 -0.008 0.196 0.017 -0.012 0.047 0.044 4* 0.410 -0.091 -0.134 -0.090 -0.042 -0.047 -0.022 -0.010 0.034 -0.020 0.036 0.001 -0.016 -0.045 0.019 0.052 0.100 0.045 0.000 -0.103 0.054 0.057 0.052 0.017 0.002 -0.010 0.033 -0.007 0.025 -0.025 -0.005 0.136 0.008 -0.074 0.025 0.167 39.778 7.947 9.262 218.034 11.506 1123.087 9.940 4.587 0.452 11.890 7.727 0.340 52.25 10.38 Pick-up Nghbd with Above Average 911 Calls Drop-off Nghbd with Below Average 911 Calls Drop-off Nghbd with Above Average 911 Calls 0.026 -0.019 Train Pick-up -0.010 Train Drop-off -0.021 Number of Observationsª 1059 841 841 841 R-Squared 0.061 0.196 0.203 0.298 Random Effects N Y Y Y Hausmann Test NOT SIG NOT SIG NOT SIG Variance Test NOT SIG NOT SIG NOT SIG *Neighborhood dummies were included in this regression -- the coefficients are not reported. **The omitted categories for the indicator variables are Driver White, Passenger White, Average Dress, Pick-up And Drop-off Neighborhood variables with Average 911 Calls. To avoid losing observations and to keep the omitted category pure, indicator variables equal to one for missing data were also included but are not reported. ***For continous variables, the effects of a one standard-deviation change are reported. † ‡ ª Survey Experience is defined on a scale of 1-3, depending on whether the driver was filling out his first, second or third set of surveys Categories are based on Total year 2000 911 calls divided by neighborhood population, with extrapolations to missing data/suburbs. Observations fell out of models for two reasons: (1) incomplete driver surveys, and (2) a mid-study change in survey design replacing type of building with neighborhood for the pick-up and drop-off information Underlined coefficients are significant at the 10% level, coefficients in bold are significant at the 5% level, and coefficients underlined and in bold are significant at the 1% level. The first regression specification (reported in column 1) simply regresses the tipping percentage on passenger and driver racial indicator variables. Because the white driver and white passenger variables are omitted, the constant term (25.2%) equals the predicted tipping percentage for a white passenger tipping a white driver. The coefficients on the remaining variables represent the incremental difference for the specified driver or passenger racial type. For example, the first specification suggests that AfricanAmerican driver tips would be 6.7 percentage points lower than the tips given to a white driver (and the bolded and underlined font indicates that this shortfall is statistically significant at the 1% level). 16 The results of the first specification are consistent with the preceding analysis. The negative coefficients on the driver variables indicate that customers tip a lower percentage to minority drivers than to white drivers and the negative coefficients on the passenger variables indicate that minority customers tip a lower percentage than white customers. For example, the regression suggests that African-American passengers tip 11.7 percentage points less than white passengers. Moreover, the regression lets us see for the first time that both of these types of racial disparities are statistically significant. But the first specification still does not control for a host non-racial factors that might be influencing the tipping percentage. Regression specifications 2 through 4 add successively more right-hand side control variables to test whether the racial effects uncovered in the previous tables (and in specification 1) are merely the byproduct of what econometricians call “omitted variable” bias. Regression 2 adds twenty-six variables related to non-racial demographic characteristics of passenger and driver (such as age and gender) and to characteristics of the ride itself. Regression 3 then adds four more variables related to the crime rate (measured by number of 911 calls per resident) found in the pickup or drop off neighborhoods. Finally, specification 4 adds individual indicator controls for 48 pick-up and drop-up neighborhoods.35 Regressions 2 through 4 also employ a “random effects” estimation method. In this context, a random effects model tries to take into account that different drivers may have idiosyncratic propensities to be tipped a particular percentage. Imagine for example that some random process makes some people better or worse drivers (and hence more or less likely to receive a good tip). A random effects model simultaneously estimates (i) the size of these individual driver random effects and (ii) whether (after controlling for the individual driver random effects) there are still statistically significant driver race effects.36 The variance test reported for these regressions suggest that estimated variance of the individual driver random effects is not statistically different than zero. This suggests that the individual driver effects are not dominant in this dataset. The real action in the data is between races and not idiosyncratic differences within race.37 After controlling for random driver effects and a host of time, manner and place effects, these specifications suggest that the customer discrimination result is quite robust. Adding additional variables to the regression does not materially impact the size or statistical significance of the driver race variables. For the most complete regression (specification 4), we find that black drivers are tipped 9.1 percentage points less than white drivers (and that this result is statistically significant at the 1% level). This regression predicts that an African-American driver would be tipped 43.6% less than a There are 61 neighborhood dummies, but many of these were dropped as a result of multicollinearity. To save space, neighborhood-dummy effects are not reported. 36 The Hausmann tests that are reported for these regressions cannot reject the hypothesis that the random effects regression is appropriate (relative to a fixed-effects regression). 37 However, later in the discussion of alternative hypotheses (part IV), we will return to this issue again. 35 17 similarly-situated white driver,38 and that this tipping shortfall causes the overall revenue per fare for African-American drivers to be 7.0% less than that of white drivers.39 The passenger discrimination imposes the economic equivalent of a 7% tax on the income of black cab drivers.40 The finding that minorities tip systematically less, in contrast, is not as robust to the inclusion of additional right-hand side controls. The size of the coefficients become smaller as additional controls are added and the Hispanic disparity becomes less statistically significant. For the most complete regression (specification 4), the size of the black passenger effect is diminished (from 11.7 percentage points in specification 1) to a 9 percentage point shortfall. While 9 percentage points is still substantial, the large point here is that even specification 4 contains only poor controls for the socio-economic class of the passenger (above or below average dress, and characteristics of pickup or drop off locations). With better controls, the seeming tendency of minorities to tip less might simply become a tendency of poor people to tip less. We will explore this hypothesis below in the alternative hypotheses discussion. But for now it is important to note that the cab driver, like the researcher, will not be able to observe many of these additional variables and, as we will stress below (in the Implications section), thus might be moved to make inferences – including racial inferences – about whom to pick up on the bases of similarly restricted data. The regressions also suggest that customers discriminate against older drivers. Specification 4 shows that a driver whose age is one standard deviation (about 8 years) above the mean driver age (about 40) should expect to receive tips that are 4.5 percentage points less than average. And this disparity is statistically significant at the 10% level (p. < .1).41 For those interested in norms of tipping more generally, the non-racial controls provide a potpourri of interesting results. In contrast to driver age, we learn that older passengers tip a systematically higher percentage. In our data, for example, a passenger whose age was one standard deviation above the mean passenger age was predicted in regression 4 to tip 3.4% percentage points more. The tipping percentage shortfall of 9.0% divided by the predicted white driver tip percentage (evaluated at the means of the non-driver race variables) of 20.9% equals 43.6%. 39 The revenue shortfall of $.76 divided by the predicted white driver revenue (evaluated at the means of the non-driver race variables) of $10.82 equals 7.0%. 40 In simple economic models, the lower revenues available for minority drivers would tend to cause minorities to substitute away from driving cabs where they would not have to pay this discrimination tax. But our conversations with both minority and non-minority drivers suggest that the drivers are not well informed about the discrepancies in customers’ willingness to tip minority drivers. And even if minority drivers learned of their ill-treatment, they may also face restricted opportunities in finding and being compensated for other forms of employment. 41 It is important to note that the model also includes a variable for weeks of driver experience, which yielded a positive but not statistically significant coefficient. Thus, the results suggest that although there may be a slight advantage to experience (all else held equal), the more significant effect is that older drivers at a given experience level receive lower tips. 38 18 We also learn in specification 4 that the tipping percentage is statistically higher: during inclement whether (13.6 percentage points); for acquaintances (10 percentage points); and, for lower fairs (10.3%).42 In contrast, we found no passenger gender effects. Men and women seem to tip roughly the same percentage in all of our specifications. A variant of the foregoing regression specifications can also be used to test whether our earlier finding -- that minority passengers seem to participate in the discrimination against minority cab drivers – is statistically significant and robust to inclusion of other right-hand side controls. The previous regressions do not allow for such a test because the passenger and driver race effects were forced by the specification to enter independently. But it is possible in a less constrained specification to have the regressions estimate effects for all 15 of the possible driver/passenger race permutations. By including these interaction effects it becomes possible to test, for example, whether black passengers tend to tip black drivers less than white drivers. Table 11. Tests of Consumer/Passenger Discrimination against Different Driver Race, by Passenger Race Minority Driver Disparity (Relative to White Drivers) -Difference in Driver Effects: Passenger Driver Regression Regression Regression Regression Race Race 1* 2* 3* 4* -0.088 -0.104 -0.100 -0.101 Black White -0.135 -0.212 -0.208 -0.166 Other -0.079 -0.075 Black -0.036 -0.064 Black Other 0.021 -0.104 -0.095 -0.071 -0.104 -0.112 Black -0.092 -0.093 Hispanic -0.161 -0.152 Other -0.062 -0.149 Black 0.020 -0.062 -0.065 -0.055 Asian -0.186 -0.176 Other -0.038 -0.158 Black -0.038 -0.037 -0.049 -0.094 Other Other -0.130 -0.058 -0.078 -0.070 *These effects are based on regressions of the same form as 1, 2, 3 and 4 where the passenger and driver racial variables are replaced with interacted variables. Underlined coefficients are significant at the 10% level, coefficients in bold are significant at the 5% level, and coefficients underlined and in bold are significant at the 1% level. Table 11 reports the tipping percentage shortfalls for minority drivers (relative to white drivers) from particular passenger races in regression specifications that are otherwise identical to specifications 1 through 4 in Table 10. For example, we see in specification The regression suggests that the tipping percentage increases with amount due squared. But we found that the squared term only began to dominate the linear term for fare amounts due that were out of sample (approximately $186). 42 19 2* that white passengers tip black drivers 10.4 percentage points less than white drivers. The black driver shortfall is large and statistically significant in all of the specifications. Again, there is some evidence that minorities discriminate against black drivers, but the differences are less significant than in the case of white passengers. The reported black disparities are all negative – indicating an estimated shortfall for black drivers, but for the most controlled specification, the black driver shortfalls are insignificant. Most of the other passenger/driver race permutations were statistically insignificant. However, it should also be noted that we may be simply running into a small numbers (or what econometricians sometimes call a “degrees of freedom”) problem. When you ask a regression to estimate 15 separate driver/passenger race effects as well as nearly 100 other right-hand control variable effects it can become increasingly difficult to identify statistically significant effects. To further analyze stiffing behavior, we regressed the stiffing indicator against the same set of independent variables in the tipping percentage specifications. Table 12 reports the results of these regressions, now relating the probability of stiffing to passenger and driver race and the nested set of non-racial right-hand side variables. 