Discrimination, Wage

Discrimination, Wage

MEASURING WAGE DISCRIMINATION

THEORY

POLICY

BIBLIOGRAPHY

Wage discrimination occurs when, due to the operation of the labor market, similar workers receive different wages on the basis of race, sex, ethnicity, age, sexual orientation, or other ascribed characteristic not directly related to productivity. Workers subject to wage discrimination may earn lower wages in a given job, be assigned to low-wage jobs within firms, or employed in low-wage firms.

Not all wage differentials are discriminatory. If more group A than group B workers are willing to work in dangerous but highly paid jobs, on average, As may earn more than Bs do. But provided that wages are based on the nature of the job and not group membership, the wage differential need not reflect wage discrimination. Similarly, if As are paid less because they are less skilled, there is no wage discrimination unless their lower skills reflect their expectation that their skills will not be rewarded.

MEASURING WAGE DISCRIMINATION

We rarely measure worker productivity directly and, therefore, we cannot test whether equally productive workers earn different wages based on their group membership. Instead, we ask whether apparently similar workers of different races, sex, etc., receive different wages. The difficulty with this approach is determining the dimensions along which workers should be similar. For example, suppose we compared men and women with undergraduate degrees in biology and found that women have lower wages. Suppose we also found that for those of this group that are teachers, wages are similar for such men and women. We might conclude that many women choose to be teachers, perhaps in order to be with their children after school and during vacations. Women (and men) with these preferences accept lower wages, so there is no wage discrimination. However, perhaps women face significant wage discrimination in jobs outside teaching, so that those who do not teach are the very talented or lucky few who find high-wage jobs outside teaching. In this case, we miss the wage discrimination by comparing men and women in the same occupation.

Moreover, if firms discriminate by offering low wages to, for example, African Americans, individuals offered these low wages may respond by not working, and we will only observe wages for those African Americans employed at good wages by nondiscriminating firms. Ignoring individuals who are not working underestimates the black-white wage differential and thus the extent of wage discrimination.

Estimates of the degree of wage discrimination, and even its existence, depend critically on the factors for which we control. There are large earnings differentials between men and women, blacks and whites, and non-Hispanic whites and Hispanics if we take no other factors into account. Skin-shade studies have found that darker Hispanics and African Americans are paid less than Hispanics and Africans with lighter complexions. Controlling for cognitive test scores in high school eliminates much of the difference between black and white men, but a significant difference reappears if we also control for years of education. Differences in the earnings of black and white (employed) women are modest, but are much larger if we take account of the potential earnings of nonworkers. Much of the Hispanic–non-Hispanic wage differential can be “accounted for” by education and by knowledge of English. Career interruptions are important in “explaining” female-male wage differentials.

THEORY

In his pioneering work The Economics of Discrimination (1971), Gary Becker argued that labor market competition will eliminate wage discrimination. If, for example, blacks are paid less than are equally productive whites, unprejudiced employers will hire blacks and make more profit than do employers who hire whites. Nondiscriminating firms will expand, hire more blacks, and drive some discriminating firms out of business. The process continues until black and white wages are equalized. Some discriminating firms that hire only white workers may survive, but there will be no wage discrimination. Similarly, if workers discriminate by requiring a premium to work with blacks, they will be employed in segregated firms, but wages will not depend on group membership.

This conclusion must be tempered somewhat if customers are prejudiced. In this case, blacks will work in jobs where race is invisible and, if there are enough such jobs, receive the same wages as comparable whites. However, in areas such as professional sports, where workers have highly specialized skills and race is clearly visible, wage discrimination may persist.

If labor markets operate less smoothly than the competitive model implies, through a variety of mechanisms they can exacerbate, not eliminate, the effect of prejudice. For example, firms that announce high wages to attract many applicants may deter black applicants who anticipate losing out to white applicants. Instead, blacks may apply to low-wage jobs in order to avoid competition with whites. Wage discrimination may also persist if workers and firms act collectively and wages are governed by bargaining. Workers may coalesce to exclude those who are “different,” and firms may pay lower wages to those with less bargaining power.

Other models also permit persistent wage discrimination. In social distance models, interactions between heterogeneous groups are costly. The market minimizes such interactions and thus encourages segregation, but complete segregation is impossible. Members of subordinate groups must either adopt the dominant group’s norms of behavior and social interaction or accept lower wages.

Social distance and the absence of shared networks may also reduce the ability of employers to evaluate potential employees from other groups. In this case, employers may engage in statistical discrimination, whereby they rely more on group membership and less on information about the particular individual. Such workers have less incentive to make unobservable investments in themselves (e.g., work hard in school) and thus will earn less than do observably similar workers whom employers evaluate individually. However, they may also have an incentive to make more observable investments (e.g., years of schooling). Similar mechanisms apply when statistical discrimination reflects self-confirming stereotypes rather than social distance.

POLICY

In the United States the Equal Pay Act of 1963 outlawed payment of different wages for the same job on the basis of race, but this had little effect because most wage discrimination probably arises through workers holding different jobs and working in different firms. The 1964 Civil Rights Act forbade employment discrimination based on race, ethnicity, sex, or religion, and Executive Order 11246 required federal contractors to take affirmative action to ensure that they did not discriminate on the basis of race. Over time, the scope of civil rights legislation in the United States has been extended so that it covers, in various degrees, age, disability, and sexual orientation, as well as race, ethnicity, sex, and religion. Many countries have similar laws. Although most analysts believe that these policies reduced wage discrimination against African Americans, women, and other groups, the extent of the effect is hotly debated.

SEE ALSO Discrimination; Discrimination, Racial; Discrimination, Statistical; Discrimination, Taste for; Discrimination, Wage, by Age; Discrimination, Wage, by Gender; Discrimination, Wage, by Occupation; Discrimination, Wage, by Race

BIBLIOGRAPHY

Altonji, Joseph G., and Rebecca M. Blank. 1999. Race and Gender in the Labor Market. In Handbook of Labor Economics, Vol. 3C, ed. Orley Ashenfelter and David Card, 3143–3259. Amsterdam, New York, and Oxford: Elsevier, North-Holland.

Becker, Gary. 1971. The Economics of Discrimination. Chicago: University of Chicago Press.

Lang, Kevin. 2007. Poverty and Discrimination. Princeton, NJ: Princeton University Press.

Kevin Lang