1 Introduction

Tertiary education attendance in Brazil varies remarkably among different population groups (^{PEDROSA et al., 2007}). Race, household income and parent´s educational attainment are among the most decisive factors that impact the student’s probability of attending a tertiary education institution (^{SILVA; HASENBALG, 2002}), as students of higher socioeconomic status (SES) have better performances in the selection process at these institutions. Consequently, students from the best private elementary and secondary schools tend to be disproportionately enrolled in the best public universities (^{TELLES; PAIXÃO, 2013}).

In order to decrease inequalities and to increase the competitiveness of Black/Pardo/Indigenous students and those from low-income families in public universities, many public institutions in Brazil implemented affirmative action policies in the last decades (^{FRANCIS; TANNURI-PIANTO, 2012b}; ^{PEDROSA et al, 2007}; ^{TELLES; PAIXÃO, 2013}). In particular, the Federal University of Minas Gerais (UFMG) implemented a bonus policy in the entrance exam of 2009. This policy increased in 10% the grades obtained in this exam for individuals who had attended public schools in the last seven years of elementary and secondary education. In addition, an extra 5% bonus was given to those who considered themselves Blacks/Pardos. This policy with minor changes was valid until the entrance exam of 2012. In 2012, it was approved the national federal law of quotas to be implemented in all federal higher education institutions in four years (TELLES; PAIXÃO, 2013). Beginning in the entrance exam of 2013, at least 12.5% of the students in each course at UFMG had to originate from disadvantaged schooling and/or race backgrounds. This number increased to 25%, 37.5% and 50% respectively in the years of 2014, 2015 and 2016, and remained in this last level since then.

Besides, other policies were also implemented by UFMG and they may have had an effect on the determinants of the performance in the selection process at higher education institutions. Mostly due to the Restructuring and Expansion of Federal Universities - REUNI, policy, there was a remarkable increase in the number of slots in federal universities in Brazil. In particular for UFMG, the number of slots increased between 2009 and 2011 from 4.6 thousands to 6.6 thousands annually (^{ARANHA et al., 2012}; ^{LIMA; MACHADO, 2016}). In the period, 27 new courses were created. Other 23 courses that already existed increased the number of slots.

Moreover, there were recent changes in the entrance exam. Until recently, most universities had their own exams, which differenced from other institutions. In the last years, the ENEM exam began to be used as part or as the unique exam of tertiary education student’s selection (^{LIMA; MACHADO, 2016}). In UFMG, there were two main changes in the entrance exam since 2009. Until 2010, there was an UFMG´s exam in two stages. From 2011 and 2013, the ENEM was used as the first stage, while there was an UFMG´s exam for the second stage. Since 2014, the ENEM and the Unified Selection System (SISU) have been used as the main methods of student selection (LIMA; MACHADO, 2016; ^{NOGUEIRA et al., 2017}; ^{SILVEIRA et al., 2015}). The federal government claimed three positive points using this last selection process: the increase in student’s mobility; the gain in efficiency of the process; and the increase in the proportion of minorities in the university, this last point associated with the quota policy. Nonetheless, ^{Nogueira et al. (2017}) and ^{Silveira et al. (2015}) concluded that none of the expectations was reasonably fulfilled.

All these policy changes might have changed the UFMG´s applicant’s pool, how students are selected, the student body and, in particular, the determinants of the performance in the selection process. Concerning this last point, many studies addressed the determinants of academic performance in tertiary education institutions (^{FRANCIS; TANNURI-PIANTO, 2012a}, b; ^{GOLGHER et al., 2015}; ^{PEDROSA et al., 2007}; ^{TELLES; PAIXÃO, 2013}).

As opposed from those studies, the main objective of this paper is to analyze the determinants of the performance in the entrance exam of students who applied at UFMG, being selected or not by this university, between the years of 2009 and 2013. In this period, there were the above-mentioned changes in the number of slots, in the affirmative action policy and, most importantly, in the type of exam used to select students. The variables of main interest are race, household income and the type of secondary school the student attended, as they are directly related to the attendance of minorities in public universities in Brazil. However, it should be clear that the paper does not discuss how the evolution of minorities´ attendance at UFMG was, but whether the determinants of performance in the selection process varied for these variables and due to these policy changes. Nonetheless, changes in these determinants may have an influence on how minorities are absorbed at UFMG. The study concerning the evolution of minorities’ attendance at UFMG is better addressed using the students actually selected at the institution, nonetheless, this topic is not addressed her as it is beyond the scope of this paper. To the best of my knowledge, this is the first analysis using econometric models that deals with the issues proposed by this paper.

