10.1177/1527002503262641Burdekin et al. / NBA FANS INDIFFERENTJOURNAL OF SPOR TO RATSCE? ECONOMICS / May 2005 ARTICLE Are NBA Fans Becoming Indifferent to Race? Evidence From the 1990s RICHARD C. K. BURDEKIN Claremont McKenna College RICHARD T. HOSSFELD Duke University School of Law JANET KIHOLM SMITH Claremont McKenna College Previous work found the racial composition of NBA teams to be positively correlated with the racial composition of their metropolitan markets in the 1980s. We find continued evi- dence of this relationship during the 1990s, with accompanying revenue gains from the inclusion of White players on teams located in whiter areas. And, as the number of White players declined significantly throughout the decade, the revenue product of a White player actually increased on the margin. The tendency for top-performing White players in the NBA to locate in cities with larger White populations also is consistent with their higher marginal value in such locations. Keywords: customer discrimination; race; sports; National Basketball Association Applied to professional sports, Becker’s (1971) concept of racial discrimination by customers is embodied by fan preference for watching players of their own race. The customer-discrimination hypothesis implies that teams lose revenue and prof- its when they adopt color-neutral hiring practices. Accordingly, unlike discrimina- tion by employers and by fellow workers, discrimination by customers is unlikely to disappear even in the long run (Nardinelli & Simon, 1990). Although discrimina- tion by customers generally cannot be disentangled from other influences on hiring AUTHORS’NOTE: The authors thank Heather Antecol, David Berri, Bill Brown, Eric Helland, Harold Mulherin, Mark Partridge, Jennifer Ward-Batts, Marc Weidenmier,Richard Smith, Tom Willett, and two anonymous referees for helpful discussion and comments. JEL codes: J15, J71. JOURNAL OF SPORTS ECONOMICS, Vol. 6 No. 2, May 2005 144–159 DOI: 10.1177/1527002503262641 © 2005 Sage Publications 144 Burdekin et al. / NBA FANS INDIFFERENT TO RACE? 145 practices and wage scales, Holzer and Ihlanfeldt’s (1998) survey evidence from four major metropolitan areas suggested that customer discrimination remains quite pervasive—with strong discrimination effects emerging when employees have significant contact with customers. It is hard to imagine a business where employees are more “visible” to custom- ers than professional basketball. Consistent with the customer-discrimination hypothesis, a number of studies, using data from the 1980s, found that National Basketball Association (NBA) teams located in more White-dominated market areas consistently had a disproportionately large share of White players (Bodvarsson & Partridge, 2001; Brown, Spiro, & Keenan, 1991; Burdekin & Idson, 1991; Hoang & Rascher, 1999; Koch & Vander Hill, 1988).1 The decline in the number of White players in the NBA since the 1980s raises the possibility that professional basketball teams may have become more “color-blind” in their hiring practices throughout the years. Evidence on whether or not customer-based dis- crimination has, in fact, declined in the NBA could offer a valuable litmus test of the relevance of customer discrimination in today’s economy—given the unusual visi- bility of not only the players themselves but also NBA teams’ hiring decisions by race. In addition to using NBA team data from the 1990s to reexamine the relationship between the racial composition of teams and the racial composition of their market areas, we address the question of whether teams have effectively boosted their attendance and revenues by “matching” the racial profile of their players with their market’s racial profile. Prior evidence on this point is mixed. Burdekin and Idson (1991) and Hoang and Rascher (1999) found that a positive match between team racial composition and the racial composition of the SMSA area increases atten- dance. Kahn and Sherer (1988) added that White players increased attendance more than enough to compensate for the White salary premium they identified for the 1980-1986 period. On the other hand, Dey (1997), using data from 1987 to 1993, and McCormick and Tollison (2001), using data from 1980 to 1988, saw no relationship between the racial match and NBA attendance.2 Our approach has three parts. We first analyze trends in the racial composition of NBA teams and document the distribution of players, classified by race, among teams. Differences in the performance of Black and White players can shed light on whether teams in the league engage in “window dressing” by adding White players to their rosters who are not as strong as Black players. We next examine what deter- mines the racial composition of NBA teams and assess whether a team can increase its attendance by matching its racial composition with the racial composition of the metropolitan area where the team is located. This is based on the premise that not just Whites but also Blacks prefer to watch athletes of their own race—and that sat- isfying these preferences leads to attendance gains. Finally, we look at the role of racial factors in the trading of NBA players and identify a tendency for the most skilled White players to stay in whiter cities. 