Principal investigator: Eduarda Miller de Figueiredo
Authors: François Gerard, Lorenzo Lagos, Edson Severnini and David Card
Location of the Intervention: Southeast Region of Brazil
Sample Size:
Sector: Labor market
Primary Variable of Interest: Log of hourly wage
Type of Intervention: Salary
Methodology: Other
Summary
The political debate regarding racial differences in education levels between whites and non-whites is intense in Latin American countries, noting that non-whites remain underrepresented in high-paying sectors. Analyzing data from the PNAD and RAIS surveys for the Southeast region of Brazil, this article estimates wage differences between races and genders, considering individual and establishment effects. The results demonstrate a wage disparity between races.
- Policy Problem
Wage differentiation between whites and non-whites occurs in many countries around the world. Randomized studies show that employer callback rates are lower for minority job applicants, implying that some companies set a higher standard for hiring non-white candidates or avoid hiring minorities altogether. This also suggests that employers assign non-white workers to lower-paying occupations, accounting for some of the racial wage disparities within companies (Penner, 2008; Giuliano, Leonard, and Levine, 2009, 2011). However, it is unclear how much these standards contribute to the average wage disparities between whites and non-whites (Lang and Lehmann, 2012).
Understanding this differentiated hiring practice is particularly relevant in Brazil, a country where almost half of all workers identify as non-white.
- Implementation and Evaluation Context
In Brazil, approximately 50% of men and women of working age are white, 42% consider themselves mixed ("pardo"), and 8% are black. In the Southeast region, 57% are white and 33% are mixed. On average, 45% of Brazilian men employed in the private sector during the study period had completed high school, with a higher rate for whites (53%) than for non-whites (38%).
Looking at average hourly wage log statistics, several factors stand out: (i) white workers of both sexes earn about 30% to 35% more than non-white workers; (ii) wage levels are more than ten log points higher in the Southeast than in the country as a whole, but wage disparities remain similar; and (iii) average wages for brown and black workers are only a few percentage points apart.
The informal sector in Brazil is large, with only 80% of private sector employees claiming to possess a valid work permit, which is the indication of formal employment in the country.
- Policy/Program Details
The study's main analysis uses administrative data for workers in the formal sector in the southeastern region of Brazil, which includes Espírito Santo, Minas Gerais, Rio de Janeiro, and São Paulo. Annual information on the labor market for formal and informal workers was collected through the National Household Sample Survey (PNAD). Data from men and women aged 25 to 54 years, with at least one year of work experience and employed in the private sector, were used.
To estimate the impacts of company policies on racial wage disparities, the authors used the Annual Social Information Report (RAIS), which provides universal coverage of formal employment data in the country (Ministry of Labor, 2015).
- Method
To assess the issue of informality, simple linear probability models for the incidence of formality were estimated. The models suggest null effects with precision of non-white race on the probability of formality. Furthermore, the size of unexplained Whitenon-White wage differences was compared based on samples that include all private sector employees and only those in the formal sector.
The results were estimated based on the AKM model (Abowd, Kramarz, and Margolis, 1999), with the dependent variable being the log of the hourly wage paid to the worker. i in the race-gender group g in December of the year tA fixed effect per person was added, as well as a set of controls that vary over time. Results were also estimated using instrumental variables, employing the estimated wage premium for workers of the same racial group but opposite gender as an instrument for the group's wage pool in each establishment.
According to the authors, it is important to emphasize that the person effects estimated in an AKM model incorporate any unobserved components of human capital, such as differences in school quality or choice of higher education. Therefore, differences in school or graduation quality between whites and non-whites were likely reflected in ability-based measures, but did not influence residual ranking measures.
- Main results
Estimates suggest that personal effects account for 51% to 62% of wage variation, while establishment effects account for 20% to 23%. Worker and firm effects are positively correlated within each gender-race group, which accounts for 8% to 11% of overall wage variation for non-whites and 18% for whites. Together, differences in wages paid by different establishments and the strong pattern of matching between workers and establishments explain about 30% to 40% of wage variation for all gender and race groups. These estimates are similar to those reported by Card, Heining, and Kline (2013) for the analysis of Germany and by Lavetti and Schmutte (2016) for Brazil.
By performing the racial wage decomposition into personal and establishment effects, the authors find a wage gap of 15,5 percentage points for men and 23,8 percentage points for women, with most of this wage gap attributed to differences in personal effects. Therefore, personal effects and time-varying covariates account for 79% and 75% of the overall wage gap between whites and non-whites, between men and women, respectively.
For education, estimates were made for three education categories: workers with less than a high school education, high school graduates who did not complete college, and college graduates. The results demonstrate that the wage disparity between whites and non-whites increases sharply in all three categories for both men and women, ranging from about 5 percentage points for workers without a high school education to 19-22 percentage points for those with a college degree. The results also demonstrate that the increase in individual effects and establishment effects, for higher education levels, is more pronounced for whites than for non-whites for both sexes. And that, together with the differences in establishment effects, the overall wage gap by race increases for high school and college graduates.
Finally, estimates show that workers with higher transferable skills are more likely to work in establishments that pay higher premiums, and that whites tend to have higher transferable skills than non-whites, earning higher wages for a given skill level.
- Lessons in Public Policy
The results of this article refer to ongoing political debates in Latin American countries where racial disparities in education levels persist and non-whites remain underrepresented in high-paying sectors. It has been shown that non-whites are less likely to be employed in high-value jobs, even in the absence of any discriminatory employment practices.
References
GERARD, François et al. Assortative matching or exclusionary hiring? the impact of employment and pay policies on racial wage differences in brazil. American Economic Review, v. 111, no. 10, p. 3418-57, 2021.
