JOURNAL OF SPORTS ECONOMICS / May 2005
team. Thus, not only do White players seem to congregate in teams playing in Whiter cities, but those same teams are most loath to trade them away.
Early studies using 1970s data find similar results. See Markmann (1976) and Karabel and Karen
Discrimination by customers may have an impact not only at the professional level but also at the
college basketball level. Brown and Jewell (1994, 1995) found that White players in the 1988-1989 NCAA season generated significantly more team (school) revenue than did Black players of comparable ability.
3. Classifying players according to their playing position does not alter the overall picture apparent in Table 2. Results are provided in a separate appendix that is available from Janet Smith (e-mail: email@example.com). Because team managers do not select players based on only one skill, we also examined whether there is any difference in the breadth of skills that Black players and White players possess on average. Correlations between pairs of performance measures are the same sign and of similar magnitudes for Whites and Blacks.
4. Hoang and Rascher (1999) also found evidence of exit discrimination in the 1980s with White players facing 36% less risk of being cut from a team than did Black players with otherwise similar char- acteristics. As Kahn (2000) documented, however, data from the early to mid-1990s generally offer much less support for salary discrimination against Blacks in the NBA (see, for example, Bodvarsson & Brastow, 1999; Dey, 1997; Gius & Johnson, 1998; Hamilton, 1997).
5. Bodvarsson and Partridge (2001) also pointed to possible coworker discrimination through the effect of team composition on salaries. They found that White players are paid more, ceteris paribus, when the team becomes less White. It is not clear, however, whether any such effect really implies coworker discrimination. It could instead reflect an attempt by teams to maintain at least some measure of ethnic diversity. Former owner of the Cleveland Cavaliers, Ted Stepien, even went on record claiming: “You need a blend of black and white. I think that draws, and I think that’s a better team” (see Karabel & Karen, 1982, p. 24).
6. It is also possible that prejudiced Black fans may have less interest in seeing foreign-born Black players on the team, again mitigating against finding significant effects of race in our empirical analysis.
7. We also estimated the relationships using clustering (where each of the 29 teams is a cluster) to consider the possibility that observations are not independent within clusters. The results comport with those in Table 3, although the coefficients are less significant. For the first equation, encompassing all players, the coefficient on POPWHITE is significant at the 0.25 level. When we estimate just on the first and last years of data, where clustering is less of an issue, the coefficient on POPWHITE is significant at the 0.01 level. When we do allow for clustering in these years, the coefficient is significant at the 0.15 level.
8. Our empirical finding (of a statistically significant relationship between the racial composition of teams and the racial composition of corresponding metropolitan areas) holds whether or not we exclude the two expansion teams, for which data are available from 1995 onward.
9. Although the influence of national broadcasting revenue has grown in recent years, more than 65% of the average NBA team’s revenue still came from local sources in 1996 (Wilson, 2001). Teams receive all revenue from the sale of tickets to regular-season home games (subject to the NBA gate assessment of 4%) and no revenue from sale of tickets to regular-season away games. Generally, teams also retain all revenues from the sale of tickets to home exhibition games. Based on SEC annual report filings by the Boston Celtics, we find that during the period 1993-2000, ticket sales represented 49.3% of total revenue and TV revenues represented 39.3%. Other sources of revenue made up the remainder.