Racial quotas in higher education and pre-college academic performance: evidence from Brazil

ABSTRACT

The effects of affirmative action on the incentives to human capital accumulation are ambiguous from a theoretical perspective and the scarce empirical evidence on the matter provides mixed results. In this paper, we address this issue by investigating the impacts of Brazil’s Law of Quotas on the students’ performance in the college entrance exam, the ENEM. We provide causal evidence that the law fostered incentives to pre-college human capital accumulation, inducing students to attain higher ENEM scores. Moreover, the effects of the quotas were greater in quantitative-intensive subjects and these impacts increased throughout the first years after the law’s implementation.

KEYWORDS:

• Racial quotas
• higher education
• equality of opportunity
• academic performance
• difference-in-differences

JEL CODES:

• J15
• I24
• I28
• H52

Disclosure statement

No potential conflict of interest was reported by the author(s).

Conflicts of interest

The authors declare the nonexistence of any conflict of interest.

Availability of data and material

Not applicable (publicly available data).

Notes

1 Exame Nacional do Ensino Médio.

2 Census for 2010 data and PNAD (Pesquisa Nacional por Amostra de Domicílio) for 2019 data

3 Institutions maintained by federal and state levels of governments are forbidden by law to charge tuition fees, but municipal institutions are allowed and usually charge some tuition fees.

4 Federal and State universities have higher average scores in the Índice Geral de Cursos (IGC), a quality index developed by the Ministry of Education, and comprise most of the higher ranked institutions in the Ranking Universitário Folha (RUF), an annual evaluation of the HEIs in Brazil developed by the Folha de São Paulo newspaper. According to Binelli, Meghir, and Menezes-Filho (2008), there were on average 9 applicants per seat at public institutions in 2003, while this ratio was 1.5 in private institutions.

5 Although an individual’s racial information is self-stated, successful candidates that were accepted to Universities through the quotas are subject to have their profiles evaluated by the University’s Commission and/or by a Public Prosecutor based on the Statute of Racial Equality, mitigating any incentives for white students to declare themselves as non-white.

6 The probability of obtaining a correct answer is assessed according to its difficulty, the probability that a student could guess a correct answer, and its ability to discriminate against students.

7 Those who attended only a part of high school in a private institution also compose the control group, since they are not eligible for the quotas. For simplicity, we shall refer to this group as private high school (or simply private school) students

8 Rambachan and Roth (2019) introduce a parameter $M$ which governs the maximum amount by which the slope of the pre-treatment difference in trends can change between consecutive periods. See Rambachan and Roth (2019) for further details.

9 We have also estimated a model with dynamic treatment effects in which we allow all the pre-treatment betas to vary, but it does not induce any significant changes in the treatment effect coefficients

10 $\widetilde{\text{β}}$ and $\tilde{R}$ account for the treatment coefficient and R-squared in the augmented regression, $\dot{\text{β}}$ and $\dot{R}$ for the treatment coefficient and R-squared in the short regression, and $R_{max}$ is the R-squared of a hypothetical model that includes both observed and unobserved controls. See Oster (2019) for further details.

11 The estimated treatment effects are, however, different among levels of parental education at a 10% significance level

12 We abstract from 2010 since the survey was not conducted in that year due to the 2010 Census and from 2016, since from that year onwards the PNAD was replaced by its latest version, the PNAD Contínua

Additional information

Funding

The authors of this research have received financial support from the EDU2016-76414-R project of the Spanish Ministry of Economy and Competitiveness. This work was supported by Ministry of Economy and Competitiveness: [Grant Number EDU2016-76414-R].