The paper uses official records of UFMG developed by the Permanent Commission of the Vestibular - COPEVE and by the Department of Academic Registry - DRCA. The databases were kindly conceded to research purposes and preserve the anonymity of the students.

Besides this introduction, the paper is further divided into four sections. Section two presents the literature review, which is the background of the paper. Section three describes the methodology used in the statistical analysis. Section four depicts the empirical results, and last section concludes the paper.

2 Literature review

Many factors are associated with academic performance in different schooling levels. Among these determinants there are the individual´s attributes (sex, race, age, etc.), household´s characteristics (parent’s schooling level, income, number of siblings, etc.) and school factors (infrastructure, administration, teachers, etc.), (^{BARROS et al., 2001}).

In particular for the elementary and secondary levels in Brazil, ^{Araújo and Siqueira (2010}) analyzed the determinants of schooling performance in Mathematics for students in the fourth year. They observed that students whose mother had higher levels of formal education, who lived with their father, in higher income households and who had a computer at home had higher performance. Moreover, individual attributes were also related to performance: girls, blacks, those who worked and those who attended public schools had lower levels of performance. ^{Machado et al. (2008}) studied the schooling performance in Mathematics at the elementary and secondary levels. They observed that white individuals and those who lived in a household with more books had higher performances. Similar to the previous study, students who did not work and had a mother with higher levels of education also had higher performance. ^{Soares and Alvez (2013}) studied the determinants of schooling performance for students of the elementary level in reading and Mathematics. They observed that individual and school variables were significantly associated with performance. Females had higher performances in reading, but lower in mathematics. As above, blacks and poor individuals had lower performances. They compared students from state and municipal schools and the second had higher performances.

Other studies described the determinants of academic performance in the tertiary level. ^{Bai and Chi (2011}) observed that females had a better GPA than males and that minorities had similar performances than non-minorities. ^{Fryer Jr. et al. (2008}) also analyzed the determinants of college GPA and observed that SAT scores, high school ranks, parental education and neighborhood’s racial demographics (fewer blacks) were positively correlated with academic performance.

In particular for Brazilian institutions, many authors addressed factors that were associated with academic performance. ^{Francis and Tannuri-Pianto (2012b}) observed that males, pardos and indigenous had lower performances. In particular for UFMG, ^{Golgher et al. (2015}) verified that students that had studied in municipal, federal or private secondary school, that hadn’t attended a specific course preparing for the exam of tertiary education institutions (pré-vestibular), who did not work or who worked up to twenty hours per week, from a higher income household and who had computer at home had higher performances.

All these cited studies addressed the determinants of academic performance at different schooling levels. However, although related, the focus of this paper is the performance in the entrance exam of tertiary education institutions. Concerning this topic, ^{Behrendt et al. (1986}) observed that students from private secondary schools, with less siblings and in regions with greater proportion of educated individuals had higher performance in the SAT exam. ^{Rothstein (2004}) observed that blacks, Asians, Hispanics and women had lower performances in the SAT.

In particular for Brazil, ^{Francis and Tannuri-Pianto (2012a}) analyzed the performance in the entrance exam at UnB in 2004. They regressed vestibular score on race, gender, and a number of family-level socioeconomic measures. They observed that blacks, females, individuals with lower SES, student’s whose mother had a lower level of formal education and who had attended public schools had lower overall scores.

This paper analyses the determinants of the performance in the selection process at UFMG between 2009 and 2013. Hence, in some aspects, it is similar to the last cited study, however, it addresses these determinants rather differently. Based on the main changes in the selection process policies, there are some possibilities of analysis regarding the determinants of performance in the entrance exam of UFMG. First, did the determinants alter due to the change of the exam, as the UFMG´s own exam was replaced by ENEM in the first stage of selection? Second, did the determinants alter because of the increase in slots between 2009 and 2011? Third, which were the changes due to the replacement of the bonus policy by the quota policy? These are the main questions being empirically addressed by this paper. In addition, the paper also associates individual´s attributes, household´s characteristics and school factors with the academic performance at the selection process at UFMG, focusing in possible changes in the analyzed period for variables related to the minorities’ attendance in higher education institutions.

3 Methodology

4 Results

5 Conclusion

Tertiary education attendance in Brazil varies remarkably among different population groups. In order to decrease inequalities, many public institutions in Brazil implemented affirmative action policies in the last decades (^{PEDROSA et al., 2007}; ^{TELLES; PAIXÃO, 2013}). In particular, a bonus policy was implemented in UFMG in the entrance exam of 2009, which was replaced by the law of quotas in 2012. Besides, other policies were also recently implemented in UFMG and other federal universities, such as the Reuni policy and the use of the ENEM exam as part or as the main exam for tertiary education students selection (^{LIMA; MACHADO, 2016}; ^{NOGUEIRA et al., 2017}; ^{SILVEIRA et al., 2015}).

