Abstract:
In this paper we analyse the movements of French young people between three states: employment, unemployment and non-participation, using data from the waves 1990-1992 of the French Labour Survey. Some of these event histories are left-censored. We therefore address the problem of initial conditions by invoking a stationarity assumption in the statistical framework of multi-states multi-spells duration models. We subsequently construct a Hausman test of the stationarity assumption We allow for interdependence between duration variables and we use an adjusted likelihood ratio test in order to choose the best distribution for unobservables. We are particularly interested in identifying characteristics that are associated with less favourable event histories. The results show that higher educational attainments are associated with shorter unemployment durations and with longer durations when employed. Another important factor is nationality; indeed to be of French nationality is often associated with better outcomes in the labour market.