The search for success: do the unemployed find stable employment?

We use discrete time proportional hazards regressions to model the impact of previous unemployment incidence and duration on job tenure. We find that jobs that follow an unemployment spell have shorter mean duration than other jobs. Only one half of jobs that follow unemployment last for 12 months. Multivariate results suggest that an unemployment spell has a severe penalty on subsequent job tenure, and that it is unemployment incidence rather than duration which has the major impact.