An Analysis of Employer Learning and Statistical Discrimination in Wages in Firms Unlikely to Comply with Equal Employment Opportunity Legislation

This dissertation explores the possibility that statistical discrimination in wages may be more likely in firms that are not subject to antidiscrimination legislation. Federal antidiscrimination regulations differ by size and type of firm, resulting in lower costs to discrimination in firms not subject to the laws or less likely to be prosecuted under them. Less intensive screening practices in these firms may also increase the benefit to statistical discrimination in wages. Utilizing employer-reported results of equal employment opportunity law compliance from the Multi-City Telephone Employer Survey, I predict the likelihood that an individual in the National Longitudinal Survey of Youth 1979 (NLSY) was employed by a firm that considers equal employment opportunity legislation in their hiring. Using observations from NLSY years 1986-2000 and an empirical test for employer learning and statistical discrimination developed by Altonji and Pierret (1997), I examine statistical discrimination in wages by education and race for male employees in firms that are not likely to actively comply with antidiscrimination legislation. I find that statistical discrimination by education is evident for black males but not for white males, suggesting that information for black males in the labor market is inferior. Statistical discrimination by education is then shown to be more pronounced for black males employed by firms that are unlikely to actively comply with equal employment opportunity legislation. An investigation of statistical discrimination by race in these firms is inconclusive in both a public learning framework and an asymmetric learning model. These results show that a complete analysis of statistical discrimination must consider differences in the incentive firms have to discriminate. While there is not clear evidence of statistical discrimination by race, the results do illustrate that productivity information for black males in firms that are unlikely to comply with antidiscrimination legislation is inferior to the information these firms have about white males. Equal employment opportunity laws may correct a market failure of underinvestment in information by firms, suggesting that policies to enhance the information available to firms not subject to these laws may be beneficial.