Economic literature has identified two potential types of racial discrimination in sports careers: exit discrimination and wage discrimination. The authors test for both types of discrimination in National Basketball Association (NBA) data in two ways. First using a modified Heckman procedure, they control for potential survival bias that may arise from exit discrimination in panel data and could possibly bias the coefficients of wage equations. Not controlling for survival bias could lead to false conclusions concerning the presence of pay discrimination. Using 1990-2008 NBA data, the authors fail to find any evidence of either pay or exit discrimination in the NBA utilizing this new technique. In one specification, however, a negative coefficient on the white dummy is found after controlling for height and being foreign-born suggesting that reverse discrimination is present. Yet, using a subset of the panel data used to examine pay discrimination in the NBA with career earnings the authors find that there is a pay premium paid to White players over their career in the magnitude of 16%–20%, ceteris paribus. Neither of these results, however, is robust and highlights the pitfalls of using the residual method in measuring both pay and exit discrimination.