20 Table 12. Regressions with Stiffing as Dependent Variable 1 Racial Effects:** Driver Black Driver Other Passenger Black Passenger Hispanic Passenger Asian Passenger Other Other Variables: Passenger Female Passenger Age Below Average Dress Above Average Dress Weeks Driving Cab Survey Experience*** Driver Age Conversation (1=yes) Repeat Passenger (1=yes) Acquaintance (1=yes) Multiple Passengers (1=yes) Dispatched Pick-up (1=yes) Amount Due Amount Due Squared Fare 25¢ Fare 50¢ Fare 75¢ Cash (1=yes) Travel Time Travel Distance Average Speed (Travel Time/Travel Distance) Night (Between 7PM and 7AM; 1=yes) Late (Between 11PM and 5 AM; 1=yes) Temperature Rain/Snow (1=yes) Luggage (1=yes) Pick-up Nghbd with Below Average 911 Calls † † † † 2 0.100 0.194 0.279 0.198 0.087 0.188 0.006 -0.032 0.069 -0.093 0.001 0.001 0.017 -0.016 0.013 -0.068 0.013 0.017 0.034 -0.134 -0.138 -0.161 -0.185 -0.360 0.029 -0.084 0.019 0.073 -0.047 0.030 -0.030 -0.104 3 0.096 0.176 0.273 0.205 0.088 0.180 0.006 -0.035 0.069 -0.087 0.001 -0.007 0.018 -0.005 -0.003 -0.067 0.013 0.006 -0.051 -0.010 -0.127 -0.154 -0.177 -0.395 0.058 -0.114 0.024 0.066 -0.034 0.029 -0.021 -0.099 0.500 -0.038 0.059 4* 0.098 0.239 0.192 0.131 0.102 0.181 0.004 -0.050 0.062 -0.065 0.000 -0.005 0.044 0.015 -0.014 -0.035 0.008 0.010 -0.154 0.088 -0.104 -0.133 -0.151 -0.285 0.110 -0.146 0.039 0.045 0.008 0.023 0.028 -0.077 0.890 0.642 -0.062 0.107 0.223 0.307 0.286 0.061 0.306 Mean 32.77 Std. Dev 13.45 39.778 7.947 9.262 218.034 11.506 1123.087 9.940 4.587 0.452 11.890 7.727 0.340 52.25 10.38 Pick-up Nghbd with Above Average 911 Calls Drop-off Nghbd with Below Average 911 Calls Drop-off Nghbd with Above Average 911 Calls -0.022 0.116 Train Pick-up 0.105 Train Drop-off 0.073 1059 837 837 837 Number of Observationsª 0.119 0.313 0.332 0.416 Pseudo R-Squared *Neighborhood dummies were included in this regression -- the coefficients are not reported. **Coefficients reported here are the changes in the probability of stiffing resulting from infinitesimal changes in the continuous variables and discrete changes in the indicator variables from 0 to 1. For continous variables, the effects of a one standard-deviation change are reported. The omitted categories for the indicator variables are Driver White, Passenger White, Average Dress, Pick-up And Drop-off Neighborhood variables with Average 911 Calls. To avoid losing observations and to keep the omitted category pure, indicator variables equal to one for missing data were also included but are not reported. ***Survey Experience is defined on a scale of 1-3, depending on whether the driver was filling out his first, second or third set of surveys. Categories are based on Total year 2000 911 calls divided by neighborhood population, with extrapolations to missing data/suburbs. Observations fell out of models for two reasons: (1) incomplete driver surveys, and (2) a mid-study change in survey design replacing type of building with neighborhood for the pick-up and drop-off information Underlined coefficients are significant at the 10% level, coefficients in bold are significant at the 5% level, and coefficients underlined and in bold are significant at the 1% level. ª † Since stiffing is a dummy variable, we ran the analogous stiffing regressions using probit models so that the output in Table 12 indicates the change in the likelihood of a stiff on the basis of a discrete 0-1 change in the dummy variables or a one-standard-deviation change in the continuous variables. Thus, in the first specification, the black driver coefficient of 10.7 indicates that a black driver is about 11 percentage points more likely to be stiffed than a white driver. Once again, the results of all specifications are quite consistent with our initial tabulations. Minority passengers are more likely to stiff and minority drivers are more likely to be stiffed. For example, in all specifications, a Black passenger is significantly 21 more likely to stiff than a white passenger, though with decreasing magnitude as we add more controls. In addition, we now see that the majority of these effects are highly significant (prob. < .01). Again, the consumer discrimination effects, now in terms of stiffing, remain robust to the addition of racial and non-racial controls. Although the second and third regressions yield black driver effects that are significant only at the 10% level, the magnitude of the effects actually remains roughly in the area of 10 percentage points. The most controlled specification is significant at the 1% level, indicating that black drivers are almost 10 percentage points more likely to be stiffed than white drivers (while, according to the same specification, Other Minority drivers are almost 24 percentage points more likely to be stiffed). As is in the case of the tipping regressions, the passenger effects tend to diminish with the addition of more controls. Comparing the first and fourth regressions, we see that the black passenger effects diminish by over 11 percentage points and the Hispanic passenger effects diminish by over 15 percentage points -- although the broad levels of significance remain the same.43 Not surprisingly, many of the non-racial controls that were significant in the tipping percentage regressions remain significant in the stiffing regressions.44 Passenger and driver age, above average dress, and acquaintance of driver with passenger yield largely significant effects with the same tendencies as in the tipping regressions.45 So, for example, in the fourth regression, we see that a passenger one standard deviation above the mean age is 5% less likely to stiff while a driver one standard deviation above the mean age is 4.4% more likely to be stiffed. Again, this suggests that passengers may have engaged in discrimination against older cab drivers as well. All in all, the regression analysis suggests that there is strong evidence of customer discrimination against minority drivers measured by tipping and stiffing differences that persists and is statistically significant after controlling for a variety of non-racial factors. And there is some evidence that black and Hispanic passengers participate in this discrimination, although the estimates of minority driver shortfalls in the most controlled regression are less significant. The second finding – that minority passengers tend to tip However, it is again important to keep in mind that none of our specifications contain good controls for the socio-economic class of the passenger so that effects the stiffing regressions are attributing to race may in fact be more dependent on the wealth of the passenger. This and its implications to cab driver inferences will be considered further in Part IV.C. 44 Note, however, that the fare dummies are all negative and highly significant. This indicates that passengers are more likely to stiff with integer fares than with any other type, perhaps in part because letting the driver “keep the change” is an option for non-integer fares. We will discuss this further infra note 68. Also, somewhat surprisingly to our minds, a passenger paying cash is less likely to stiff. Some drivers may have neglected or refused to record a non-cash tip. 45 Though rain and snow are not significant disincentives to stiffing in these regressions, the temperature effects show that passengers are about 2.3% less likely to stiff with every 10% drop in temperature, suggesting that cold/inclement weather does have some prohibitive effect on stiffing. Additionally, passengers are more likely to stiff at night under the cover of darkness (4.5% more likely in the third regression, and significant at the 5% level). 43 22 a lower percentage and stiff more frequently – is confirmed as statistically significant for African Americans in all specifications (but the absolute size of these effects systematically decreases as more non-racial controls are introduced). IV. Alternative (Non-Racial) Hypotheses Although the prior regressions control for a host of non-racial variables, there are still many aspects of the transaction that we do not observe and hence cannot include in the analysis. As is often the case, the omission of right-hand side control variables creates a possibility that the racial effects reported above may in fact be caused by non-racial factors for which we did not control. This section outlines the major alternative hypotheses and assesses with the best data available the extent to which they qualify the two racial results of the last section. But before proceeding to consider the particular omitted variables that may be driving the “minority driver” and the “minority passenger” effects, we first take on more global concerns about the quality of the data reported by the cab drivers. A. Censored Data? It is imported to remember that all the results reported above are based upon surveys filled out by individual cab drivers. Either misreported or censored data would importantly reduce our confidence in the results. A weak indication of survey reliability can be found in the non-significance of the “Survey Experience” variable reported in both the Table 10 and 12 regressions. The coefficients on this variable are both very small and not statistically significant. This indicates that reported tips of the drivers did not vary as they filled out more surveys. If the drivers were misreporting fares, they at least seem to be consistently misreporting them over time. The earlier discussed rejection of individual driver random effects (as evinced by rejecting the hypothesis that the random effects variance was different from zero) also provides some small measure of assurance that drivers were accurately reporting fairs. The random effects regressions suggest that drivers of the same race were treated similarly: If white drivers were misreporting fare data, they seemed to be doing it consistently as a group.46 The possibility that drivers explicitly colluded to misreport is In sharp contrast, an earlier pilot study conducted by Suzanne Perry found that one of the drivers’ surveys had markedly different (and implausible) survey answers. This driver reported that virtually all of his passengers failed to tip. Our impression from interacting with the participating drivers was that each took the study seriously. Most asked questions about the study and several expressed interest in obtaining a copy of the results. One driver returned a subset of his 50 surveys, apologizing that he could not complete the set because he was going to be unable that month to make the lease payment on his cab. 46 23 unlikely, for the simple reason that the drivers in the survey did not have good information on the universe of drivers participating in the study. Nor did the drivers have any obvious motive to act collectively in such a manner. But beyond misreporting of the data, there is also the possibility that drivers reported only a non-random sample of their total universe of fares. If drivers “censored” the transactions that they reported, we would be less sure whether the results of the last section would be robust to analysis of a broader (less censored) sample. Unfortunately, although drivers were instructed to collect data for their “next 50 rides,” there is evidence of driver censoring. To begin, we find that there are a disproportionate number of integer fares reported in the data. Fares (not including tip) are regulated in New Haven to come in 25 cent increments. A full 44.4% of our observations were reported to have final meters equal to integer amounts (e.g., $6.00).47 The length of trips, however, may not be random -- so there are some benign reasons to explain why the trailing digits of the final meter would not be random. But still the disproportionate number of integer fairs – far exceeding 25% of the data – strikes us as some evidence of censoring. But even more directly we found that the drivers stretched their survey answers over several shifts of work. If we restack our observations, we find that our data comes from 105 distinct driver-days (say, driver 1 reporting data from Tuesday, April 12th, 2001). But many of these driver-day observations had relatively few fares. Indeed only 55 of these driver-day observations included ten or more fares. This suggests that at least some of the time our drivers were not reporting the universe of fares encountered on a particular shift. On the other hand, in the process of arranging the distribution and collection of the survey forms it became clear that the drivers worked irregular hours. Working multiple shifts in a single day – centered on, say, morning and evening rush hours – would not be a surprising way for a driver to structure his day.48 This evidence of censoring importantly qualifies the reliability of the foregoing results. It is possible, for example, that drivers were more likely to record the results of outlier fares that made more of an impression on them, or that black drivers were more suspicious of the motives of the white male co-author who solicited their participation in the study and sought to conceal the full magnitude of their tipping income out of, say, fear of being 19.4% of the fares were 25 cents over the dollar; 20.4% of the fares were 50 cents over the dollar; and, only 15.9% of the fairs were 75 cents over the dollar. 