The formal-informal labor market segmentation hypothesis revisited

Despite the existence of a large empirical literature, there is no conclusive answer to whether labor markets in developing countries are segmented. Moreover, a prior and perhaps more fundamental question remains largely unanswered, namely, whether the formal-informal segmentation hypothesis actually has empirical content. This paper tackles the latter question by developing a stochastic dynamic discrete choice model to analyze workers' choices between unemployment, formal and informal jobs. This general framework is then used to assess the empirical implications of dierent models of labor market functioning. As a complete taxonomy exercise is beyond the scope of this paper, I focus on the two most polar cases in the present context: the fully integrated and fully segmented labor markets. The nal goal is to analyze to which extent these two extreme benchmarks can generate implications that are empirically distinguishable and, more broadly, whether the ndings in the empirical literature can be supported by a single model or by a broad array of models. The analysis shows that both models can support the observed regularities regarding formal-informal wage gaps, transitions and job duration. Hence, it does not seem possible to identify the existence of segmentation from the moments typically analyzed in the empirical literature.