Academic journal article
By Hill, James Richard
Quarterly Journal of Business and Economics , Vol. 43, No. 1-2
The use of the residual method to measure wage discrimination is fraught with problems. Oaxaca (1973) stated:
If it were possible to control for virtually all sources of variation in wages, one could pretty well eliminate labor market discrimination as a significant factor in determining wage differentials by sex (or race) ... The other extreme is to control for virtually nothing and thereby minimize the role of productivity differences, [DELTA]Z. This is tantamount to declaring at the outset that the two labor inputs are near perfect substitutes and thereby attributing virtually all of the observed wage differential to labor market discrimination. (p. 699)
Between these two extremes the researcher must find adequate and appropriate proxy variables to measure the productivity of workers as well other market factors that could impact wages. Accomplishment of this task still can lead to an understatement or overstatement of wage discrimination using wage equation residuals. For example, if female or minority workers do not invest as heavily in education or training as a rational market response to a lower rate or return on investment due to market discrimination, then the resulting estimates using the residual methodology understate the overall effect of discrimination on wages. On the other hand, if race or sex is correlated with an omitted variable that helps to explain productivity, then the resulting estimates of wage discrimination overstate wage discrimination or can lead to a false positive conclusion of the existence of wage discrimination.
Labor market studies of wage discrimination in professional sports do not suffer from data shortages. According to Kahn (2000), the plethora of statistics designed to measure the performance of our professional athletes provides a virtual labor market laboratory for analysts. Problems of correlation between race and productivity measures may still present problems.
This study uses panel data over the 1990-2000 seasons to examine for evidence of pay discrimination. (1) Models are estimated using the standard OLS technique and the random effects formulation.
Kahn and Sherer (1988), using salary data from the 1985-86 season, found evidence that black players, who represented almost 75 percent of team rosters, received approximately 20 percent less in pay, ceteris paribus, compared to white players. The source of this discrimination was suggested to be fans. An article by Brown, Spiro, and Keenan (1991) using 1984-1985 season salary data found similar results, about a 15 percent pay premium for white players; another article by Koch and Vander Hill (1988) using the same salary data as Brown, Spiro, and Keenan (1991) but more dependent variables estimated about a 9 percent pay premium for whites. Both studies also cite fan discrimination as the driving force behind pay discrimination.
Several articles (Dey, 1997; Gius and Johnson, 1998; and Bodvarsson and Brastow, 1999), using salary data from the late 1980s and 1990s failed to find pay discrimination in this more recent time period. Hamilton (1997) found evidence of racial pay differences only at the upper end of the 1994-1995 season's salary distribution. Dey (1997) replicated the regression models of Brown, Spiro, and Keenan (1991) for each year of the period 1987-1993 as well as a pooled sample. In two of the years and in the pooled sample there was a premium for white players, but this premium disappeared when a dummy variable for centers was added to the model. Dey (1997) concludes that the salary premium for whites reported by Brown, Spiro, and Keenan (1991) may actually be a pay premium for centers because they were more likely to be white. Kahn and Sherer (1988) did have position dummy variables for center and forward, however, and still found pay discrimination.
Bodvarsson and Brastow (1999) found evidence to suggest that the disappearance of the pay discrimination was the result of a decrease in owner monopsony due to the negotiation of a new NBA collective bargaining agreement between the owners and players in 1988 combined with the addition of four new teams. …