48 More direct evidence of this censoring can be found in our failed attempts to audit the reporting of the drivers participating in our sample. We sent a handful of student testers to take cab rides during the period our drivers were filling out forms. It proved to be logistically difficult to put one of the auditors into the cab of a participating driver. We ultimately were able to match testers to drivers for only ten fares. See infra text accompanying note 54 (discussing other aspects of the testing data). We had hoped to check whether the drivers’ surveys matched with the testers’ reports (same fare, same tip, etc.). But none of the ten testers’ fares were reported by drivers in their survey data. Again, this strongly suggests that drivers were not reporting the full universe of fare data during the period in which they were participating in the project. However, a likely cause of censoring would seem to be insufficient time between rides for a driver to fill out the survey. Our testing regime, in which the second of two paired testers caught the same cab immediately after the first tester exited it, may have contributed to this problem. 47 24 audited. (That every driver reported some substantial tips, however, argues against this troubling hypothesis.) If the recorded fares are not representative of the larger universe, then the prior results may not be indicative of the broader tipping experience. However, as one of the first studies of taxicab tipping practices and the first quantitative study of consumer-side discrimination, even qualified results raise important concerns about the possibility of disparate treatment (and at a minimum suggest the appropriateness of further testing). Putting aside for the moment these important censoring concerns, we next turn to the possibility of omitted variable problems with the two core racial results. B. Lower Tips for Minority Drivers Our finding of customer discrimination against minority drivers was both quite stable and robustly significant in Tables 10 and 12’s series of nested regressions. But it is always important to consider whether omitted variables may be driving this disparate-treatment result. Here we pause to consider three possibilities: individual driver effects, disparate driver quality, and disparate customers. Individual Driver Effects. To begin, it is useful to assess whether what we reported as driver race effects might instead be caused by idiosyncratic characteristics of the twelve individual drivers in the data. As an initial matter, we found that our random effects models, controlling for an increasing range of non-racial factors, rejected the presence of individual driver effects (and after attempting to control for them nonetheless found pronounced evidence of customer discrimination). Second, if we simply calculate mean tipping percentage for each of 4 white and 6 black drivers, we find that white drivers garnered three of the four highest tipping percentages, while black drivers garnered three of the four lowest tipping percentages.49 But it is possible in non-random effect specifications to alternatively control for individual driver effects by asking the regression to “cluster” the data by individual driver.50 When we rerun the regressions clustering by individual cab drivers, we still find evidence of customer discrimination but the results are not as statistically significant.51 In the clustering regressions that parallel the first specification in Tables 10 and 12, we still find that black and other minority drivers receive lower percentage tips and are more likely to be stiffed, but the results are only marginally statistically significant.52 One of the white driver’s average tipping percentage was particularly high (30.5%) and one of the black driver’s average tipping percentage was particularly low (5.8%). 50 See STATA USER’S GUIDE 256 (1999). 51 The clustering procedure, by its nature, generates the same coefficients as reported in Tables 10 and 12, but lets the data test whether clustering increases or decreases the level of statistical significance. 52 The Table 10 analog suggests that the Black Driver tipping percentage result is only significant at the 11.9% level, while the Other Minority Driver result was no longer statistically significant. The Table 12 analog suggests that the Black Driver stiffing result is only significant at the 6.9% level, while the Other Minority Driver result is only significant at the 9.5% level. The much smaller pilot study of Suzanne Perry, discussed supra note 46, also was not able to identify statistically significant customer discrimination against minority drivers. 49 25 On net, we still have some lingering concerns about individual driver effects. But using a variety of reasonable approaches that alternatively control for these effects, we still find what seems to be an independent and statistically significant disparity in the tips received by minority and non-minority drivers. Disparate Driver Quality. The driver race disparity might alternatively be explained by potential differences in the quality of service that minority and non-minority drivers provided. If minority drivers provided systematically poorer service than white drivers, then minority drivers may have received lower tips not because of their race, but because passengers may give lower tips for poorer service. We do not have much information to respond to this theoretical possibility. The speed variable and the indicator variable for whether the driver conversed with the passenger crudely control for two dimensions of quality. Also, other studies of tipping generally have found that variation in service quality does not explain a very large percentage of differences in the amounts that people tip.53 These studies would at least suggest that the degree of the racial disparity is not likely to be caused by differences in quality. We also undertook a modest amount of auditing of the drivers in our study to see if there were gross differences in the quality of service they provided.54 Based on a total of only ten audit rides with participating drivers (six rides with white drivers, four rides with black drivers), we did not find support for the hypothesis that minority drivers provided lower quality service. Indeed, our testers subjectively rated the quality of service higher for black drivers than for white drivers (average 4.5 out of 5 for black drivers vs. an average rating of 3.3 for white drivers). This miniscule sample does not allow us to confidently test for quality differences. But when combined with the authors’ own experience of taking numerous cabs in New Haven, we are fairly confident that the driver race effect is not well explained by racial disparities in driver quality. Disparate Customers. Finally, we considered whether the driver-race disparity might be caused by minority drivers serving different types of customers than non-minority customers. Under this hypothesis, minority drivers would receive lower tips than white drivers not because customers discriminate but because minority drivers tend to provide service to low-tipping customers while non-minority drivers tend to provide service to high-tipping customers. As discussed above, there are some structural variables that tend to push New Haven drivers toward a more random selection of customers. Both dispatchers and cab stands purport to operate on a queued basis – allocating the next customer to the next driver in line. 53 Michael Lynn & Michael McCall, Gratitude and Gratuity: A meta-analysis of Research on the ServiceTipping Relationship, 29 JOURNAL OF SOCIO-ECONOMICS 203, 212 (2000) (“Although the average relationship between tip size and service evaluations was statistically significant in this review, it was also quite small—accounting for less than two percentage of the variability in tip percentages.”). 54 See supra note 48 (discussing the driver audits). 26 But there are several dimensions on which non-random allocations of passengers can occur. Dispatchers may, contrary to stated policy, give poorer fares to minority drivers. Drivers may engage in different strategies as to which cab stand they wait at or whether they queue for the next dispatchers call. Waiting at the airport may expose drivers to a different mix of passengers than choosing to wait at the train station or at the Shubert theater.55 And passengers may directly call a driver to schedule service.56 Table 2 already showed one dimension of non-random allocation of passengers. Minority drivers were more likely to service minority passengers and white drivers were more likely to be paired with white passengers. We also explored a few other dimensions of non-randomness. We found, for example, that a Pierson chi-squared test of statistical independence rejected the hypothesis that African-American and white drivers pickup passengers from the same neighborhoods (p = .01). But we found no statistical difference in the average fare of black, white or other minority drivers. While the evidence of non-randomized allocations makes it more difficult to test for customer discrimination, our regressions controlled for a host of non-racial differences and still found robust statistical evidence that minority drivers were tipped less – even after controlling for the heightened chance that minority drivers have of serving minority customers and even after controlling for their non-random allocation of neighborhoods. Accordingly, the disparate customer hypothesis does not, in the end, present a strong challenge to our earlier results. C. Lower Tips by Minority Passengers It is superficially inviting to categorize our earlier results as that the minority status of either the driver or the passenger cause the expected tipping percentage to be lower. But these two racial effects stand on a very different theoretical and empirical footing. As a theoretical matter, it is straightforward to test whether one person’s race influenced the decisions of another person – for example, X refused to sell to Y because Y was black. But it is much harder to test whether the behavior of a decision-maker herself was affected by the decision maker’s own race. Such attempts run the risk of crudely essentializing racial characteristics. At the extreme, such interpretations suggest far fetched genetic predispositions. But, as is well understood in the identity literature,57 the other extreme of emptying race of all experiential content is a social scientist’s nightmare. Of course, cultural differences in behavior – such as a different propensity to tip – are likely at some level to be the byproduct of differential cultural experiences. But it is literally impossible to independently control for the myriad of present and past Steve Salop helpfully suggested that we reanalyze the data to try to better control for more aggregated driver strategies over the course of a shift. In this more aggregated analysis, we would have tested whether driver racial disparities persist at the shift level when we take into account that waiting at the airport takes longer but is expected to generate a larger fare. However, the problem of incomplete shift data, discussed supra Section IV.A, unfortunately precludes us from analyzing shift data in a systematic way. 56 But this last possibility may still be an example of customer discrimination, if customers systematically prefer to schedule with white drivers. 57 See, e.g., Kenji Yoshino, Covering, 111 YALE L.J. 769 (2002). 55 27 disparities in cultural experiences. So to demand an experiential based quantitative test of culture differences is in effect a request for statisticians to put away their regressions. The prior results, as a theoretical matter, exist in the intermediate range between the essentializing and non-essentializing extremes. The passenger race coefficients should be interpreted (at best) as crude measures of current cultural differences – taking the (historical and present) ambient experiential differences as given. The regressions ask whether, after controlling for aspects of the observed transactions, individuals of particular racial cultures (exposed to different experiences) tend to tip differently.58 As an empirical matter, it is important to remember that passenger race effects were not as robust as driver race effects. While the black and Hispanic passengers were found to tip systematically less than whites in all the regressions of Table 10 and to stiff systematically more in all the regressions of Table 12, the degree of the disparity narrowed as better neighborhood controls were added to the regressions. Nonetheless, analysis of data from other tipping studies makes us fairly confident that passenger race – in the limited sense just described – is likely to play an independent role in expected tip amount. It turns out that an independent literature exists examining whether customer race affects the tipping size – particularly (but not exclusively) in restaurants. The least persuasive studies merely survey service providers about their general impressions. For example, a Houston survey of 51 waiters and waitresses found that 94% of the servers classified African Americans as poor tippers – whereas none of the servers classified whites as poor tippers.59 There are also numerous racialized incidents concerning server perceptions of African Americans as being poor tippers.60 For example: On October 23, 1999, Charles Thompson and Theresa White went out for dinner at Thai Toni Restaurant in Miami Beach, Florida. When they got their bill, they found that a 15 percent gratuity had been added even though no similar charges were added to the bill of a nearby couple. Mr. Thompson asked the restaurants’ Most residents of New Haven (96.3%) are native United States citizens and of the foreign born residents most (86.5%) are from Puerto Rico. Census Table, supra note 32. But two or three of our six AfricanAmerican drivers were born abroad. This raises the possibility that some of the tipping disparity that we attribute to race discrimination may instead be attributable to national origin discrimination. Unfortunately, at the time of collecting the data we did not ask the national origin of the drivers and hence cannot separately control for this affect. 