Survival analysis of transitions from benefit to work

What factors affect the probability that a person makes a transition from benefit to employment? What is the effect of those factors? Given information such as age, sex, most recent occupation and industry, can we estimate the probability of such a transition? We applied the proportional hazards model to Linked Employer-Employee Data (LEED) to answer those questions. The anonymous longitudinal administrative data is from Inland Revenue and is based on monthly returns. Our principal finding was that, of the limited variables available, age and sex have the most significant impact, and that the difference between sexes is greatest in under-35-year-olds. We also found differences by industry and occupation, as well as some regional differences and time effects.