Affirmative Action in Brazil

1
Affirmative Action in Brazil

• Andrew Francis
• Emory University
• Maria Tannuri-Pianto
• University of Brasilia

2
Introduction

• Large black and mixed-race population in Brazil
• High rate of ethnic intermarriage
• Inequalities in many aspects of life including
  income, education, health, etc.
• Race and socioeconomic status correlated
• Policy makers have developed initiatives aimed at
  reducing inequalities

3
Introduction

• Recently, some universities have implemented an
  affirmative action system involving racial quotas
• University of Brasilia
• Unique opportunity to examine a number of
  questions about the effects of affirmative action
  in higher education
• Before/after the implementation of the policy
• Racial continuum

4
Summary of Talk

• Background and policy
• Survey and data collection
• Preliminary results
• Racial socioeconomic change at Unb
• Academic performance of quota students
• Incentive effects on effort and racial identity
• Conclusions

5
Background and Policy

• College admissions highly competitive
• New entering cohort every semester
• Vestibular system
• Most students attend college in their region of
  residence
• Unb one of the best in Brazil
• Established quota system in 2nd semester of 2004

6
Background and Policy

• 20 slots reserved for negros (black racial
  identity)
• includes students who are pardo (brown skin
  tone) and preto (black or dark skin tone)
• Quota system lowers the minimum score necessary
  for acceptance
• To apply under quota system
• Declare oneself negro (black racial identity)
• Photo panel

7
Survey and Data Collection

• Population of interest
• Cohorts entering Unb two semesters before
  (2-2003, 1-2004), three semesters after (2-2004,
  1-2005, 2-2005) implementation of quota system
• Both quota and non-quota students
• All races, all majors
• Three data sources
• Admissions data
• University grades
• Our own questionnaire, University Education
  Survey

8
Survey and Data Collection

• Admissions data
• Vestibular exam scores for all accepted and not
  accepted students (150,000 entries)
• Multiple records per person across time
• Short 18-question socioeconomic survey (QSC) at
  registration for vestibular exam
• However, no race questions until 2004
• Incomplete sample 2005
• University grades
• Official grades for all students

9
Survey and Data Collection

• University Education Survey
• Questionnaire
• Demography and family background
• Pre-university education
• University admissions
• University education
• Employment
• Future and expectations
• Race and society
• Photos
• Affirmative action, race only at end to avoid
  priming
• We never mention true focus of survey in any
  contacts with respondents

10
Survey and Data Collection

• Respondents contacted by email and phone
• Face-to-face interview with interviewer
• Full questionnaire
• 30 minutes
• About 1,000 participants
• Respondents contacted by email
• Surveymonkey (internet)
• Shorter questionnaire with essential questions
• 10-15 minutes
• About 2,000 participants

11
Sample Quality Checks

• We are able to compare our sample with the
  population on a number of characteristics
• Biases (in order of importance)
• Econ majors
• Gender (5 pp difference between respondents and
  non-respondents)
• Average GPA (difference is about 10 of one
  standard-deviation)
• No significant differences in
• Race, family income, most other majors, score on
  vestibular, etc.

12
Preliminary Results
13
Race Poverty

• Big socioeconomic differences among whites,
  browns, and blacks (pretos)
• Blacks (pretos) are by far poorer than the rest
• Using race to target poorer families is certainly
  reasonable, all else equal

14
Benefits of Applying Under Quota System

• The difference in cutoffs for non-quota and quota
  students is enormous, especially at the
  beginning most quota students wouldnt have
  gotten in otherwise
• Initially, a lot of candidates with better scores
  than selected quota students were passed up
• Later, the difference is cutoffs narrows
• Later, most quota students would have gotten in
  anyways

15
Racial Change at Unb

• Did the quota system increase the percentage of
  darker students at Unb?
• Self-reported race
• white decreased slightly
• brown and black (preto) increased slightly
• Analysis of photos
• Panel rated photos in terms of skin tone (light
  to dark)
• Scores were standardized by rater and averaged
  across raters

16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
Socioeconomic Change at Unb

• Did the quota system increase the percentage of
  poorer students at Unb?
• There are small differences in fathers
  education, having a car, and public school
  attendance such that somewhat poorer students
  are now attending Unb
• There are moderate differences in family
  structure and having domestic workers
• Students are poorer but the changes are modest

22
Performance at Unb

• Quota students tend to have slightly lower grades
• This difference decreases by half when
  controlling for their score on the vestibular
• Remaining difference is significant but small
  relative to the gender gap

23
Incentive Effects on Effort

• Darker students appear to have studied more and
  gotten better grades in high school
• Darker students were more likely to have taken a
  college-prep course (cursinho)
• No change in the reported number of hours studied
  for the vestibular, however
• White students applied to a greater number of
  universities
• No change in study effort at Unb

24
Incentive Effects on Racial Identity

• It is possible to look at multiple responses on
  the QSC for the same individuals
• Applicants who said they were white in semester
  t were more likely to say they were brown in
  semester t1
• To a lesser extent, applicants who said they were
  brown in semester t were more likely to say
  they were black (preto) in semester t1
• PAS vs. vestibular on vestibular QSC, more
  likely to say they are brown

25
Incentive Effects on Racial Identity

• Whites rarely considered themselves negro (black
  racial identity), while pretos almost always did
• Most browns didnt consider themselves negro
  prior to the quota system, but a very high
  percentage of browns did after quotas
• The percentage of non-quota browns who considered
  themselves negro also went up with time
  (strategic behavior or social externality)

26
Incentive Effects on Racial Identity

• On one hand, about half of browns who said they
  were negro on the QSC said they were not on our
  survey
• Evidence of misrepresentation, cheating
• On the other, many browns reaffirmed that they
  were negro on our survey
• In regressions, being a quota student is positive
  and significantly related to identifying as
  negro, while controlling for gender, race, skin
  tone, family background variables, semester, and
  major

27
Conclusions

• Using a race-based affirmative action program in
  Brazil has some justification, since race and
  socioeconomic status are related
• At Unb, the quota system may have raised the
  percentage of darker and poorer students, but
  effects modest
• Incentive effects of raising pre-university
  effort of darker students
• But there also were large effects on racial
  identification
• Some students misrepresented their race, while
  others may have been more likely to identify as
  negro thus polarizing racial identity
• Perhaps targeting the darkest students or using
  family income or public school attendance could
  be more effective