Incentive Effects of Affirmative Action

Abstract

This article illustrates with a formal economic model a concern often raised by critics of affirmative action—that the policy may discourage its beneficiaries from acquiring work skills. Ironically, this can happen for reasons analogous to those evoked to explain why discrimination may discourage its victims from investing in skills: when skilled workers are less likely to succeed, fewer find it worthwhile to become skilled. Similarly, when unskilled workers are more likely to succeed, fewer deem it necessary to become skilled. Discrimination can lead to the former situation; affirmative action can lead to the latter. The analysis shows how affirmative action can lead employers to patronize minority workers, that is, hold them to a different standard. This patronization can have the effect of making skill acquisition less beneficial for minority workers. The labor market conditions under which this counterproductive effect of affirmative action is most likely are identified.

Get full access to this article

View all access options for this article.

References

1. For an extended discussion of problems with a pure color-blind approach to public policy in the face of racial inequality, see my essay “Why Should We Care about Group Inequality,” Social Philosophy and Policy, 5(1):249-271 (Autumn 1987). I also provide there an informal discussion of some negative unintended effects of affirmative action other than the one analyzed in the present article. An important theme in that essay, having answered in the affirmative the question “Should `color' ever be taken into account?” is that preferential treatment is often not the best method of doing so. I make the case that targeting social service benefits to disadvantaged minorities may be a superior means of taking into account the history of racial discrimination.

2. The argument set out in the model is largely expressed verbally and is, therefore, less rigorous than the mathematical model that it approximates. Due to space limitations, mathematical proofs of the propositions have been omitted. They are available from the author on request, or, for a more complete treatment, see Stephen Coate and Glenn C. Loury , “Will Affirmative Action Eliminate Negative Stereotypes?” mimeographed (Cambridge, MA: Harvard University, May 1991).

3. Specifically, let p ₁ (p ₀) be the probability that an investing (noninvesting) worker gets an unclear test result. Then 1 - p ₁ is the probability that an investing worker passes the test, and 1 - p ₀ is the probability that a noninvesting worker fails it. I assume p ₀ < p ₁.

4. A more direct way to eliminate discrimination would be to forbid the employer to treat whites and blacks with unclear tests any differently. That is, the government could merely insist on color-blind behavior from the employer, without regard to results. This would be difficult to enforce in practice. The government would have to observe all information upon which an employer might base his assignment—interviews, work history, and so on—to determine if he is really treating blacks and whites the same. In most employment situations this is not possible. The analysis offered here applies to those situations where affirmative action takes mainly a results-oriented rather than a process-oriented form, with the government's focus being on the numbers hired, not the hiring procedures.