Labor Turnover and Labor Legislation in Brazil

One of the main characteristics of the Brazilian labor market is its impressively high job and worker turnover rates. This contrasts with many analysts' view that Brazil has an overly regulated labor market. While the Brazilian labor code is, in fact, very restrictive, dismissal costs are not high when compared with other countries in the region. ¹ Moreover, many authors claim that the design of some job security programs in Brazil creates perverse incentives that stimulate labor turnover to greater levels than would otherwise be attained.

Critics of job security provisions usually argue that dismissal costs tend to create obstacles to functional labor market flexibility, while supporters of dismissal costs stress the potential benefits of reducing income volatility and increasing investment in specific human capital, which could raise productivity in the medium run. According to this latter view, an excessively high turnover rate is a problem, since it might cause underinvestment in human capital and signal a low commitment between employers and employees.

Brazil displays one of the highest labor turnovers in the world for some comparable measures. An average of 3.4 percent of the formally employed enter and leave every month. A high labor turnover partially explains the low quality of jobs observed in Brazil, since labor productivity depends essentially on the level of human capital, either general (through basic education) or specific (through on-the-job training). High labor turnover is a disincentive for training investment because it lowers specific human capital and, therefore, labor productivity.

This paper fully describes the legislation on dismissal costs in Brazil. Following the argument of other authors, I conclude that the design of the Fundo de Garantia por Tempo de Serviço (FGTS) system, a seniority severance payment fund, is inefficient, is a source of conflict between firms and workers, and creates labor turnover.

According to the FGTS system, every firm in Brazil must deposit 8 percent of its current formal employees' wages (8.5 percent since September 2001) into accounts opened in each worker's name in a state bank. With a few exceptions, workers can only withdraw money from these accounts if they are fired without justification, in which case they have access to the FGTS account balance plus a firing penalty paid directly by the employer. Since the returns to the fund are much below market rates, workers have a strong incentive to get hold of their FGTS funds. The firing fine can be de facto negotiated between firms and workers, creating ample room for fake dismissals, by which firms simulate they are firing workers without just cause and are paying the firing penalty. This tends to increase labor turnover. Every year in Brazil, there are around nine million withdrawals from FGTS accounts. ²

The paper uses two episodes of increases in the dismissal fine to empirically identify the effects of firing costs on labor turnover. Specifically, the two episodes--namely, the 1988 Constitution and a labor law introduced in September 2001--are used to test the prediction that higher job termination costs would, other things being equal, reduce turnover for formal workers affected by the legislation. The paper does not advocate that such firing cost increases are good policy responses to the original distortion, which causes turnover to be suboptimally high in Brazil. In fact, a removal of the original distortion through a reform of the whole functioning of the FGTS system would be preferable.

A simple difference-in-differences methodology is used to test the implication of the increased fines based on monthly individual data from the monthly employment survey (Pesquisa Mensal de Emprego, or PME) carried out by the Brazilian Geographical and Statistical Institute (IBGE); this household survey covers the six main metropolitan regions in Brazil and includes information on previous employment spells for those currently unemployed. The methodology exploits the fact that the change to the legislation should have affected some groups of workers differently than two control groups: informal workers and formal workers with...