59 Michael Lynn, Servers’ Perceptions of Who are Good and Poor Tippers (unpublished manuscript 2000). 60 For example, an anonymous posting from a discussion board at www.tipping.org expressed the following perception, which it described as common among servers: When I worked at T.G.I Fridays, I noticed that many tips from African American parties were not based upon a percentage of the check, but were typically an arbitrary amount and were usually in the two-to-five-dollar range. This is not to say that all African Americans left small tips, but a significant number did. 58 28 owner/manager for an explanation and was told: “You black people don’t tip well.”61 The Miami city counsel reacted by passing an ordinance forbidding discriminatory tipping practices in eating establishments.62 More authoritative studies take one of two forms. Some of the studies, like ours, have restaurant servers record information about their customers.63 Other studies interviewed customers as they depart the restaurant.64 Indeed, Michael Lynn and Clorice ThomasHaysbert conducted something akin to a meta-analysis in which they pooled the restaurant data from five different studies of restaurant tipping in Houston.65 The authors found that African-American customers were expected to tip 3.59 percentage points less than white customers (p < .0001).66 While the restaurant studies are suggestive, they too may be infected by the problem of omitted variable bias. Next we examine two dimensions in which our prior tipping regressions are incomplete and may misattribute effects to passenger race that are not really there. Service, Not Passenger Race. If drivers systematically offer poorer service to minority customers, then the regressions may mistakenly suggest that African Americans and Hispanics tend to tip less at a given level of service. Instead, it may just be that passengers of all races tend to tip less for poor service, and drivers disproportionately give poor service to minority passengers. This alternative hypothesis is a concern because we have only very weak controls for driver quality (chiefly speed and conversation).67 Michael Lynn & Clorice Thomas-Haysbert, Ethnic Differences in Tipping: Evidence, Explanations, and Implications 3 (unpublished manuscript 2003). 62 See http://www.co.miami-dade.fl.us/csd/tipping.htm This single event brought down a firestorm of protest. The NAACP organized a picket line outside the restaurant; the Greater Miami Convention and Visitors Bureau removed the restaurant from its membership list; the Florida Attorney General charged the restaurant with violating the state’s Deceptive and Unfair Trade Practices Act and forced the restaurant to pay a $15,000 fine. See Lynn & Thomas-Haysbert, supra note 61; Ruben Castaneda, Restaurant Sued Over Tip Dispute; 15% Was Added To Lunch Bill for Black Women, WASH. POST (Apr. 13, 2000). 63 Michael Lynn, Tipping: A Reward for Server Effort? (unpublished manuscript 2000); Michael Lynn et al., Reach out and touch your customers, 39 CORNELL HOTEL & RESTAURANT ADMIN. QUARTERLY 60 (1998). 64 Michael Conlin et al., The Social Norm of Restaurant Tipping (unpublished manuscript 2000); Michael Lynn & Jeffrey Graves, Tipping: An Incentive/reward for Service? 20 HOSPITALITY RES. J. 1 (1996); Connie Mok & Sebastian Hansen, A Study of Factors Affecting Tip Size, 3 J. RESTAURANT & FOODSERVICE MARKETING 49 (1999). 65 Lynn & Thomas-Haysbert, supra note 61, at 13 (the five studies that are pooled are listed in notes 63 and notes 64). 66 Id. at 16. The pilot study of Suzanne Perry, discussed supra note 46, also found statistically significant shortfalls in the percentage tipped by African Americans relative to whites. 67 Our limited attempt at using testers to audit the drivers themselves (which yielded only ten observations) provides a smidgen of information related to the service issue. Our testers included a pair of men, one white and one African American, and a pair of women, one white and one Indian American (Asian subcontinent). The testing regime was structured so that both members of each pair took rides with the same 61 29 There are, however, several different forms the poor service hypothesis might take, depending on the scope of poor treatment, which might actuate lower tips. For example, minority passengers might tip less because (i) the current driver provided poorer service; (ii) a prior driver provided poor service; or (iii) a prior retailer provided poorer service to the minority passengers (or their families or friends). Indeed, it might be theoretically possible that minority passengers and drivers may be caught in mutually reinforcing, unhappy equilibrium in which minority passengers give little because they are generally exposed to poor retail treatment, while drivers generally provide poor treatment because they expect a poor tip. Class, Not Passenger Race. Second, as discussed above, the tendency of non-minority passengers to tip more may be driven by socio-economic class differences. Possibly, rich people tip more, not white people. But our regression does not pick up this impact because we have again only very poor correlates with class (passenger dress, pickup and drop-off locations).68 Fortunately, a new study by Michael Lynn has good controls for both income and service in a national telephone survey he conducted eliciting information from approximately 900 consumers on their tipping behavior with regard to 9 different service providers – including taxi cab drivers.69 The respondents were asked not only their race, sex, and age, but also their income (in ten ordered categories) and education (in seven ordered categories). The survey additionally controlled for service quality in the way the tipping question was framed. For example with regard to cab tipping, respondents were asked: “If you received good service from a cab or limousine driver would you tip a percent of the total cost of the service, tip them a flat amount or not give them a tip?”70 drivers. We found that our minority testers actually rated the quality of service slightly higher than white testers (3.98 on a 5-point scale for 5 rides by minority passengers vs. 3.56 for 5 rides by non-minority passengers). In addition, the testers’ relative rankings of driver quality were identical. Hence, the hypothesis that minority passengers might tip less because they received poorer service does not find support in this handful of observations. 68 There is another piece of evidence that could be construed to suggest that poverty might be driving poor tipping. If we break down the stiffing rates by fare type, we find that relatively few whites or blacks are willing to stiff when the fare ends up being 75 cents over the dollar. But when the fares end up being 25 cents over the dollar African Americans show a much greater propensity (31 times greater) to stiff drivers. This is consistent with poverty being disproportionately important to African-American passengers: when more money (75 cents) is at stake African Americans are more willing to stiff. Table 13. Mean Stiffing by Passenger Race and Fare Type Race/White Race/White Disparity 25¢ over Disparity RACE Integer Fare Ratios the dollar Ratios White 22.1% 1.0% Black 61.5% 2.78 32.7% 31.42 Hispanic 50.7% 2.29 13.6% 13.09 Asian 26.2% 1.18 10.7% 10.29 Other 50.0% 2.26 0.0% 0.00 69 70 50¢ over the dollar 3.6% 23.3% 20.8% 0.0% 33.3% Race/White Race/White Disparity 75¢ over the Disparity Ratios dollar Ratios 5.1% 6.42 6.9% 1.36 5.73 6.3% 1.23 0.00 11.1% 2.19 9.17 0.0% 0.00 Michael Lynn, Black-White Differences in Tipping of Various Service Providers (Working Paper 2003). Id. at 5. 30 Respondents who said they would leave a percentage or flat tip were then asked: “What amount?” Lynn found that African-American respondents were 11% more likely than white respondents to say that they would stiff a cab or limo driver (p < .03). There was no significant racial difference among those who said they would leave a tip in the propensity to leave a flat amount and no significant racial difference in the amount of flat tips. But among percentage tippers, Lynn found that African-American respondents were likely to report tipping 1.99 percentage points less than white percentage tippers – however, this result falls slightly short of statistical significance (p = .14). Lynn did not aggregate the three types of responses (stiff, percentage, and flat) to get an overall assessment of racial disparities. But controlling for service and income does seem generally to narrow the passenger race disparities in propensity to stiff and in average tipping percentage as compared with our previous findings. History. A final explanation of lower minority tipping resuscitates the brief history discussed in Part II. At one time, accepting tips was seen as a symbol of the recipient’s degradation. Minorities may at one time have been unwilling to tip – not because they were disproportionately poor or because they were receiving poorer treatment – but because they did not want to participate in a practice that had been so often framed as a token of their own inferiority.71 Of course, this degradation conception of tipping may have long passed. But both minority and non-minority consumers today may still be affected by this now withered perception – as one generation passes its tipping practices on to the next.72 Stepping back, we are most concerned about the possibility that drivers non-randomly censored the data in ways that might undermine the reliability of our primary results. This censoring effect by itself should qualify anyone’s reading of the data and underscores the preliminary nature of the study. In contrast, we are less concerned with the various omitted variable concerns discussed with regard to the driver-race and passenger-race effects. The evidence of customer discrimination against minority drivers is relatively stable and robustly significant. The evidence that African-American and Hispanic passengers tip less is slightly less stable – and declines in size as better controls for class (captured by pickup and drop-off locations) are added. But even after taking out the class component (to the extent our data permit), there seems to be an independent and robustly significant passenger-race disparity in tipping (particularly with respect to stiffing propensities). V. Why Are Consumers Discriminating? As reported above in Table 6, 39.2% of African-American passengers left no tip, as compared with just 10.9% of white passengers. 72 Recall that children generally know when and how much their parents tip. See supra notes 11-12 and accompanying text. 71 31 Statistically identifying the cause of disparate treatment is usually a daunting task.73 And our survey of cab drivers seems particularly ill-suited to uncover the well-springs of customer motivation. Nonetheless, this section will try to tease out of the data a few statistical intimations that at least are suggestive of the extent to which the disparate treatment is conscious. Our first inclination was to be skeptical that conscious discrimination could be playing much of a role in the overall shortfall for minority drivers. Is it really possible that many passengers would overtly tip a lower percentage because the driver was a minority? But the racial disparity in the rates of stiffing gives one pause. Recall from Table 4, that black drivers were 80% more likely to be stiffed than white drivers (28.3 vs. 15.7%). And Table 12 showed that this disparity was sustained as we controlled for more variables. The disparity in the stiffing rate suggests that stiffing is not just caused by a hard-wired set of passengers who never tip. Rather passengers tend to make a conscious decision to sometimes leave no tip. This does not mean, however, that passengers were consciously stiffing based upon the cab driver’s race. But it does suggest that conscious decisionmaking of some kind was at work. In contrast, there is another part of the data that resonates with unconscious thought processes. People who are deciding how much to tip often choose to either round up or down to the dollar nearest their preferred total (including tip).74 If we use the average passenger race tipping percentages (from Table 5) to make predictions about the passengers’ preferred tipping percentages, it is possible to identify those observations in which the tip given was just sufficient to round the total payment up or down to the nearest dollar. Table 14 divides the tipping data into instances where passengers rounded the total amount to the integer nearest this target amount. Table 14. Distribution of Rounded and Non-Rounded Total Amounts Paid Fares Percentage Rounded down to nearest integer 286 27.0% Rounded up to nearest integer 294 27.8% Rounded to nearest integer 580 54.8% Rounded to another amount 176 16.6% Not rounded 303 28.6% Total 1059 100.0% We find that just over half of the tips in the dataset have been rounded by the passenger to the dollar nearest our prediction of the passenger’s preferred tipping percentage. And, as one might expect, there is substantial