Wage differentials between different demographic groups are a prevalent phenomenon in most developed countries. In spite of powerful equal-pay legislation and other anti–discrimination measures, these gaps remain considerable, even after controlling for workers’ qualification and experience. While wage differentials at the job cell level (occupation–establishment) are small, occupational segregation and establishment segregation are important causes of the wage gaps, and there is only weak tendency of segregation to decline.
The aim of this paper is to demonstrate that equal–pay legislation can, by itself and under certain circumstances, increase job segregation, reduce wages of minority workers, and contribute to the persistence of discriminatory behaviour on the side of employers.
Economic theories of labour market discrimination can be divided into two strands. One is statistical discrimination in the presence of imperfect information (Phelps (1972) and Arrow (1973)), the other is taste–based discrimination where some agents have a distaste against interacting with minority workers (Becker (1957)). Such discrimination may occur on the side of employers, co–workers or customers. As Becker pointed out, taste–based employer discrimination can have no permanent effect on the wage gap in a competitive labour market.
In the long run (i.e. under free entry or perfect capital mobility) there will be enough non–discriminatory employers who hire all minority workers at the same wage as majority workers. In a sense, segregation helps to reduce and eventually eliminate discrimination in the competitive model (see Cain (1986), p.711-712). In an imperfectly competitive labour market, however, such a result need not hold. First, monopsonists (and similarly monopsonistic competitors) may easily discriminate if labour supply elasticities differ between demographic groups (Robinson (1933)). Second, taste–based employer discrimination can persist in models of monopsonistic wage competition since discriminators earn positive profits and will not necessarily be forced out of the market by non–discriminatory employers (Bhaskar et al. (2002)). Search models, as another departure from the competitive paradigm, also permit persistence of taste–based discrimination (Black (1995), Bowlus and Eckstein (2002), Lang et al. (2005)).