All these policy changes might have changed qualitatively and quantitatively the UFMG´s applicant pool and the factors that may affect the student´s performance in the selection process. The main objective of the paper was to analyze the determinants of the performance in the entrance exam of UFMG between the years of 2009 and 2013, period when these changes in the selection process took place.

Many findings of the paper corroborated the results of other studies, such as males, singles, individuals who did not work and who had already graduated had higher performances. However, the focus of the paper were the differences in these determinants due to the above-mentioned policies and also related to the minorities’ attendance at higher education institutions.

The bonus, quota and Reuni policies were effective to promote an increase in the proportion of minorities attending Brazilian Federal universities. The presence of more minority students on campus may cause positive externalities to the minority students themselves or to their peers, in particular through mentoring or role model effects on future minority students (^{HOLZER; NEUMARK, 2006}).

Nevertheless, the point addressed here is whether the performance of minorities varied with the implementation of these policies and due to the introduction of the ENEM as a tool for student’s selection. Affirmative action and Reuni policies did not affect remarkably the performance of minorities at the UFMG´s selection process. However, the use of ENEM as the first stage in this process relatively decreased the performance of minorities. That is, due to this last change, minorities became less competitive vis-à-vis non-minorities. Some factors that can be related to this finding are described below.

^{Moeto et al. (2017}) discuss that the entrance exams of the selection process of higher education institutions are the main guidance to define the content of secondary level disciplines. The ENEM was created in 1998 to evaluate abilities and competences of students ending their secondary education. However, the exam also began to be used to select students at the tertiary level. Thus, the use of ENEM as a tool to select students for tertiary level institutions implicated that the secondary level adapted to its demands. In this vein, ^{Santos (2011}) argue that the implementation of the ENEM, since its beginning, had as one of its objective to be an instrument to promote changes in the secondary level, incorporating new abilities and competences to students of this educational level. Hence, consequently, secondary schools, in particular private ones, adapted their curriculum and changed their forms of teaching to increase their students’ performance. Moreover, many extra courses were created to ameliorate the student´s preparation to the ENEM (^{SANTOS, 2011}; SANTOS; CORTELAZZO, 2013).

Associating these points to our empirical findings, private schools might have adapted more effectively to changes in the selection process at UFMG, increasing the gap between minorities and non-minorities. Moreover, minorities may not have the resources to further pursue extra preparation for the ENEM exam.

In addition, the ENEM exam address basic competences, articulating conceptual contents of different disciplines and contents from everyday life. These competences are: language expertise, phenomena comprehension, the resolution of applied problems, and the facility to build arguments and proposals intervening in real problems (^{SANTOS, 2011}; SANTOS; CORTELAZZO, 2013). Besides, the ENEM exam is considered a more interesting exam than traditional selection process exams at the tertiary level, with less emphasis to memorized aspects of learning, and a focus on competences (SANTOS, 2011; SANTOS; CORTELAZZO, 2013). ^{Moeto et al. (2017}) argue that there are more conceptual and contextualized questions in the ENEM when compared to the traditional selection process. For instance, ^{Santos and Cortelazzo (2013}) argue that the ENEM emphasizes reading comprehension instead of explicit grammar questions. However, although these points may represent an advance, the questions still require memorization and contribute only slightly to the development of abilities that require interdisciplinarity and scientific thinking (^{MOETO et al., 2017}). Nonetheless, it incentives a more active attitude of the students toward leaning, giving less weight to the passive mindset of a simple receptor of content (SANTOS; CORTELAZZO, 2013).

These topics described in the last paragraph can be related to our empirical findings. Individuals with a more precarious educational background may have relatively more difficulty to have an active attitude towards learning in order to acquire abilities and competences to deal with conceptual and problem solving questions. Moreover, it might be more difficult to access different concepts of different disciplines to build arguments and proposals intervening in real problems for someone with lack of basic knowledge in most disciplines of secondary school.

^{Barros (2014}) argue that some researchers advocate that the ENEM promotes the democratization of tertiary education. However, the author adds that reality continues to be highly unequal concerning educational opportunities and that not much has changed with the implementation of ENEM as a tool for students selection. Following our findings, the use of ENEM actually increased inequality in attendance at higher education institution.

According to ^{Cunha and Heckman (2007}), ability gaps across children with different backgrounds open up at early ages. Thus, they argue that early childhood investments should be distinguished from later investments, and that interventions targeting adolescents may be less effective. That is, changes in tertiary education, as described in this paper, are a second best solution for the education problem. For instance, SES levels and the type of secondary school attended were particular decisive and stable to determine performance, hence future labor market outcomes and competitiveness further in life.