uncertainty as to whether the rounding will be up or down. See AYRES, supra note 1, at 54 (for an attempt concerning disparate treatment in new car sales). Tipping your preferred percentage is harder to accomplish in cash transactions than in credit card transactions (where the customer may have an easier time filling in a non-rounded total amount). Accordingly, we would expect the rounding phenomena to be more important with regard to taxi cab tipping than with regard to restaurant credit card tips. 74 73 32 These rounded observations, to our minds, are prime candidates for unconscious discrimination. Imagine the following scenario: Just before your cab arrives the fare clicks over to $7. Think fast. What to do you do? Do you leave $8 or $9? We imagine that passengers are called upon to make a quick decision about whether to round up or down. Even people who think that they are hard-wired percentage tippers may find that unconscious factors influence this rounding decision. Just as people who confront the IAT sorting game often can’t help but treat blacks and whites differently,75 people who confront the dichotomous rounding game may have trouble purging the influence of race. In fact, when we analyze the rounded observations, we find a bit of evidence to support this idea form of unconscious discrimination. Overall, passengers were 6 percent more likely to round up than round down when they have a white driver than when they have a black driver.76 Finally, we tried to decompose the total Black-White driver tipping disparity of 7.73 percentage points (from Table 3) into two components parts: a stiffing disparity and a rounding disparity.77 We find that 27.2% of the overall disparity comes from the propensity to stiff minority drivers, and that 35.5% of the overall racial disparity comes from different propensities to round when paired with a minority driver. This decomposition suggests that both unconscious and conscious motivations may be playing a role in consumer discrimination. But we should again emphasize the weakness of these conclusions. We are pushing the data to the limits of their competence and while the evidence of disparities in stiffing are quite robust and by themselves explain a substantial portion of the customer discrimination, the evidence of disparities in rounding is much more tentative and turns, among other things, on our particular method of predicting what were individual customers’ preferred tipping percentages. As one of us has noted: The IAT asks the subject to complete four different sorting tasks. In the first task, the subject is asked to simply sort photographs into two categories labeled ‘African American’ and ‘European American’… In the second task, the subject is asked to sort words (such as ‘love,’ ‘war’) into two categories labeled ‘Good’ and ‘Bad.’ These first two tasks allow the test to establish a baseline metric of the subject’s ability (measured by speed and accuracy) to sort photographs and words. The third task then asks the subject to sort combinations of photographs and words into two categories. One of the categories is labeled ‘African American or Good’ and the other category is labeled ‘European American or Bad.’ And the final task asks the subject to sort the categories into the two categories ‘African American or Bad’ and ‘European American or Good.’ AYRES, supra note 1, at 419-20. Subjects perform the final sorting task significantly more quickly than the third, suggesting an implicit attitudinal preference for European American over African American. 76 But somewhat surprisingly this overall effect is driven by the behavior of minority passengers. Minority passengers were more likely to round up when paired with a white driver (than when paired with a minority driver). But white passengers were less likely to round up with white drivers (than African-American drivers). However, in an alternative form of this analysis using regression-based predictions, other passenger races, including black passengers, were also less likely to round up with white drivers. 77 The rounding disparity is calculated as the weighted difference of the black and white rounded tipping percentage averages. The stiffing disparity is calculated as the (generally) increased probability of a black driver stiff multiplied by the white passenger tipping percentage average over non-rounded fares. (Note that stiffs are defined not to be rounded fares.) 75 33 VI. Normative Implications To our minds, there are two primary normative implications to be drawn from the foregoing analysis. First, the government should be careful in its tax laws not to add insult to injury by directly or indirectly attributing phantom tipping income to minority drivers.78 The second and more important implication is a tentative proposal to raise taxi fares by 15% and to require “tip included” decals to be prominently displayed in all cabs. At a minimum, this paper has identified two new rationales for such “mandated tipping” regulation. But before proceeding, we should mention at least in passing that the finding of lower tips for minority drivers raises interesting issues about the scope of our civil rights laws. Does the evidence of customer discrimination against minority drivers suggest a violation of the Civil Rights Act of 1866, 42 U.S.C. § 1981? Section 1981 is a broad prohibition against race discrimination in contracting.79 But we have not been able to find a case challenging discrimination by buyers. Indeed, Jones v. Alfred H. Mayer famously found that the goal of Section 1981 was to insure that "a dollar in the hands of a Negro will purchase the same thing as a dollar in the hands of a white man."80 The focus is on protecting minority buyers, not minority sellers. And even if the statute’s broad language was read to regulate buyer behavior, it is far from clear that a passenger’s tipping decision concerns discrimination in the terms of a contract. Even though the law has increasingly treated tips as wages for a variety of tax purposes,81 disparate treatment in tipping does not concern a formal term of agreement and thus might not impair the right of minority drivers “to make and enforce contracts.”82 But whether or not these legal obstacles could be cleared, there are abundant logistical problems that would preclude the use of Section 1981 or any other civil rights statute to discourage passenger discrimination. The difficulty and costs of proving that an individual passenger tipped less because of the driver’s race would block even the most subsidized litigation. Our findings of passenger discrimination illuminate interesting issues of civil rights law, but they do not suggest (to our minds) a viable new form of litigation. A. Adding Insult to Injury? There are several ways in which the disadvantages to minority drivers (and, more generally, minority employees) resulting from consumer discrimination could be 78 79 Bruce Ackerman pointed this implication out to us. See 42 U.S.C. § 1981 (1994) (“All persons . . . shall have the same right . . . to make and enforce contracts . . . as is enjoyed by white citizens . . . .”). 80 392 U.S. 409, 443 (1968). 81 Seagrave, supra note 11, at 123. 82 See Patterson v. McLean Credit Union, 491 U.S. 164 (1989). 34 exacerbated by methods of estimating tipping income for tax purposes.83 For example, a business establishment enrolled in the Tip Rate Determination Agreement (TRDA) is required to work with the IRS to determine a baseline tip rate for its employees.84 If this imputed tipping income is based on the overall average – which is likely to be greater than the minority average -- minority employees will be taxed on money they did not earn or, put another way, minority drivers will be forced to pay taxes at an effectively higher rate than their white counterparts. Furthermore, “TRDA provides that if employees fail to report at or above the determined rate, the employer will provide the names of those employees, their social security numbers, job classification, sales, hours worked, and amount of tips reported.”85 Hence, a minority employee conscientiously reporting tip income could be threatened with a higher risk of audit and job loss. An alternative to the TRDA is the Tip Reporting Alternative Commitment or TRAC. Although TRAC does not require the determination of a baseline tipping rate -- instead requiring methods for directly reporting tip income -- it does provide that “if the employees of an establishment collectively underreport their tip income, tip examinations may occur but only for those employees that underreport.”86 Hence, in cases where the IRS regards employees as collectively underreporting, minority drivers are likely to bear an unfair portion of the blame and therefore again are more likely to be audited or lose their jobs. Presently, TRDA and TRAC are available to food and beverage and hairstyling industries as well as casinos, so that in these businesses in particular, minority employees might be shouldering the multiplied burdens of lower tipping income, greater tax rates and higher possibility of audits and job loss. According to the IRS, “Plans are underway to extend this program to all industries where tipping is customary.”87 An immediate implication of the results presented in this paper is that, unless adjustments are made to take into account lower minority tipping incomes, programs such as TRDA and TRAC may compound the problem of consumer discrimination. Furthermore, studies of consumer discrimination are necessary for those industries in which TRDA and TRAC are already in use, to determine whether minority workers do in fact earn systematically lower tipping incomes and are therefore disadvantaged by the programs.88 Apart from programs estimating tip income, minority drivers or employees may be unfairly exposed to audits and penalties simply on the basis of the fact that their reported tipping incomes may appear suspiciously low. According to the IRS, failure to report tips It might be questioned why literal gratuities count as income at all since they more closely resemble gifts than contracted wages. Nonetheless, courts universally treat tips as part of wages for a variety of different legal purposes. For a history of the treatment of tips as wages, see SEGRAVE, supra note 18, at 12. 84 Tips on Tips: A Guide to Tip Income Reporting for Employers in Businesses Where Tip Income is Customary, IRS Publication 3144 Department of the Treasury, Internal Revenue Service at 6. 85 Id. at 7. 86 Id. at 7. 87 Id. at 3 88 A further possible problem that might result from the imputation of tipping income is in the withholding of wages for tax purposes. If a uniform percentage of income is withheld for all employees on the basis of an assumed baseline tipping rate, then that sum will be disproportionately large for minority employees. 83 35 can result in “a penalty equal to 50% of the social security and Medicare taxes or Railroad Retirement tax [owed] on the unreported tips” (as well as a “negligence penalty of 20% of the additional income tax, plus interest”).89 Hence minority drivers may be unfairly subject to additional penalties as a result of consumer discrimination.90 B. Service Compris While our “phantom income” concern is analytically sound, we are agnostic about its empirical importance. Driver non-compliance with tax laws may be so rife that it is harder to tell whether minority drivers are given less room to cheat on their reporting of income or would be subject to a higher chance of audit on conscientiously reported income.91 Our second proposal is more important, but also more controversial. The central idea is that prohibiting all tipping would likely reduce two types of disparate racial treatment. It would directly stop passengers from discriminating against minority drivers. And secondly, a prohibition on tipping might reduce some driver discrimination against minority passengers. This second point will be more fully elaborated in just a moment. But first a word about implementation. 1. Implementation It would be possible for the law to directly prohibit tipping. Indeed, anti-tipping statutes of just this kind were passed at the beginning of the twentieth century by a handful of states.92 But enforcement of a strict prohibition is impossible. Who is to know if a passenger slips a driver a few extra dollars before exiting the cab? And why would police have an incentive to investigate or prosecute such small potatoes infractions? Moreover, we imagine that the prohibition would incense many passengers (“What right does the government have to say that I can’t tip?”) and insight noncompliance. Instead, we tentatively propose that cab commissions simultaneously increase the metered price by 15% and require cabs to prominently display “Tip Included” decals. Enforcement of the decal requirement is much more feasible, because decals are rather durable,93 and enforcers don’t need to observe the private interactions between drivers See Reporting Tip Income, IRS Publication 531, Department of the Treasury, Internal Revenue Service, at 3; Tips on Tips: A Guide to Tip Income Reporting for Employees who Receive Tip Income, IRS Publication 3148, at 6. 90 The overall risk is probably not very high. According to Moody’s: The chance of an individual tax return's being audited last year was less than one in 200, down from one in 112 in 1999 and one in 60 in 1996. For taxpayers who make more than $100,000, and who pay 62 percent of all individual income taxes, the audit rate last year was slightly less than one in 100, down from one in 50 in 1998 and down from one in 9 in 1988. http://www.efmoody.com/newsletter/may2001.html (We conducted a number of audit tests in this regard using Turbo Tax filing software without triggering their audit checks.) 