The labor market in brazil: analysis of number of hires and layoffs

Abstract

The objective of this article is to evaluate the relationship between the number of workers hired and dismissed in Brazil. To do so, we used data provided by the General Register of Employment and Unemployment (CAGED) of the Ministry of Labor and Employment (MTE) of Brazil. The study period covers the years 2011 to 2015 and analyzed all economic sectors as well as all regions of the country. It was considered that the factors that could influence the number of hires and layoffs would be the average salary range of the workers (a qualitative variable comprising 5 levels, measured in the number of minimum wages), the level of education (a qualitative variable with 5 levels: Illiterate, Elementary School, High School, Higher Education, and Unclassified), and sex (qualitative variable). Using Excel software, data was standardized and arranged in a panel database in balanced format, considering the levels of education as the identifying factor of individuals.
Initially a linear regression model was adjusted, considering as dependent variable the Number of Hires (designated in Brazilian Portuguese by "Admitted") and as predictors the remaining variables under study. When the significance of the regression was tested, it was concluded that the Number of Hires was not influenced by Sex. A linear regression model was obtained for the Number of Hires as a function of the Number of Layoffs (designated in Brazilian Portuguese by "Desligados") and the Salary Range in which all variables are significant. Subsequently fixed and random effects models were applied and the relevant statistical tests were carried out.
The fixed effects model intends to control the specific effects of individuals considering that the intersection of the model varies from one individual to another, but is constant over time, while the remaining parameters are constant for all individuals and in all periods of time (Duarte, Lamounier & Takamatsu, 2007). The fixed effects model is also known as LSDV - Least Square Dummy Variables model, since it is a generalization of the linear regression model for panel data, introducing a dummy variable for the effects of variables that designate individuals and / or time.
The random effects model treats individual effects as random variables, considering that individuals on whom data are available are random samples from a larger population of individuals (Duarte, Lamounier & Takamatsu, 2007). According to Pinheiro and Bates (2014, p.3) they are associated with individual experimental units randomly withdrawn from a population. In this model, it is assumed that there is no correlation between the individual effects and the other regressor variables. Their estimation is done using generalized least squares (GLS).
After the estimation of the fixed and random effects models, statistical tests of comparison between the models were applied: Hausman test and Breusch-Pagan test of Lagrange multipliers. Thus, we opted for the random effects model, which returned the best results in the tests.
It was observed that there was a decrease in the number of admissions in Brazil in the last year analyzed (2015), which is believed to reflect the political and economic crisis that began in the
country. It has also been identified that the number of people hired has greater movement in the range that includes the salary between 1 and 3 minimum wages, which corresponds to the average salary of a Brazilian citizen. The highest number of people hired is in the 2nd salary range, that is, they receive between 1 and 3 minimum wages, proving that the largest number of workers is in the average salary range of Brazil. Likewise, it was identified a greater movement of workers with elementary and high school education, due to the fact that higher education has only recently developed in the country.