Ensaios sobre o uso repetido do seguro-desemprego

Abstract

This thesis is composed of three essays that evaluate the repeated use of the Brazilian Unemployment Insurance. The first essay investigates the extent and nature of repeated use of unemployment insurance from socioeconomic, geographic characteristics, time worked, and receipt of benefits. For this, the accesses of each user of the unemployment insurance program between the months of July 2019 and July 2020 were analyzed using the Unemployment Insurance Management Database (BGSD) as a source. The results obtained from the cumulative logit model without proportional odds, indicate that the most likely characteristics in determining recidivism to unemployment insurance were individuals in the largest age groups, in lower levels of education and salary range, male, workers in the construction and industry sectors, residents in metropolitan regions, and receipt of benefits. Longer working hours and residence in metropolitan areas contributed to reducing recidivism. Given the adverse effects regarding the repeated use of unemployment insurance and the insertion of stricter criteria for the receipt of benefits tied to participation in a qualification program, the second essay analyzed the impact of PRONATEC on the repeated use of unemployment insurance. The sample comprises applications for unemployment insurance daily between 2013 and 2015 obtained from the Unemployment Insurance Management Database (BGSD) in conjunction with RAIS (Annual Social Information Report). To achieve the objective, the PSM (Propensity Score Matching) was employed, by the NNM (Nearest Neighbor Matching) method, associated with the difference-in-differences model with doubly robust estimators. The findings suggest that the insertion of an active policy in unemployment insurance promoted by PRONATEC was not an obstacle to the repeated behavior of beneficiaries of the Brazilian Unemployment Insurance program. Finally, the third essay analyzed the exhaustion effect and the determinants of unemployment duration among non-recidivist and recidivist users of the unemployment insurance system. BGSD microdata was used jointly with RAIS between January 2010 and July 2020 with nonparametric (Kaplan-Meyer), semiparametric (Cox), and parametric (Weibull) models of duration. The results reveal that the conditional probability of leaving unemployment increases from the first to the seventh month of unemployment, when it reaches a maximum point, declining thereafter, characterizing the exhaustion effect between recidivist and nonrecidivist workers. Considering the individual characteristics, it was observed that recidivists had a higher probability of unemployment duration. It is also noteworthy that the parametric estimates, corroborated by the semiparametric ones, reveal that characteristics such as white, with higher wages, more hours worked in the last job, male, with a lower level of education (elementary school), and living in the Midwest and South regions increase the probability of unemployment duration.