DURAÇÃO E TAXA DE SAÍDA DO DESEMPREGO: EVIDÊNCIAS DE AUSÊNCIA DE DEPENDÊNCIA NA DURAÇÃO PARA AS REGIÕES METROPOLITANAS DO BRASIL (1984-2000)

The proportion of long-term unemployed (6 or more months) in the six major metropolitan areas in Brazil has grown 72% between 1986 and 2000. At the same time, the average duration of unemployment has risen 43% for the same period. This worrisome information motivates our purpose in this work: study the relation between exit probabilities and unemployment duration. Former studies for Brazil have found a negative relation between those two variables, considered the effect of the existence of a negative unemployment duration dependence. We try to respond below to what extent this is due to the heterogeneity of the unemployment group and to what extent to a true effect of duration dependence, being that question most relevant for policy-oriented decisions. In order to evaluate this matter, we uses the eyeball test proposed in Jackman and Layard (1991) applied to PME data (1984 to 2000). This test analyzes the impact of the average duration variation along the years on the probability of leaving unemployment for different duration classes, separating the effect of heterogeneity from genuine duration dependence. The results contradict former studies for Brazil since we consider that the fall in exit probabilities raises with unemployment duration due to the effect of heterogeneity of unemployed in observed and unobserved characteristics and not to unemployment negative duration dependence. Nevertheless, our findings are in accordance with studies for developed economies, with exception for England.