GCER Featured Research Profile Series

The Real Impact of Affirmative Action: Nano Barahona, Cauê Dobbin, and Sebastián Otero Take it to the Data on Brazilian Education 

One of 2023’s biggest news stories was the Supreme Court’s historic and highly controversial decision to gut affirmative action in college admissions. This decision ended the ability of US higher education institutions to consider the race and ethnicity of an applicant as part of their admission decision. 

A recurring critique from opponents of affirmative action is that the practice creates a system where students of lower educational attainment can displace students who display greater academic ability, resulting in college campuses that aren’t as academically optimal as is possible. Economists actually have a term for this type of problem: an equity-efficiency tradeoff. This phrase describes the phenomenon by which prioritizing social equity can simultaneously compromise economic efficiency. The classic example is taxation: taxing the rich and redistributing it to the poor can decrease inequality, while also dulling incentives to work and thus shrinking the economy as a whole. More equal slices, but a smaller overall pie. 

In the context of affirmative action, this means more diversity on our college campuses, but lesser overall educational attainment as a whole. But, this is all in theory. What happens if you actually take this idea to the data? Well that’s exactly what economists Nano Barahona, Cauê Dobbin, and Sebastián Otero do in their paper “Affirmative Action In Centralized College Admissions Systems,” and the findings may surprise you.

So what does this paper look at? Well, it narrows in on affirmative action in a specific country – Brazil. Why Brazil? Well, Brazil has what the authors refer to as a highly “centralized” admissions system. This refers to the fact that for admission to the prestigious, flagship, and tuition-free federal universities all students must complete Brazil’s national university entrance exam (ENEM), and their score in this exam, along with their choice of degree program, is the sole determinant of admission. In fact, the admissions system produces daily cutoff scores during the application process, which are the scores needed to gain admission to a given program, based on which students can alter their application. 

This highly centralized admission system contrasts the black box-style admissions we have in the US, where admission is based on a combination of factors, and the specific reason for a given student’s acceptance is anything but clear. The Brazil system gives the researchers direct insight into why a given student gets in, literally presenting the exact score a student needed to beat in order to be admitted. The centralized system also gives the economists access to large, reliable datasets on students’ degree program preferences, test scores, and socioeconomic background. Combining this information with data from Brazil’s census and employer-employee data results in an accurate, multifaceted educational database from which the economists can construct varied, powerful, and robust statistical models. 

Finally, there’s the actual affirmative action policy of Brazil. In August of 2012, the Brazilian government passed a law requiring federal institutions of higher education to reserve half of their seats for students from public high schools, a group which represents over eighty percent of all high school graduates. The bill passed with overwhelming support in Congress, and was structured to further break down the reserved slots into reserved slots for poor applicants and applicants of black, brown, and indigenous backgrounds. See the below chart for a clear visualization of this policy: 

On top of the comprehensive data available, this affirmative action policy neatly bucketed applicants into different categories – ideal for the regression model that the economists used to draw their insights. Not only can the researchers create a model that looks at the effect of the policy as a whole, as other economists have done previously, but can actually make comparisons between students benefited and displaced by the policy. It’s worth noting, that while one of these categories is race-based, Brazil’s affirmative action policy is primarily socioeconomic, as the overarching 50-50 slot split is done on by public versus private high school attendance. This is an important distinction from the US’s previous system, especially considering the paper focuses mainly on this socioeconomic distinction, and not the racial one. 

So what do the researchers do with all this data? By leveraging the Census and employer-employee data, the economists complement their analysis on the policy with a model projecting students’ income in the years after graduation based on their degree program, allowing insight into how the policy actually helped or hurt the students affected by it. To avoid a statistical bias problem by which students with higher test scores may be earning more than students’ with lower scores simply because of latent intellectual ability, the authors use a clever measure of grader strictness. Since graders are assigned randomly, and some will inevitably grade the more subjective essay portion harsher than others, the economists are able to compare students with higher and lower test scores that weren’t a result of their actual ability. 

After surmounting that potential statistical bias, the paper is able to finally estimate the impact of the affirmative action policy, seen above. The model returns an average 1.02% increase 

in projected earnings of students who could access the reserved seats, and 1.42% decrease in the corresponding non-targeted group. These trends are what we would expect – an increase in the targeted population that comes at the expense of the non-targeted population. What is most interesting though, is that when changes across both groups are summed, the average student experiences only a 0.014% decrease in projected income – essentially zero! This implies that the change in income benefit to the targeted population is an almost one-to-one transfer from the non-target population, i.e. this affirmative action policy comprises an increase to social equity with practically no backlash to economic efficiency.

This flies in the face of what traditional economic theory says under the equity-efficiency framework: the increase in social equity should bear with it some overall economic cost. So why is this happening? The authors’ present two central explanations. First, there is substantial academic literature focused on the idea that standardized testing is biased against students of a lower socioeconomic background. If this is the case, prioritizing lower socioeconomic backgrounds in the admissions process can actually correct for the economic inefficiency resulting from biased standardized testing, which would counterbalance the loss in efficiency from displacing non-targeted students in the affirmative action policy. 

