Modelling employment spells from Emilian labour market

The analysis of employment spells obtained from a Cross Population Survey (CPS) should take into account the phenomenon of length-bias because CPS-type data are generally right-censored. The paper presents the procedure to analyse a data set which includes only censored or incomplete spells. First of all, the distribution of the incomplete spells of employment are analysed without explanatory variables (such as age, sex, etc.), assuming that the unobserved completed spells have a Weibull, a logistic, or a Gompertz distribution, and the best assumption is adopted (the Weibull). Using the constant and the time-dependent proportional hazards models, the methodology to study the effects of explanatory variables on spell distribution is then described for both incomplete spells and incomplete spells considered as completed ones. These methods are then applied to the lengths of ongoing spells of employment for employed and self-employed workers.