New evidence on state dependence in unemployment histories

Using administrative data records from the Spanish Employment Agency we examine whether or not there is evidence of state dependence in unemployment under benefits in Spanish young workers. For this fact, we use a mixed proportional hazard model that allows for state dependence through lagged duration dependence in order to disentangle the effects of unobserved individual heterogeneity and the true state dependence. We have found evidence that past unemployment experience and unobserved individual components affect the experience of longer future unemployment spells under benefits. However, we appreciate in workers with completed past unemployment spells that the correlation between the duration of succesive unemployment spells is only due to the unobserved components across individuals. Besides, we observe that workers in their second unemployment experience under benefits present higher hazard rates that in their first unemployment experience under benefits.