The contribution of minimum wage valorization policy to the decline in household income inequality in Brazil: A decomposition approach

ABSTRACT

There is a vast literature that estimates the effect of the minimum wage on wage inequality in various countries. However, as the minimum wage directly affects nonlabor income of families in some countries (in the Brazilian case via the benefits of the pension system and of certain social programs), this article extends the empirical analysis by studying the effects of the minimum wage on the level of inequality of household income as a whole. To accomplish that we employ a decomposition method that gauges the contribution of the increases in the minimum wage that occurred in recent decades in Brazil through the labor and nonlabor sources of household income. The results show that the minimum wage had a contribution of 64 percent to the observed fall in income inequality between 1995 and 2014 and that pensions were the most relevant channel over this period.

KEYWORDS:

• Decomposition
• inequality
• labor and nonlabor income
• minimum wage
• pensions

JEL CLASSIFICATIONS:

• D31
• D63
• E24
• E64
• H53
• H55
• J31

Notes

¹See Anuário Estatístico da Previdência Social, 2013.

²Though the 1988 constitution precludes the use of the minimum wage as an index for other wages, this type of informal arrangement is often used in the labor market.

³We ignore the eventual influence of the minimum wage on other nonlabor income of households.

⁴Note that the contribution of the change in the value of minimum wage can be computed separately through the labor and nonlabor (pensions and BPC) channels.

⁵This is a simplified example since we do not show in Table 1 the counterfactuals associated with changing the minimum wage value for each component (labor, pensions, and the BPC) separately.

⁶See Fortin, Lemieux, and Firpo (2011) for a general review of decomposition methods in applied microeconomics.

⁷Another difference between the studies is that Barros et al. (2006) simulate the counterfactual distributions in a fashion that also incorporates the association between some subsets of two variables. Azevedo et al. (2013a) include one variable at a time disregarding the contribution of these associations of variables. The point raised by Azevedo et al. (2013a) for sustaining their method is that isolating the contribution of the association of subsets of two specific variables is partial since the change in one variable can potentially affect all other variables. See Azevedo et al. (2013a) for a comparison of the two methods.

⁸As we have six variables, there were 6! = 720 possible paths to consider. Operationally, this was done via the Stata ado file “adecomp” developed by Azevedo, Nguyen, and Sanfelice (2012).

⁹Because the rural areas of the states of Rondônia, Acre, Amazonas, Roraima, Pará, and Amapá were not included in the PNAD sample until 2004, households in those areas were excluded from the analysis. However, these households represented a small share of 2.6 percent of the sample, and the population of these areas accounted for only 1.9 percent of the population in 2014.

¹⁰Although the Lula administration started in 2003, it was decided to always use the final year of the previous administration as the first year of the next administration to maintain the continuity of the series. Because the second Lula administration ended in 2010, a demographic census year in which the PNAD is not conducted, we used the PNAD 2009 for its closing year. The Cardoso administrations also lose a year with the absence of the PNAD in 1994. Thus, we kept seven years for the Cardoso administrations, seven years for the Lula administrations, and five years for the Dilma administration.

¹¹This occurs when at least one source of income of any household resident either is not reported or does not comply with the IBGE’s data review and is therefore ignored to avoid affecting the income level.

¹²These results are available from the authors upon request.

¹³Available at www.mds.gov.br/relcrys/bpc/1_tab_evolucao_concessao.htm and www.mds.gov.br/relcrys/bpc/download_beneficiarios_bpc.htm.

¹⁴According to Kerstenetzky (2012), in 2003, the elderly statute reduced the age of eligibility for the BPC from sixty-seven to sixty-five, and in 2007, there was a relaxation of the concept of disability. Both measures have contributed to increasing the target audience for welfare benefits.

¹⁵The line for 1995 is not plotted because of the low incidence of BPC recipients in the PNAD that year.

¹⁶The first group consists of workers with a formal employment contract and self-employed workers and employers who contribute to the pension system; the second group is complementary to the first group, that is, it consists of workers without a formal labor contract and self-employed workers and employers who do not contribute to the pension system. It should be noted that according to Reis, Corseuil, and Brito (2015), the use of the criterion of contribution to the pension system for self-employed workers and employers leads to a formality classification similar to that obtained when using the criterion of formal business registration (National Register of Legal Entities).

¹⁷We used the command adecomp, which was developed by Azevedo, Nguyen, and Sanfelice (2012) for Stata. Although this command delivers results for the Gini and Theil indexes, inequality differences calculated by the Theil index between periods do not coincide with those calculated by other procedures. Therefore, we decided to present only the results for the Gini index, whose differences were confirmed by other calculation methods.

¹⁸Because of the low incidence of the BCP in some years of analysis, we did not break down its contribution between the regions.

Additional information

Notes on contributors

Alessandra Brito

Alessandra Brito is a Researcher at the Instituto Brasileiro de Geografia e Estatística (IBGE), Brazil.

Miguel Foguel

Miguel Foguel is a Researcher at the Instituto de Pesquisa Econômica Aplicada (IPEA), Brazil.

Celia Kerstenetzky

Celia Kerstenetzky is a Full professor at Economics Institute of Federal University of Rio de Janeiro (UFRJ), Brazil.