Low-Wage Mobility in the Italian Labour Market

This paper uses SHIW panel data for 1993 and 1995 to model individual transition probabilities at the bottom of the Italian wage distribution. The analysis is based on a bivariate probit model with endogenous switching which allows tackling the initial conditions problem, i.e. the potential endogeneity of the conditioning starting state. Results show the appropriateness of such a choice: the correlation between state and transition probabilities is significantly different from zero, while overlooking endogeneity leads to overstatement of both size and significance of coefficients in the transition equation. The paper shows that while some factors such as education, sex and geographical location have an effect on low-pay persistence, job related variables are more effective in avoiding falls into low-pay from higher pay. It is also shown how raw persistence involves a considerable share of true state dependence, pointing towards the existence of low-pay stigma