Abstract
This paper studies potential explanations of the declining wage inequality in Brazil such as changes in demographic/skill composition, wage structure, occupations/sectors and minimum wage. I perform a wage inequality decomposition to quantify composition and price effects and use a CES production function to estimate the effects of the skill supply on relative wages. I find that the fall in upper-tail inequality is driven by changes in the returns to education and experience, while that in lower-tail inequality is also given by those to minimum wage and female workers. These patterns are consistent with the decline in relative wages between skill groups which are given by the increase in both the supply of skills and the real minimum wage.
Keywords
wage inequality; skill premium; minimum wage
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Thumbnail
Source: Socio-Economic Database for Latin America and the Caribbean (CEDLAS and The World Bank). Version: May 2018. Gini coefficient for the distribution of household per capita income excluding zero income. The average Gini coefficient for Latin America is an arithmetic average of 18 Latin American countries (Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay and Venezuela) from 1989 to 2015. Data became available only after 1989 for most countries and there are missing observations in specific countries and years, thus around 25 percent of the observations were obtained by interpolation.
Source: PNAD data from 1981 to 2015. Log changes in hourly wages at the 10th, 50th and 90th percentiles from the“wage sample” are normalized to zero in 1981.
Source: PNAD data for 1981, 1990, 2001 and 2015. Relative changes in log hourly wages from the“sample”between two years. The change in the log hourly wage at the median is normalized to zero for each period.
Source: PNAD data from 1981 to 2015. Log hourly wages for full-time salary workers are regressed by gender in each year on four education dummy variables (less than 11, 11, 12 to 14, and 15 or more years of schooling), a quartic in experience, two race dummies (black/indigenous and others non-white/non-mix-race) and the corresponding interactions between education and experience. I calculate a set of fixed weights equal to the participation in employment of each one of the 490 gender-education-experience groups. The composition-adjusted log wage for each educational group is the weighted average of the predicted log wage of white/mix-race workers evaluated at each demographic group. The skill premiums are the weighted average of the composition-adjusted log wages between the corresponding educational categories. Tertiary and non-tertiary educated workers are aggregate categories. The former comprises workers with at least some tertiary education and the latter comprises illiterate, primary and secondary educated workers.
Source: PNAD data from 1981 to 2015. Employment participation of the 490 gender-education-experience groups from the“supply sample”is weighted by using a set of fixed weights equal to the mean wage in each cell normalized to the wage of male workers with secondary education and 10 years of potential experience (base group) over the sample period. Efficiency units of labour supply are given by the weighted average of labour supplies in each demographic group. Relative supply of tertiary/non-tertiary and secondary/primary educated workers is the logarithm of the ratio between efficiency units of labour supply of the corresponding skill group.
Source: PNAD data from 1981 to 2015. Composition-adjusted log hourly wages for full-time workers are the weighted average of the predicted log wage in each of the 490 gender-education-experience groups. Each series is normalized at zero in 1981. See 
Source: PNAD data from 1981 to 2015. Efficiency units of labour supply are the weighted average of the employment participation of 490 gender-education-experience groups. See 
Source: PNAD data from 1981 to 2015. Sample comprises salary/wage workers and self-employed, aged 18-65 years old. Workers employed in the military are excluded from the sample. PNAD provides information on individual occupations and eight aggregate categories (Managers, Professionals, Technicians, Office and administration, Sales, Production and repair, Personal services and Agricultural occupations) from 2002 to 2015. This categorisation is not consistent with that in prior years. I sort around 380 occupations from 1981 to 2001 into the previous eight occupational categories by matching the definitions of individual occupations between the two periods. Occupations that are not assigned to any category are excluded from the sample (around 7 percent). 
Source: PNAD data from 1981 to 2015. Log hourly wages of salary/wage workers and self-employed excluding those in the military, aged 18–65 years old. Each series is normalized at zero in 1981. See 
Source: PNAD and MTE data from 1981 to 2015. Sample comprises salary/wage workers and self-employed, aged 18–65 years old. The ratio is the proportion of minimum wage workers in each demographic/occupational group to the proportion of minimum wage workers in the workforce. Minimum wage workers are defined as those who earn +/5 percent the nominal minimum wage in each year.
Source: PNAD and MTE data from 1981 to 2015. Figure on the top shows changes in log real hourly minimum wage normalized at zero in 1981. The remaining figures show the observed and predicted wage gap between 50th/10th and 90th/50th percentiles for full-time salary/wage workers aged 18–65 years old. Predicted values are obtained from separate OLS regressions of wage gaps on a constant term and the log real minimum wage. Coefficients, robust standard deviations in parentheses, and R-squared are reported.
Source: PNAD and MTE data from 1981 to 2015. The predicted skill premium is obtained by regressing the log of the compositionadjusted wage gap between skill types on a constant, linear time trend, the corresponding log relative supply in efficiency units and the log of the real minimum wage (equation (
Source: PNAD data from 1981 to 2015. Tertiary/non-tertiary and secondary/primary wage gaps are given by the ratio between the weighted average of the composition-adjusted log wages of the corresponding education and experience categories. See 
Source: PNAD data from 1981 to 2015. Log tertiary/non-tertiary and secondary/primary labour supplies are given by the ratio between the weighted average of efficiency units of labour supply of the corresponding education and experience categories. See 
Source: PNAD data from 1981 to 2015. Composition-adjusted log hourly wages for full-time workers are the weighted average of the predicted log wages in each one of the 490 gender-education-experience groups. Each series is normalized at zero in 1981. 
Source: PNAD data from 1981 to 2015. Efficiency units of labour supply are given by the weighted average of the employment participation of 490 gender-education-experience groups. See 
Source: PNAD data from 1981 to 2015. Sample comprises salary/wage workers and self-employed, aged 18–65 years old. Workers employed in the military are excluded from the sample. See 
Source: PNAD data from 1981 to 2015. Log hourly wages of salary/wage workers and self-employed excluding those in the military, aged 18–65 years old. Each series is normalized at zero in 1981. 
Source: PNAD data from 1981 to 2015. Detrended skill premiums and relative labour supplies are the residuals obtained from separate regressions of the composition-adjusted wage gap and the relative labour supply in efficiency units on a constant and a time trend term. See the notes of