146 JOURNAL OF SPORTS ECONOMICS / May 2005 TEAM RACIAL COMPOSITION AND PLAYER PERFORMANCE Annual data from the 1990-1999 period bring out the sharp decline in total White representation in the NBA, falling from just more than 25% in the 1990-1991 season to 20% in the 1998-1999 season (see Figure 1). This result corresponds with declining White representation on the bench, which fell from 30% to 23%. There has been little change in the percentage of starters who are White, however, which remained below 15% across most of the sample period. As Table 1 shows, White players and Black players remain unevenly distributed among NBA teams. Although, on average, the Chicago Bulls and Utah Jazz teams were just over 33% White during the 1990s, at the other extreme, the New York Knicks and Toronto Raptors were under 10% White. Available data on total min- utes played during the 1996-1997 through 1998-1999 seasons suggest an even greater disparity (see column B of Table 1). The percentage of total minutes accounted for by White players ranges from a high of 44.7% for the Utah Jazz to4% for the Toronto Raptors. Admittedly, many factors besides customer preferences could lie behind the varying racial compositions of NBA teams—including the race of available draft picks, existing contractual obligations, the availability of free agents, and increased player demand arising from two expansion teams (Toronto Raptors and Vancouver Grizzlies). There is no evidence, however, at least in recent years, that fan preferences for watching players of their own race have led to inferior White players entering the NBA in place of more skilled Black counterparts. Not only has the overall represen- tation of White players in the NBA declined in recent years, but also the perfor- mance levels of White players and Black players appear similar. In Table 2, we use annual player statistics for 1996-1997 through 1998-1999 to compare the average performance of Whites and Blacks. We examine starters and bench players sepa- rately throughout a range of measures—points scored, assists, total rebounds, blocks, and field-goal percentage—that have consistently been found to impact sal- ary (see Berri, 2003).3 Starters are the top five players on each team in terms of min- utes played per game, whereas bench players constitute the remainder. Significant differences between average Black player and average White player performance emerge in only 3 cases out of 10, and there seems to be no systematic tendency for players of one race to outperform those of the other. Indeed, White starters outper- form Black starters in field-goal percentage but are statistically tied in all other cate- gories. Black bench players outperform their White counterparts in assists and total points per game but are statistically equal in the other three categories. We base our White and Black performance-level findings on player data from the last three seasons of the 1990s, a period during which there is little evidence of overall salary discrimination against Blacks in the NBA. Indeed, the NBA appears to have changed markedly in this regard since the 1980s, when a number of studies Burdekin et al. / NBA FANS INDIFFERENT TO RACE? 147 35% 30% 25% 20% 15% 10% Bench 5% Starters Team Percentage of Players Who Are White 0% 1990 1991 1992 1993 1994 1995 1996 1997 1998 Figure 1: Average Racial Composition of NBA Teams Over Time NOTE: A regression of percent of team that is White (TWHITE) on a time trend generates a negative coefficient (t statistic of 2.80, significant at the .01 level). A corresponding regression of percentage of bench that is White (BWHITE) on time generates a negative coefficient (t statistic of 3.30, significant at the .01 level). The coefficient in the regression of percentage of starters, however, who are White (SWHITE) on time is not statistically different from zero (t statistic of 0.36). found that Whites were paid more than Blacks for the same level of performance (see Kahn, 2000).4 Although most of the salary-based studies focus on aggregate measures of performance and salaries, a few consider how the fan base in a metro area may differentially affect salaries. For example, Bodvarsson and Partridge (2001) found that Black players in two seasons (1985-1986 and 1990-1991) were paid more in areas where the Black population was higher. They argued that this finding may reflect “black fans wanting to see a team with a greater percentage of blacks” (Bodvarsson & Partridge, 2001, p. 413). Although no such population effect was significant in their regressions of White player salary levels, the possibil- ity that Black-White salary levels may vary with the racial composition of the mar- ket area is consistent with the customer-discrimination hypothesis.