Abstract

This paper analyses financial distress among Italian households using the longitudinal component of the Bank of Italy Survey on Household Income and Wealth (SHIW) for the period 1998-2006. It aims to test whether the probability of experiencing financial difficulties is persistent over time. First we review the methodologies for estimating dynamic nonlinear panel data models, drawing attention to the problems to be dealt with to obtain consistent estimators. Specific attention is given to the initial condition problem introduced by the presence of the lagged dependent variable in the set of explanatory variables. Second we provide an in-depth discussion of the alternative approaches proposed in the literature - subjective/qualitative versus quantitative indicators - to identify households in financial distress. We define a quantitative measure of financial distress based on the distribution of net wealth. Finally we apply dynamic probit models to test for true state dependence in financial distress. The estimation uses four different methods: the pooled probit; the random effects probit with exogenous initial conditions; the Heckman model; and the more recent Wooldridge model. The results of all the estimators confirm the null hypothesis of true state dependence and show that, in line with the literature, less sophisticated models, namely pooled and exogenous models, tend to over-estimate this persistence.