91 Joseph Bankman & Stuart Karlinsky, Cash Business Owners and Their Accountants (Working Paper 2002). 92 See supra note 18. 93 See Mark F. Grady, Why Are People Negligent?: Technology, Nondurable Precautions, and the Medical Malpractice Explosion, 82 NW. U.L. REV. 293 (1988). 89 36 and passengers -- they just need to check off whether a decal is in place when they are already inspecting the cab. “Service compris” regulation of this kind is also likely to motivate a much higher degree of passenger compliance. Passengers are likely to know that the fare has been increased by 15% and will not feel an obligation to double pay. To be sure, in many countries, there has arisen a norm of paying a small tip in addition to the “service compris” amount added to one’s check.94 A similar add-on norm might also develop under our proposed system. But the key point here is that amounts added on are likely to be a lot less than the 16% mean tip in our study. If the most generous passengers add on at most 5% under a “service compris” regime, there simply would not be enough room for either of the race effects uncovered above. Both (i) the amount of customer discrimination against minority drivers, and (ii) the shortfall in tipping by African-American and Hispanic passengers are likely to decline. 2. Reducing Driver Discrimination The impact of “service compris” regulation on customer discrimination against minority drivers is straightforward.95 Anything that reduces customer discretion with regard to tipping should predictably reduce the opportunity for customers to discriminate. But reduced tipping discretion may also reduce a second type of race discrimination – discrimination by drivers against minority customers. If drivers expect to be tipped less by minority passengers, they may refuse to pick up these customers in favor of bettertipping white customers. The leading empiricist on tipping in the United States, Michael Lynn, has previously diagnosed the possibility that low minority tips would lead to service discrimination: See www.info-france-usa.org/visitingfrance/tipping.asp (“[In service compris establishments,] it is customary to leave small change unless you are dissatisfied.”). 95 Customer race discrimination may be an important impediment to the growth of minority owned business. Indeed, two of us have argued that evidence of downstream discrimination by customers provides a constitutional basis for government affirmative action. Ian Ayres & Fredrick E. Vars, When Does Private Discrimination Justify Public Affirmative Action?, 98 COLUM. L. REV. 1577, 1614 (1998). Courts and commentators appreciate how “upstream” discrimination (that is, discrimination by a minority firm’s suppliers) can impede a firm’s ability to compete, but often fail to consider how “downstream” consideration by one’s customers can create racialized barriers to entry. Jennifer Lee makes this point with respect to downstream discrimination by black customers against black-owned firms: [B]lack merchants often charge higher retail prices because their suppliers charge them higher wholesale prices than they do other business owners . . . . Black merchants, on the other hand, complain that black customers refuse to patronize their own, opting instead to buy from neighboring Jewish and Korean merchants.” Lee, supra note 4, at 93 (emphasis added). Customer discrimination of the kind uncovered in this Article may therefore also be relevant to government attempts – via affirmative action in procurement – to counteract the impact of private discrimination. Ayres & Vars, supra, at 1612. 94 37 [I]f Blacks do tip less than Whites, then managers should try to change their Black customer’s tipping behavior and/or should closely monitor their tipped employees’ treatment of Black customers. Otherwise, their tipped employees are likely to deliver inferior service to Black customers whom they believe are poor tippers.96 But our proposal inverts his proposed intervention. Instead of encouraging minorities to tip more, we propose a (much more feasible) regime to induce non-minorities to tip less.97 We propose to eliminate (or at least to substantially decrease) discretionary tipping across the board. While it might at the moment seem that the practice of tipping in certain service industries is too embedded in American norms to be displaced,98 it is useful to remember our earlier discussion of the strong antipathy that many Americans had for tipping less than 100 years ago. The Anti-Tipping Society of America, an alliance of 100,000 traveling salesmen who from 1905 to 1919 managed to have the custom outlawed in seven states.99 As late as 1946, Life magazine concluded that “tipping was a national nuisance and as such should be eliminated.”100 At that time, fully 69.7 percent of general public would have preferred that tipping be eliminated—provided that services workers were provided with fair wages.101 All this is to say that a social practice that today seems an inevitable or inherent part of our economy may not seem so inevitable if we look slightly into the past (or at other parts of the world102). To make this theoretical possibility more empirically plausible, the next two sections of the paper lay out the evidence that (i) driver discrimination is a serious problem and (ii) drivers engaged in various forms of statistical discrimination are likely to infer that minority passengers will tip them substantially less. Lynn, supra note 69, at 4. Indeed, one of the surprising possible implications of our research is that the enhanced restaurant tipping norm for progressive patrons which has gradually developed over the last 50 years of ratcheting up the expected tipping percentage for servers – first from 10% to 15% and then from 15% to 20% – may have actually had the non-progressive effect of enhancing server discrimination against minority customers. 98 That Americans tip only certain, apparently arbitrary, categories of service workers itself suggests that we are not hard-wired tippers. For example, we tip the shoe-shine person but not the salesperson who helped us pick out the shoes. 99 Lynn Snyder, No Tips Please, available at www.speakeasy.org/wfp/36/tips.html (“By 1915, state legislatures in Wisconsin, Illinois, Iowa, Arkansas, Mississippi, Nebraska, Tennessee and South Carolina were trying to enact anti-tipping laws., Some were passed, some were vetoed by governors, but in the end, they were struck down by the courts, which said, in effect, that Americans can spend their money as they see fit.”). See Dunahoo v. Huber, 185 Iowa 753 (1919) (statute violates the state privileges & immunities clause because there was no reasonable ground for allowing employers to accept tips and prohibiting employees from accepting tips when “engaged in like services.”); Ex Parte Farb, 178 Cal. 592 (1918) (statute violates the due process provision of the U.S. Constitution and the freedom of contract provision of the California constitution). The anti-tipping statutes were repealed in Arkansas, Mississippi, South Carolina, and Tennessee in 1925, 1926, 1922, and 1925, respectively. Note, 17 CORNELL L.Q. 183, 188 (1932). 100 Leo P. Crespi, The Implications of Tipping in America, 11 PUB. OPIN. QUART. 424, 424 (1947). 101 Id. at 426. 102 See infra note 113 and accompanying text. 97 96 38 a) Evidence of Driver Discrimination There is abundant empirical evidence that cab drivers have discriminated against AfricanAmerican passengers by refusing to pick them up.103 Dozens of prominent African Americans have presented detailed descriptions of discrimination. For example, Cornell West describes an incident of driver discrimination in 1993 that provoked him to write Race Matters: This past September my wife, Elleni, and I made our biweekly trek to New York city from Princeton. I dropped my wife off for an appointment on 60th Street between Lexington and Park Avenues. I left my car—a rather elegant one--in a safe parking lot and stood on the corner of 60th Street and Park Avenue to catch a taxi. . . . I waited and waited and waited. After the ninth taxi refused me, my blood began to boil. The tenth taxi refused me and stopped for a kind, welldressed, smiling female citizen of European descent. As she stepped in the cab, she said, “This is really ridiculous, is it not?”104 More recently, actor Danny Glover filed a complaint with the New York Taxi and Limousine Commission (TLC) after five cab drivers refused to pick up his daughter and him.105 Following newspaper reports of the Glover incident, several other prominent African-American men – including Harry Bellefonte, David Dinkens and Denzel Washington – came forward reporting that cab drivers had refused to pick them up as well.106 The TLC at the time was averaging seven complaints of driver discrimination each day.107 In response to the Glover incident, New York Mayor Rudi Giuliani started “Operation Refusal” – a sting operation in which 150 undercover police were empowered to See Cole, supra note 28. This discrimination seems particularly pronounced in cities like New York and Washington where “flagging down” a cab is more prevalent. In the past – and possibly as a response to driver discrimination – informal “jitney” cabs would serve African-American neighborhoods. See http://usembassy.state.gov/kampala/wwwhinsidenews_jitney. html (“Throughout the past century, jitneys were often the only cab service available to African American neighborhoods, where special hire taxi drivers refused to go.”); see also AUGUST WILSON, JITNEY (1982); RUBEN SANTIAGO-HUDSON, LACKAWANA BLUES (2000). 104 CORNELL WEST, RACE MATTERS (1993). 105 J. M. Shenoy, African Americans Flag Down NY Cabbies (Nov. 6, 1999) www.rediff.com/news/ 1999/nov/06us1.htm. 106 Id.: Several African American celebrities and senior executives working for top law and financial firms in New York were all over the media in the past 24 hours detailing their own experiences. Makeup artist Dynode Marcie said he was in a tuxedo and waiting for a cab near the Radio City Hall in Manhattan after the Grammy Awards were given out. Several cabbies passed by him. … A top Wall Street executive said that on a cold night, four cabs passed by her and her three children. When a white man appeared a few minutes later and stood a few feet ahead of her, two cabbies almost ran into each other trying to stop for him. See also www.dackman.homestead.com/files/TaxiClips.htm 107 My ‘Zero-Tolerance’ Plan To End Taxi Discrimination, available at www.cvfieldsmbp.org/newsle. 103 39 immediately impound the cabs of drivers who refused to pick up minority passengers.108 First-time offenders are fined up to $350; third-offenders can have their license permanently revoked.109 Evidence of this “drive-by racism” can also be gleaned from broader surveys of the population. For example, an ABC news poll found that a plurality of blacks (42%) believe taxi drivers avoid picking up blacks (only 19 percent of whites agree).110 And one in six blacks (18%) report that they have personally been refused a cab ride. Finally, a 1989 audit study sponsored by the Washington Lawyers’ Committee for Civil Rights provides controlled evidence of disparate racial treatment by cab drivers. One black and one white tester stood three car lengths from each other and attempted to hail a cab in Washington, D.C. The study found that taxis were 11.2 percent more likely to stop for whites than blacks and that as a result blacks had to wait, on average, 27 percent longer for a cab to stop.111 b) Tests of Statistical Discrimination All in all, there seems to be overwhelming evidence of pickup discrimination by drivers against minority passengers. Indeed, while there has been a sustained debate about cab driver discrimination, it does not concern whether such discrimination exists – but whether it is justified. To date, the defenders of the discrimination (including many cab drivers themselves) have argued that driver discrimination is a reasonable reaction to a heightened risk that minority passengers will hurt them during the commission of a robbery. For example, a Boston Globe article entitled “Racism or Fear” emphasized: “Taxicab drivers and chauffeurs face unusually high risks of becoming homicide victims,” noted the Bureau of Labor Statistics in a 1994 report. “This occupation accounted for almost one tenth of all victims of job-related homicide, but less than one-half of one percent of the nation's workforce. Nocturnal trips, especially those to secluded areas, make these drivers particularly vulnerable. Almost half the cab drivers died from 9 p.m. to 3 a.m.” . . . For most, racism isn’t a reason to avoid picking up black men or driving to black neighborhoods. Prudence is.112 See Padberg v. McGrath-McKechnie, 108 F. Supp. 2d 177 (E.D.N.Y. 2000); Randy Kennedy, Cabbies Entitled to Hearings, Judge Rules, N.Y. TIMES (May 1, 2002) (“A federal judge in Brooklyn has ruled that the city's Taxi and Limousine Commission must give cabdrivers a chance to defend themselves before it can revoke their licenses for illegally refusing to pick up people who hail them.”). 109 Jeff Jacoby, Racism—or Fear?, BOSTON GLOBE (Nov. 18, 1999) 216.247.220.66/jacoby/1999/jj11-1999.htm. 110 abcnews.go.com/onair/2020/2020_000217_abcpoll_races.html (Jan. 12, 2001). 111 Siegelman, supra note 1, at 77. 