Affirmative action policies in college admissions are often judged by how they change outcomes for the students who directly gain access—and for those who lose a place. But in large, centralized admissions systems, the ripple effects can be just as important: when one student shifts into a new program, the seat they leave behind is taken by someone else, who in turn may bump another applicant elsewhere. A new study by economists Nano Barahona, Cauê Dobbin, and Sebastián Otero examines these cascading reallocations and asks a simple question with a complicated answer: who ultimately benefits, and by how much?
The researchers develop an empirical approach designed to separate two layers of impact. First are the “marginal” effects—what happens to students who are just barely helped by the policy and those just barely displaced by it. Second are the overall effects once the admissions market re-sorts itself in response, potentially affecting many more students than those initially at the cutoff.
They apply this framework to a major Brazilian reform affecting federal higher education institutions. The policy required federal colleges and universities to reserve 50% of seats for applicants who are racially and socioeconomically disadvantaged. Because admission in these settings depends on score thresholds, the authors can compare applicants close to the relevant cutoffs to estimate causal impacts.
At the margin, the results suggest that the targeted students who are newly admitted because of affirmative action enroll in programs of higher quality than they otherwise would have. Those improvements translate into stronger degree completion outcomes: the graduation gains for marginally benefited targeted students are roughly twice as large as the graduation losses experienced by marginally displaced non-targeted students. The authors attribute that asymmetry to differences in outside options—disadvantaged students appear to have fewer and weaker alternative pathways absent the policy, so gaining access yields a larger payoff.
However, marginal comparisons do not capture the full system-wide reshuffling triggered by seat reservations. To assess the total impact, the team estimates a model of students’ degree choices and their potential outcomes across programs, allowing them to simulate how admissions equilibria adjust when the policy is in place.
That broader accounting preserves the basic pattern: among all students who are ultimately reallocated because of the policy, the average gains for benefited students remain close to twice the average losses for displaced students. Yet the equilibrium dynamics change the scale of displacement. Once knock-on moves are included, the policy ends up displacing about 30% more students than it directly benefits. As a consequence, the aggregate losses among non-targeted students offset about two-thirds of the graduation gains achieved by targeted students. The net effect across all students is therefore positive but modest.
The paper also acknowledges extensive feedback from a large group of economists and seminar participants, as well as financial support from multiple Stanford-affiliated centers and fellowship programs. The authors note that their conclusions represent their own views and not necessarily those of the National Bureau of Economic Research.
Why it matters
Affirmative action debates often focus on direct winners and losers, but this study shows that in centralized admissions systems, indirect reallocations can substantially reshape the policy’s overall impact. For policymakers, the key takeaway is that large gains for disadvantaged students can coexist with wide-ranging displacement that reduces net system-wide improvements. What to watch next is how alternative designs—such as different quota sizes or complementary capacity expansions—might preserve equity gains while limiting equilibrium spillovers.