Effort, Race Gaps and Affirmative Action: A Structural Policy Analysis of U.S. College Admissions

Author

Brent R. Hickman, University of Iowa Department of Economics

Abstract

Using the college admissions model of Hickman (2010), I study the implications of Affirmative Action (AA) in US college admissions for student academic achievement (prior to college) and college placement outcomes. I argue that the competition between high-school students for admittance to college is strategically similar to a multiobject all-pay auction. The link to auctions provides access to a set of tools which create tractability and empirical power. This allows me to compare various college admissions policies in terms of three criteria: (i) the induced level of overall academic achievement, (ii) the racial achievement gap, and (iii) the college enrollment gap. To estimate the model I first develop a method for measuring AA practices in the entire college market without the need for student-level application data. Then I recover distributions over student heterogeneity using empirical auctions techniques which avoid imposition of distributional assumptions. These estimates facilitate a set of counterfactual experiments to compare the effects of the estimated US policy with alternatives not observed in the data: color-blind admissions and quotas. AA policies as implemented in the US significantly diminish the enrollment gap, but at the cost of lower academic effort on average, and particularly among talented minorities. A ranking between the colorblind rule and the US policy is ambiguous without knowledge of a social choice function assigning weights to criteria (i)-(iii). In contrast, a quota system produces a substantial improvement on all 3 criteria, relative to both alternatives. However, quotas are illegal in the US and cannot be implemented as such. Nevertheless, I propose a variation on the AA policy already in place that is outcome-equivalent to a quota, is simple to implement, and automatically adjusts according to the amount of asymmetry across demographic groups.