Affirmative Action, On and Off

"The Persistent Effect of Temporary Affirmative Action." Conrad Miller. American Economic Journal: Applied Economics, July 2017, vol. 9, no. 3, pp. 152-90.

In 1965, President Lyndon B. Johnson signed Executive Order (EO) 11246, requiring that federal contractors take "affirmative action" to prevent discrimination in their hiring and employment practices; firms of a certain size and contract value were subject to more strict requirements, such as specifying goals and timetables for hiring of minorities. Conrad Miller, an economist at the University of California, Berkeley, has described EO 11246 as "arguably [one] of the most controversial labor market interventions in U.S. history."

Current theoretical models of affirmative action in hiring tend to focus on statistical discrimination and human capital accumulation: If employers believe that members of a minority group are less productive or if employers have difficulty evaluating minority candidates, their hiring will be biased against members of that group. That minority group as a whole would then have less incentive to invest in human capital, such as education and training. If so, temporary affirmative action might have persistent positive effects on minority hiring by encouraging minority human capital accumulation.

Other economic models of affirmative action, often cited by skeptics, treat the policy as introducing inefficiency into labor markets by forcing employers to lower their hiring standards for minorities. Still another possibility is that EO 11246 has simply had little effect, possibly as a result of limited enforcement. An innovative 2016 article by Fidan Ana Kurtulus of the University of Massachusetts, Amherst compared contractors to noncontractors, finding that the regulation had only small effects — a less than 0.1 percentage point increase in a firm's black share of employees that disappeared as soon as four years after that firm became a federal contractor.

But in a recent article in the American Economic Journal: Applied Economics, Miller has argued that the more proper comparison is between firms that had ever been federal contractors and firms that had not; if affirmative action did have lasting effects on individual firms, then simply comparing contractors to noncontractors would obscure effects at firms that stopped contracting with the federal government but continued to increase their minority hiring.

Such an effect — both persistent and large — is what Miller found. In the five years after a firm became subject to EO 11246, its black share of employees increased by an average of 0.8 percentage point. To provide perspective, Miller noted that "a 0.8 to 1.3 percentage point increase in the black share of the U.S. workforce would eliminate the black-white jobless gap over this period." Moreover, this effect persisted after a firm was no longer subject to the regulation; in the five years after a firm stopped being a federal contractor, its black share of employees grew, on average, by another 0.8 percentage point.

Why did the firms continue to increase their minority hiring when no longer required? One possibility is that they anticipated becoming federal contractors again; if there are adjustment costs to adopting affirmative action, an employer expecting a future contract might find it best to keep complying with the executive order. Firms might also think that compliance would increase their chances of winning contracts. Miller argued that these explanations were not supported: Whether a firm won subsequent contracts did not show any relationship with their black share of employees or their persistence in affirmative action compliance.

Thus, Miller argued, the persistent positive effect on hiring of black workers suggests that compliance with the executive order was profitable: Firms' hiring of blacks might have been inefficiently low before the regulation, or there might just be multiple equi­libria for the racial composition of new hires.

Miller argued that firms may respond to EO 11246 in a more complex manner than simply lowering their hiring standards for the affected groups. He put forward a model of "screening capital," in which employers can respond to such programs by improving their recruiting and selection processes. These improvements include investments such as developing tests, employing and training personnel specialists, and building relationships with intermediaries such as employment agencies and schools.

Miller's screening capital model makes two main predictions. First, the model predicts that screening capital investments will reduce all racial disparities in individual firms' hiring rates; if employers tend to underestimate or have trouble screening certain racial groups, better screening should reduce that gap. Second, it predicts that affirmative action will increase the returns to screening capital (by improving the expected quality of minority hires).

Miller found that large employers, who tend to spend more time on screening and use more screening methods, also have a higher black share of workers among their employees. While data limitations prevented Miller from ruling out alternative mechanisms, he concluded that the evidence suggests that screening investments play a role in the persistent effects of affirmative action.