112 Jacoby, supra note 109; see also DINESH D'SOUZA, THE END OF RACISM 282 (1995); Glenn C. Loury, Discrimination in the Post-Civil Rights Era: Beyond Market Interactions, 12 J. ECON. PERSPECTIVES 117 (1998) (“If the set of those seeking taxis when the odds are low that a given taxi will stop for them has an especially large fraction of potential robbers in it, the drivers might rationally be reluctant to stop.”). 108 40 Drivers emphasize that they are often not scared of the passengers themselves, but of where the passengers will ask them to go. One driver explained: The man who gets into my cab could be the best man in the world but who knows what could happen to me after I have dropped him home?113 By this argument, it might be rational for cab drivers to refuse to pick up even welldressed elderly African-American women (who present very low probability of committing crime themselves) if they are more likely than others to live in high-crime neighborhoods. Of course, opponents of driver discrimination have appropriately questioned whether the driver discrimination might instead be driven by irrational statistical inferences, stereotypes or more traditional forms of racial animus.114 But this paper suggests the possibility of another dimension of statistical discrimination. Instead of (or, in addition to) making inferences about crime, cab drivers might refuse to pickup minority passengers because they expect a lower tip, a lower fare, or a higher chance of having to “deadhead” back from the drop-off location without a return fare. The idea is not new that servers might give poorer service or refuse to serve a demographic group that they predict will give a poor tip. For example, the California Railroad Commission concluded in 1914 that women traveling alone “because they are known to tip less ‘generously’ than men, receive the aid of the porter last or not at all.”115 While such statistical discrimination might be theoretically coherent, actual inferences made on this basis may be irrational or stereotyped. Indeed, psychologists have found that people who believe they are merely rational statistical race discriminators are more likely than others to harbor unconscious biases against minorities.116 To see whether tipping or one of the other revenue-based inferences might plausibly induce driver Shenoy, supra note 105. See Jody David Armour, Race Ipsa Loquitor: Of Reasonable Racists, Intelligent Bayesians, and Involuntary Negrophobes, 46 STAN. L. REV. 781 (1994). Professor Armour has also argued that racial inferences are irrational in a more dynamic sense. A prospective employer could more accurately assess the merit of a candidate by fairly straightforward investigation and examination. JODY DAVID ARMOUR, NEGROPHOBIA AND REASONABLE RACISM: THE HIDDEN COSTS OF BEING BLACK IN AMERICA 58 (1997). Race is often not the best evidence of merit if the employer were doing her due diligence. But this dynamic critique of statistical discrimination has less force in thinking about whether cab drivers’ refusal to serve is rational. A hypothetical driver approaching the curb to pick up a potential passenger has very little opportunity for acquiring additional information about the passengers’ qualities. About the only information that the driver can access are the things that are visible from inside the cab – and it is just these variables that we collected in our surveys. 115 Tips Really Don’t Go To Tiptakers, NEW YORK TIMES 10 (May 5,1914). Similarly, a New Jersey restaurant owner pushed legislation outlawing tipping because he felt that “poor people were sometimes insulted because of their small tips.” Law To Cut Down Tips, NEW YORK TIMES 1 (Jan. 21, 1911). 116 W.A. Cunningham et al., Implicit and Explicit Ethnocentrism: Revisiting the Ideologies of Prejudice (unpublished manuscript 2002). 114 113 41 discrimination, it is helpful to assess the magnitude of the inference that more or less rational drivers would make. Fortunately, the data are fairly well suited for this task. While our data omit many nonracial factors that might better explain why some people leave low tips, they do contain a rich set of variables describing the information that was available to the cab driver. Accordingly, it was possible for us to run “observational” regressions controlling for the factors that a driver could observe about a passenger when pulling up to the curb.117 These regressions allow us to directly assess what kind of inferences drivers would make about different passenger races after controlling for other curbside observable information. This turns a computational vice into a virtue. For these purposes, we don’t need to be embarrassed by the omitted variable problems that dogged our earlier attempts to test whether a passenger’s race actually causes the passenger to tip more or less. Instead, we are interested on what kinds of inferences a driver would make given his or her limited information. Because the salience of race may swamp the informational content of other variables, we ran the observational or statistical discrimination regressions two different ways. First, we estimated the racial inferences that an “irrational” statistical discriminator would make if he saw only passenger race (and ignored all non-racial factors). Second, we estimated the racial inferences that a “rational” discriminator would make after taking into account our full panoply of curbside observable information. To our knowledge, this is the first time that anyone has quantitatively measured the inferences that a statistical discriminator might make. The results of these regressions are reported in Table 15. 117 In these regressions, we ignored information (such as drop-off location, conversation, and distance) that only became knowable during or after the fare. 42 Table 15. Estimating the Differences (relative to White Passengers) in Fare, Tip Amounts, Probabilities of Far Suburb Destinations and Stiffing that Drivers Could Expect Observing Minority Status of Customer Irrational Statistical Discriminator: Uncontrolled observational regressions Far Suburb Stiff Outcome Variables Fare ($) Tip ($) Tip % Indicator* Indicator* Passenger Black 0.405 -1.791 -1.220 -0.053 0.317 Passenger Hispanic -1.873 0.350 -0.019 0.293 -1.009 Passenger Asian 0.243 -0.023 0.076 -2.466 -0.792 Passenger Other -0.995 -0.977 0.495 n/a 0.320 Observations 1064 1059 n/a 835 1059 R Squared 0.0634 0.0624 n/a 0.065 0.0976 Rational Statistical Discriminator: Controlled observational regressions** Far Suburb Stiff Outcome Variables Fare ($) Tip ($) Tip % Indicator* Indicator* Passenger Black -1.093 0.466 -0.953 -0.049 0.255 Passenger Hispanic -0.486 0.594 0.002 -0.711 0.228 Passenger Asian -0.512 -0.436 0.460 -0.017 0.087 Passenger Other -1.300 -0.864 0.399 n/a 0.205 Observations 968 963 n/a 436 958 R Squared 0.1665 0.2281 n/a 0.2116 0.2328 *Coefficients reported here are the changes in the probability resulting from discrete changes in the indicator variables from 0 to 1. **Other variables in the controlled regression are passenger sex, age, and dress indicators; driver age, experience and survey experience; repeat passenger, acquaintance night, late, snow/rain, and luggage indicators; and continuous pick-up location variables, categorical pick-up location variables and pick-up location specific indicator variables. Underlined coefficients are significant at the 10% level, coefficients in bold are significant at the 5% level, and coefficients underlined and in bold are significant at the 1% level. The first two columns of Table 15 report the inferences that a driver would make about the relative size of dollar fares and tips that different passenger races would produce. The top panel estimates that an “irrational” statistical discriminator (who sees only passenger race) would, for example, expect African-American passengers to have fares that were $1.79 lower than white passengers and to leave tips that were $1.22 less.118 And both these effects are statistically significant (p < .05). The raw fare differential turns out be bigger than the tipping differential (which is only 40.5% of the overall revenue shortfall) and hence would loom large in the inferences of this type of a discriminator. Analogous results are found for Hispanics and Asian passengers. Indeed, the irrational statistical discriminator would expect the fares of Asian passengers to be almost $2.50 less than white passengers and would expect the tip to be only 79 cents less.119 As a theoretical matter, a lower fare does not necessarily mean a less profitable fare. Because cab fares are non-linear, starting with a fixed $2 amount (commonly referred to by cab drivers as “the drop”), it might be more profitable for drivers to service a larger number of small fares than a smaller number of large fares. But in New Haven, the likelihood of finding numerous small fares is low so that profitability is largely monotonic with total revenue. 119 This may be due in part to geography and housing patterns in New Haven. Many Asian passengers were likely Yale University students making relatively short trips between the train station and campus. This hypothesis is also consistent with the diminution and loss of statistical significance of the passenger Asian 118 43 The second panel, however, tells a very different story. A rational discriminator (who takes into account not only passenger race but non-racial factors as well) would come to very different conclusions. Most importantly, the rational discriminator would not infer that minority passengers would have smaller fares. While the fare column shows some (but more modest) shortfalls in the size of the expected fares, none of these shortfalls are statistically significant. But the tipping shortfalls remain highly significant (in the statistical sense)120 and represent a higher proportion of the overall revenue shortfall. In short, while both rational and irrational discriminators would infer that minority passengers are likely to tip less, irrational discriminators would be much more concerned by revenue shortfalls caused by lower fare amounts (that statistically disappear after controlling for observable non-racial factors). The final column of Table 15 reports the statistical inferences that drivers would make about the relative likelihood of being stiffed. Here, we find that both rational and irrational discriminators would make largely the same kind of inference: AfricanAmerican and Hispanic passengers are much more likely than white passengers to leave no tip. Indeed, the likelihood that these minority passengers will stiff more than whites is in all estimates on the order of 20 to 30 percentage points higher. Hyper-rational drivers would discount the importance of these racialized stiffing inferences. They would care only about expected total revenue and put no independent weight on whether part of the expectation concerns stiffing fares. But a slightly more behavioral approach suggests that the stiffing disparity might powerfully complement the overall estimates of revenue shortfalls. Incidents of stiffing are likely to be particularly vivid and salient to drivers. Moreover, stiffing might produce an independent reflex of indignation (“Why should I pick up this person, who’s so much more likely to insult me by stiffing me?”). These estimated tipping shortfalls and the indignation effects are the strongest evidence that revenue-based statistical discrimination may play a part in the observed reluctance of drivers to service minority passengers. Finally, Table 15 estimates the likelihood that different passenger types will ask to be driven to far suburbs. The idea here was to assess the inference that a cab driver would likely make about the cost of having to deadhead back without a return fare. In an independent analysis, we did find that white passengers asked to go to drop-off neighborhoods that were slightly more likely to have a dispatch pickup request than the drop-off neighborhoods of the average African-American or Hispanic passenger.121 But these drop-off disparities are not good measures of the true deadhead cost because most New Haven drop-off neighborhoods are so close to high dispatch areas. effect when, among other things, pick-up location and age are controlled for. See bottom panel of Table 13 and text infa. 120 Although not for Asian passengers. See supra note 119. 121 The average white passenger asked to be dropped off in neighborhoods that generated 15.2% of the pickup dispatch requests, while the average African-American and Hispanic passenger asked to be dropped off in neighborhoods that generated only 11.1 and 9.1 percent of the pickup dispatch requests respectively. 