The second explanation is one that the authors land on as most convincing, and even find evidence for in the data. The idea is that the non-public school students who aren’t targeted by the affirmative action policy tend to be wealthier, and can afford the more expensive, private alternatives to the federal universities. Understanding this requires some more detail on Brazil’s higher education system. In Brazil, the public universities are prestigious, akin to many of the US’s flagship universities. This prestige, along with their free price, makes them very hard to get into. There are also private universities that are considered very good, but since they’re expensive, they’re much more easy to be admitted to due to less competition from poorer applicants. 

This is important to understand, because it’s essential to how Brazil’s affirmative action system avoids an efficiency loss. When the affirmative action policy displaces some of the richer, non-targeted students, many are able to just go get educated at comparable alternatives provided by the private sector. Therefore, the efficiency loss is much smaller, as many of the displaced students are still getting quality educations. The targeted students, on the other hand, are much more likely to not be able to afford private education, so the ones who get into the federal universities through the affirmative action policy are benefitted much more than the non-targeted students are by losing out on the federal schools. 

The paper concludes by restating its central findings, but goes further to offer future avenues for research. The authors suggest examining the interaction between affirmative action policy and the presence of differences in student loan availability, as targeted students tend to have access to student loans more often. Additionally, the authors also note the potential economic gains that come from a more diverse college campus, as the peer effects of a more diverse learning environment can itself influence social behaviors as well as academic and professional outcomes for students. If you’re interested in reading the full article, you can find it on Professor Cauê’s website, linked here: https://cauedobbin.com/, alongside his other papers. Also, with some bilingual skills or help from Google Translate, you can read about its media coverage in these various outlets: Valor Econômico, Nexo, AgênciaBrasil, Exame, R7, OPovo.

Previously Featured Research Profiles:

Fall 2024 Featured Research Profile: Climbing the Economic Ladder, and Counting the Rungs: Economists Debraj Ray & Garance Genicot Explore a New Method to Measure Upward Mobility.

Fall 2022 Featured Research Profile: GU economists hold serve:  New research by a team of GCER Fellows and grad students show how Federer, Djokovic, and Nadal can up their game. 

Spring 2021 Featured Research Profile: Finance-Thy-Neighbor? Georgetown Economist Margit Reischer analyzes the macroeconomic implications of trade credit in supply chains.

Fall 2020 Featured Research Profile: Impactful differences and differential impacts: Georgetown Economist Louise Laage shows how to recover average effects in panels with unobserved and unexplained sources of heterogeneity.

Spring/Summer 2020 Featured Research Profile: Spouse and house? GCER Fellow Minsu Chang connects the dots between the two.

Spring/Summer 2019 Featured Research Profile: GCER Fellow Alexandre Poirier’s and Matthew Masten’s study of treatment effects carves out the circumstances under which circumstances carve out economic behavior. 

Fall 2018 Featured Research Profile: “You’re [not] fired!”–GCER Fellow Toshi Mukoyama and Sophie Osotimehin explore the dynamic linkages between employment protection regulations and firms’ innovation decisions.

Winter/Spring 2017-2018 Featured Research Profile: Child Care in Reverse: GCER Fellow Ami Ko explores the subtle effects of informal health care on the long-term care insurance market.

Fall/Winter 2016-2017 Featured Research Profile: Markets with Search and Matching Frictions: Georgetown economists James Albrecht and Susan Vroman discuss directed search in the housing market.

Fall/Winter 2015-2016 Featured Research Profile: How (not) to run a bank: Georgetown economist Martin Ravallion examines World Bank successes and failures.

Spring/Summer 2015 Featured Research Profile: Leaning in,… sort of: Georgetown economist Mary Ann Bronson explores reasons why men and women make different post-secondary educational investments.

Fall 2014 Featured Research Profile: Carbon emissions make the global economy tipsy… Harrison and Lagunoff study a “business-as-usual” scenario in a tipping model.

Spring 2014 Featured Research Profile: Collateral Damage to Standard Economic Theory… GCER Fellow Dan Cao shows how incorrect beliefs can fuel a crisis.

Fall 2013 Featured Research Profile: Oh what a tangled web we weave… Anderson and Smith explore the dilemmas of deception.

Winter/Spring 2013 Featured Research Profile: Unintended Consequences in the Struggle for Equal Rights: Anderson and Genicot explore the surprising relationship between suicides and female property rights in India.

Fall 2012 Featured Research Profile: Gale and O’Brien sing the blues over Use-or-Lose!

Spring/Summer 2012 Featured Research Profile:”What a piece of work is a man! How noble in Reason! How infinite in faculties!” … But how much is he worth? Huggett and Kaplan provide an answer.

Winter 2012 Featured Research Profile: Happiness is in the air! Levinson uses happiness surveys to put a dollar value on air quality.

Fall 2011 Featured Research Profile: Junior, the Risky Investment, Grandma, the Insurance Contract, and other bedtime stories as told by Gete and Porchia.

Spring/Summer 2011 Featured Research Profile: Ludema, Mayda, and Mishra show that when firms talk, governments listen.

Winter 2011 Featured Research Profile: Bachmann and Bai examine the effects of wealth bias in the policy process.