The Error Structure of Earnings: An Analysis on Italian Longitudinal Data

Abstract

The aim of this paper is to characterize the time series properties of earnings in Italy, using the panel data set drawn from the Bank of Italy Survey of Household Income and Wealth (SHIW). The Bank of Italy Survey is drawn every two years: this feature raises identification problems as the first-order autocovariance is not observed. However, it is possible to use the panel dimension of the data set in order to discriminate between several specifications that imply different covariance patterns. In order to exploit the differences that may arise due to heterogeneuos education attainments, estimates are performed by education group. Results show that the AR(1) plus individual effect model provides the best characterization of the unobserved component of the earnings process. The estimated autoregressive parameter, however, is well below unity, indicating stationarity.