JOURNAL OF SPORTS ECONOMICS / May 2005
sesses a propensity to discriminate, a team’s management would be less likely to accommodate such customers if it could fill its arena seats with nondiscriminating customers. It is more likely that nondiscriminating customers are marginal demanders of seats when the market-area population is large relative to stadium size. In the regressions below, we account for this by entering the ratio of stadium capacity to total population in the market area as an independent variable. We expect a positive relationship between the ratio of stadium capacity to population and the percentage of White players on a team. A time-trend variable controls for time-dependent changes in the racial composition of teams.
Table 3 shows OLS regression results for the three dependent variables that measure team racial composition. As a determinant of overall team composition, the percentage White of the SMSA population (POPWHITE) is significant at the 95% confidence level. It is not significant in the bench regression, however, sug- gesting that the primary focus of customer discrimination is on starters (for whom this effect is significant at the 90% level).7 The stadium capacity variable is signifi- cant at the 90% level in the team and starter regressions, whereas the time trend is significant at the 99% and 95% levels in the team and bench regressions, respec- tively, but not in the starter regression.8 As documented earlier, the ratio of White to Black starters remained relatively constant throughout the decade.
RACIAL MATCHING AND REVENUE GAINS
Basketball teams generate revenue primarily from two sources: television con- tracts and home-game attendance. The NBA bylaws and constitution require that teams pool and share television and cable revenue, but allow teams to retain all home-game attendance revenue (less a 4% assessment fee). For our purposes, the NBA’s revenue-sharing arrangement is important because a substantial fraction of a team’s revenue comes from local sources.9 Accordingly, if customer demand for tickets depends on a team’s racial profile, profit-maximizing managers will inter- nalize the revenue impact of the local market’s racial preferences.10
In a cross-sectional analysis, the empirical effect that team racial composition has on ticket revenue is complicated. Assuming that Black and White players are of equal performance quality, finding no relation between team racial composition (TWHITE) and ticket revenue is consistent with either of two hypotheses: (a) Fans are indifferent to race so that team race per se does not affect ticket revenue, or (b) fans are not indifferent but markets are in equilibrium with regard to race except for random errors. Similarly, finding a positive relationship between TWHITE and revenue also raises two possible interpretations: (a) Black players and White play- ers are of equal quality, but customers are not indifferent to race and demand a greater percentage of White players on the team than managers supply; or (b) con- sumers do not care about race, but players are sorted among teams so that the aver- age quality of White players on a team is positively correlated with the percentage of the team that is White (TWHITE).