Statistical discrimination

From Diversifying Economic Quality: A Wiki for Instructors and Departments

Revision as of 18:23, 18 August 2012 by Jvenato1 (talk | contribs)
Jump to: navigation, search

Statistical Discrimination occurs when agents use an observable characteristic of an individual to make inferences about another attribute that is relevant to the transaction but more difficult to observe. The agents' beliefs about group averages affect their treatment of an individual member of the group. This theory is typically attributed to Kenneth Arrow's 1973 work "The Theory of Discrimination" and to Edmund Phelp's 1972 paper "The Statistical Theory of Racism and Sexism. In Phelp's theory, the cause of inequality is an exogenous difference in two group's innate qualities. Without full information about different workers' productivity, the "rational" employer would make hiring decisions based on their knowledge of the applicants' group as a whole. For example, employers may be reluctant to hire and train teenage workers because their past experience with teenagers is that they only want a short-term position (e.g., a summer job). Arrow's later theory followed a similar line of thought, but did not assume that a person's or a group's traits were determined exogenously. Instead, these traits are endogenous and develop through society's beliefs about a group. For a more complete overview of statistical discrimination, read this paper, Theories of Statistical Discrimination and Affirmative Action: A Survey by Hanming Fang and Andrea Moro.

Unlike animus-based discrimination, statistical discrimination assumes that the actor is rational. His or her actions are not due to any direct prejudice or dislike for the disadvantaged group, but rather are meant to maximize the actor's utility. Though the outcomes are based on average characteristics of a group, these generalizations do result in unequal outcomes for group members. A common example of statistical discrimination is seen in the car insurance market. Because male drivers are statistically more likely to get in an accident, they are charged higher premiums. An example more pernicious than car insurance is labor market discrimination against women, particularly married women, who are statistically more likely to take time off to raise a family. Another example is racial profiling in law enforcement: minorities are more likely to be pulled over than white drivers.

Compare to animus-based discrimination, implicit associations, and institutional discrimination.