Habit Formation and Precautionary Savings: scelte di consumo delle famiglie Italiane pre e post la crisi finanziaria del 2008

In the following pages we try to find evidence of habit formation/durability and of precautionary savings through the estimation of a dynamic panel regression of consumption rate of nondurable goods on the Survey of Household Income and Wealth (SHIW) gathered by the Banca d'Italia on Italian Households. The estimation was implemented with the the first-difference GMM estimator (Arellano and Bond(1991) and Blundell and Bond(1998)). The main goal that we try to achieve in this brief research is to find out if the influence of the 2008's Financial crisis has modified Italian's households preferences, and we do such thing estimating a dynamic panel regression for the consumption rate of nondurable goods that includes in its explanatory variables, one that measures the interaction between time and consumption, i.e. we create a dummy variable for the period crisis/post crisis (2008/2010) and multiply it for the consumption rate (also this regression was estimated with the first-differenced GMM estimator). After having presented a review of the theoretical background starting from the illustration of the basic theory of intertemporal choice and then allowing in the utility function the possibility of time varying preferences, we illustrate the reference model, taken from Rhee (2004), which consists in a maximization problem of the utility function, chosen to be a power utility, and allowing for habit formation, meaning that we relax the hypothesis of additive and time separable preferences. At the end of the maximization problem, we get an Euler equation for the consumption rate to be estimated. We selected the waves of observation from 1998-2010, taking only the panel families. Among all the possible estimators available in literature, we chose the first-difference GMM estimator, that was developed by Arellano and Bond(1991) and Blundell and Bond(1998), we made such choice in order to eliminate the fixed effects contained in the panel dataset. Looking at the results of the dynamic panel regression (with no interaction), we find evidence of durability of consumption since the coefficient is negative and different from zero, but we do not find evidence of precautionary savings, since the coefficient is not significant. On this parameter we may say that the result could be due to the variable chosen to detect the incentive for precautionary savings, so probably changing the variable could signal evidence of this phenomenon. Considering the regression with the interaction variable, we see that there are significant differences between the two sub-periods selected in the data (pre-crisis and crisis). We may say so because the parameter for the interaction is different from zero. We still find no evidence of precautionary savings. With respect to dynamic panel regression, we see in the results for this estimation that there are differences between managers, white-collar and unemployed in consumption choices, once the interaction time-consumption is inserted in the model. The results of the dynamic regression were in line with all the previous works found in literature (Dynan (2000), Guariglia and Rossi (2002) and Rossi (2005)).