44 Accordingly, Table 15 focused instead on a dimension where the deadhead cost was more substantial. Fewer than 1% of our pickup dispatches were to the far suburbs. So a cabdriver dropping off in these suburbs had virtually no chance of picking up a return fare for the long ride back to New Haven. On this dimension, we found that either a rational or irrational discriminator would infer that African Americans were statistically less likely to be dropped off in the far suburbs. So at least on this dimension, AfricanAmerican passengers should be favored (in comparison with white passengers) as having a lower expected deadhead cost on the return trip. But on net, deadhead inferences are likely to be second-order effects. White passengers are only about 5 percentage points more likely to ask to go to the far suburbs. We doubt that this small percentage point difference translates into a substantial difference in expected revenue.122 In the end, these estimates of racial inferences suggest that a previously unreported form of statistical discrimination may be driving some of the well-documented reluctance of cab drivers to serve minority passengers. Instead of cost-based inferences about the probability of crime, driver discrimination may in part be actuated by revenue-based inferences about the likely tips that will be earned. On net, rational discriminators would expect that African-American tips will be 56.5% less than white passenger tips,123 and that this tipping shortfall causes the overall revenue from an African-American passenger to be 13.8% less than that of a white passenger.124 Analogous calculations suggest that rational discriminators would expect Hispanics to tip 42.2% less than whites, which represents a 7.9% shortfall in revenue.125 Moreover, the size of the inferences that rational and irrational drivers would make about short falls in minority tipping are an order of magnitude larger than the inference that rational drivers might make about the heightened crime risk of serving African-American One way to get a crude handle on this magnitude is to estimate the effect of assuming that a far suburb fare on average forced a driver to sacrifice an average size fare and tip. 5% of the average far suburb fare amount represents only 52 cents and this amount is almost surely inflated because it does not take into account the increased revenue to the cab driver of driving the passenger to the far suburb in the first place. On net, drivers tend to look on far suburb trips as “good news.” 123 The rational-discriminator tipping shortfall of 95.3 cents divided by the predicted white passenger tip (evaluated at the means of the non-passenger race variables) of 1.68 equals 56.5%. 124 The rational-discriminator revenue shortfall of $2.06 divided by the predicted white passenger amount paid (evaluated at the means of the non-passenger race variables) of $14.99 equals 13.8%. 125 Table 16 summarizes the statistical shortfall inferences for all four minority passenger types: Table 16. Estimating the Percentage Shortfalls in Tips ($) and Total Revenue ($) (relative to White Passengers) that Would Be Inferred By "Rational Discriminators" Inferred Percentage Shortfall Passenger Total Race Tip ($) Revenue ($) Passenger Black 56.5% 13.8% Passenger Hispanic 42.2% 7.9% Passenger Asian 25.9% 6.5% Passenger Other 51.3% 14.6% 122 45 passengers. Metrocab informs us that there have not been more than 5 robberies of cab drivers in any recent year.126 There are approximately 3000 fares in New Haven each day – suggesting that there is one robbery for every 219,000 fares. Even if we assume that all robberies are committed by minorities, the inferred additional cost of serving minority passengers would only be 3.8 cents per fare.127 Of course, for irrational or particularly risk-averse drivers, inferences about the additional crime “costs” of serving minorities might loom particularly large. But the difference in the magnitude of what a rational discriminator would infer about tipping shortfalls and heightened crime costs is striking. This then is our core evidence that the perceived minority tipping disparities may be a cause of driver discrimination. Our data do not allow us to test whether drivers in fact make these kinds of inferences or whether such inferences translate into discriminatory behavior by the cab drivers against minority passengers. But inferred disparities of this magnitude might be responsible for at least part of the driver discrimination. A movement toward mandated tipping service compris regulation by reducing this perceived racial disparity in tipping might accordingly reduce the amount of revenuebased discrimination. 3. Countervailing Effects This paper has identified two new reasons that militate in favor of mandated-tipping regulation. But showing that service compris might reduce two different types of disparate racial treatment does not prove that this intervention would be on net beneficial. In this section, we briefly consider two countervailing effects that militate against our proposal. First is a countervailing civil rights effect. Mandating that a tip be included in the fare could make taxis too expensive for poor people (who are themselves disproportionately people of color). The current tipping norm may provide less well-off citizens the opportunity to purchase transportation services at an effectively discounted price if they choose not to tip the driver. A service compris regime may reduce two forms of disparate treatment but it may simultaneously also create a disparate impact against minorities. Hence, there may be a disparate impact/disparate treatment tradeoff. African Americans may find it easier to find a cab under service compris, but less able to afford the offered ride. 126 127 Interview with Metrocab executive (March 12, 2003). The probability of a robbery (.0000046) multiplied by a standard measure of robbery costs ($8416) yields an expected cost of .038 (1997) dollars per fare. See Ian Ayres & John Donohue, Shooting Down the “More Guns, Less Crime” Hypothesis, STANFORD L. REV. (forthcoming 2003) (estimating dollar impact of crime). If we make the even more conservative assumption that all the crimes are also aggravated assaults, the additional cost of serving minority passengers increases to 11.5 cents per fare. But the Metrocab official emphasized that the most serious crime committed against cab drivers in New Haven in recent years was merely theft – so inferring the additional costs of violent crime seems unwarranted in assessing the inferences of a rational discriminator. Of course, drivers may look outside New Haven in assessing the risk of harm. A single highly publicized murder, for example, could have national repercussions on riskaverse drivers. 46 Traditionally, disparate treatment violations have been seen as the more central civil rights concern.128 How we as a society choose to manage this tradeoff may also turn in part on whether we view a tip as a discretionary gift to a driver or as an earned portion of driver compensation. If the latter, it is hard to be too concerned about retaining a system that facilitates two types of disparate treatment so that a subset of passengers have the opportunity to chisel on paying for their service. The availability of alternative, affordable public transportation would also seem a relevant consideration.129 The second countervailing effect concerns service incentives. Requiring “tip included” decals might reduce cab drivers’ incentives to provide high-quality service. While the service theory is certainly coherent,130 its empirical plausibility is less certain. Researchers have found that patrons’ perception of service accounts for less than two percent of the variability in restaurant tipping percentages.131 They’ve also discovered that tipping is not significantly related to servers’ or third-parties’ evaluations of the service.132 However, most of these studies suffer from the problem of restricted range; the potential for a tip (or the threat of being stiffed) may induce servers to provide essentially uniform, excellent service. Finding a weak correlation in such circumstances says almost nothing about how much the quality of service might decline if discretionary tipping were eliminated. High-quality service is also often observed in “pooled house” restaurants where the servers’ individual incentives are already substantially dampened: Even if tips were a purely economic message sent from diner to waiter, the widespread practice of pooling tips jams the signal. In what is known as a “pooled house,” which includes almost all fine-dining restaurants in New York and many in such cities as Toronto, Montreal and Vancouver, a tip does not usually go directly into the pocket of the waiter. Instead, all cash and credit-card tips are added up at the end of the evening and distributed to waiters, runners and busboys according to a point system. A point is the total tip pool divided by the total hours worked by the pool participants.133 For example, the disparate impact cause of action was only expressly added to the text of Title VII in 1991. See Civil Rights Act of 1991, 25 U.S.C. § 1301(4) (2000). 129 Examining how normal, periodic increases in fares impact taxi-cab ridership could provide some basis for estimating the potential negative effects of our proposed 15% fare hike. So too policy-makers would be wise to consider how switches to “service compris” regimes have actually worked in practice. 130 However, with regard to non-repeat customers, there remains the question of why a server would work harder, knowing that the one-off customer has no economic rationale to tip. 131 Michael Lynn & Michael McCall, Gratitude and Gratuity: A meta-analysis of Research on the ServiceTipping Relationship, 29 J. SOCIO-ECONOMICS 203 (2000). 132 Id. at 205; see also Afar, supra note 10, at 5 (“When asked about it hypothetically, [people] therefore indicate a large sensitivity of tips to service quality. When faced with an actual tipping situation, however, the social pressure and the embarrassment that one feels when he tips poorly bring people to tip for poor service more than they thought they would tip when asked about it hypothetically.”). 133 Grimes, supra note 12. As a theoretical matter, the “pooled house” may actually be a better way to induce quality service. Although it diminishes the incentives for the waiter, it may more than compensate for that by better motivating busboys and runners and by instilling esprit de corps. 128 47 Other countries without tipping traditions – or without traditions of tipping nearly as significant amounts – seem to do just fine without granting so much payment discretion.134 Indeed, the earlier discussion on the racial antecedents of the United States tipping norms raises the possibility that the American practice may be more a vestigial attribute of our racial subordination and have less to do with a way of giving servers appropriate incentives. Conclusion The word “TIP” is thought by some to have originated in British pubs where signs with just these three letters were posted on boxes as a reminder that gratuities were welcome. The letters were an acronym for the phrase “to insure promptness.”135 But the evidence from this paper is suggestive of a new acronym, “to insure prejudice.” In this preliminary study (of just one thousand fares in a single city), we have shown that discretionary tipping facilitated prejudice in two different ways: it allowed (i) customers to discriminate against minority drivers and (ii) possibly gave cab drivers a revenuebased incentive to refuse to pick up minority passengers. Changing to “service compris” regulation would predictably mute both types of discrimination, although such a change would likely discourage some poorer passengers from taking taxis and could reduce the quality of taxi service provided. This paper also provides a test of you the reader. The “minority driver” and “minority passenger” results are a natural place to test your own tendency to pick and choose among civil rights results. Informally, we find that liberals tend to minimize the minority-passenger effects – immediately asking about omitted-variable issues (“Might it be poverty instead of race?”). Conversely, we find that conservatives tend to minimize the minority-driver effect – immediately asking about omitted-variable issues (“Might minority drivers provide poorer service?”). Omitted variable questions are almost always appropriate, but the emphasis of different questioners seems interestingly non-random. While there are some important qualifications to our results,136 this paper provides an initial test of consumer-side discrimination. It also provides the first quantitative Information on international tipping norms can be found at www.quinwell.com/vaca/tipping.html. See also www.todotravel.com/english/tips/items/money/t-mon3.htm (no tip is expected in “Australia and New Zealand (except for top restaurants), Scandinavia, Singapore, mainland China, Japan, South Pacific islands, and in Zambian hotels”). 135 See E. L. Konigsburg, A View From Saturday (1997); www.metroactive.com/10.24.96/dining9643.html (“An oft-repeated story is that tipping [is] supposedly an acronym for ‘to insure promptitude’”). But this acronym etymology is probably apocryphal: In many languages, the words for 'tip' are associated with drinking, because in many countries the tip began as a gratuity to enable the tippee to buy himself a drink. In French 'pourboire' means literally 'for drink;' the German 'trinkgeld' is 'drink money;' the Spanish 'propina' is from 'propinar' meaning invite to drink; Russia's 'nachai' is the equivalent of 'for tea;' and the Chinese 'cumshaw' is 'tea money.' It may be reasonable to surmise the word tip is a short form of 'tipple'--to drink. K. SEGRAVE, supra note 18, at 5. 136 See supra Part IV (particularly with regard to censoring). Among other things, future researchers in this area would do well to consider: randomizing the selection of participating drivers, more closely monitoring 134 48 estimates of rational and irrational statistical discrimination. It is our belief that exposing the dual racial determinants of tipping suggests more generally that consumer discretion in retail transactions may give rise to unexpected civil rights concerns. the drivers to avoid omitted data, and obtaining better controls for driver quality and passenger wealth (perhaps by surveying passengers directly). 49