Social employment of welfare recipients in Belgium: an evaluation

In Belgium welfare agencies are subsidized to employ welfare recipients for a period sufficiently long to entitle them to benefits of the contributory social insurance program. This type of employment is called Social Employment (SE). This paper investigates the effect of SE on the exit rate from welfare. We extend the minimum chi-square approach of Cockx (1995) for grouped duration data to deal with selection into SE at any instant during the welfare spell. Moreover, we introduce a specification error as in Amemiya and Nold (1975) to account for omitted variables, which may be correlated with the selection into SE. We show that a variable that is not correlated with the omitted variables, but affects the probability of selection into SE identifies the SE-effect, and we propose a consistent estimator of this effect that is free of selection bias. We argue that region is such a variable, because in Belgium regional governments determine the subsidy for SE, which affects the participation rate in SE by some political process without consideration of the average characteristics of the welfare recipients in their region. The empirical analysis discovers creaming in the selection process. Without correction for selectivity we find that SE reduces welfare dependency. After correction this conclusion is reversed. These results are in line with the incentives faced by the welfare agencies, that administer the program.