On the covariance structure and mobility of Italian wages

This Thesis uses Italian panel micro-data to investigate the intertemporal wage covariance structure and the extent of transition probabilities at the bottom of the wage distribution, producing new and original evidence on the degree of persistence of cross-sectional wage differentials over individual life-cycles and on the features of the process governing wage mobility across low-pay thresholds. Chapter 1 presents a survey of the debate generated by the rise of wage inequality observed in many industrialised economies and stresses how longitudinal analyses of wage persistence and mobility shed light on the long term impact of rising cross-sectional dispersion; a survey of the two research areas to which this Thesis contributes, i.e. variance components models of the wage covariance structure and econometric modelling of transition probabilities, is also presented. Variance components models of the wage covariance structure are estimated in Chapters 2 and 3, where two unbalanced panels drawn from the Social Security archive on the 1974-88 and 1979-95 intervals (respectively) are analysed by applying the minimum distance technique. Chapter 2 shows that while permanent wage profiles converged within the overall wage distribution, divergence can be detected for white collar workers, suggesting that the former could have been imparted by the egalitarian wage policies of the late 1970s. Results from Chapter 3 indicate that the rising wage inequality observed in Italy over the 1980s and the early 1990s permanently affected the evolution of wage profiles especially during the second half of the 1980s; on the other hand, increases in the relative importance of wage volatility are shown to characterise the first half of the 1990s, thus mirroring the higher labour market “flexibility" of recent years. The Chapter also takes into account the relationship between covariance structure components and observable workers characteristics; in particular, a model which shifts the parameters of interest with respect to workers’ occupations is developed, finding that permanent differentials arise from the wage distribution of white collar workers. A bivariate probit model with endogenous switching is developed in Chapter 4 to analyse low-wage mobility taking the endogeneity of starting wage states into account, using survey data from the Bank of Italy. Results indicate the appropriateness of such a framework, the correlation between state and transition probabilities being statistically significant. While workers’ attributes are found to have a limited impact on the probability of leaving low-pay, a considerable share of aggregate low-pay persistence appears to be the consequence of true state dependence, i.e. the experience of low-pay raises, per se, the likelihood that the phenomenon occur in the future. Chapter 5 checks the robustness of these conclusions to the presence of endogenous attrition from the wage distribution over time by augmenting the model with a third equation for the probability of belonging to the balanced sample. The computational difficulties posed by the required evaluation of trivariate normal integrals are overcome by implementing simulation estimation techniques. Results indicate that exits from the wage distribution over time are an ignorable source of sample selection for the estimation of low-pay transition probabilities on these data, thus pointing towards the robustness of the findings of Chapter 4 to this generalization of the model