Unobserved Heterogeneity and Intertemporal Nonseparability: Evidence from Consumption Panel Data

In this paper we analyze the importance of intertemporal non-separabilities for consumption decisions using household data. We follow the test suggested by Meghir and Weber (1996), so we exploit the variability of the within-period marginal rate of substitution (MRS) between commodities. We also check for the presence of liquidity constraints by comparing the results obtained from theMRS to those of the Euler equations. For that purpose, we use a Spanish data set in which households are observed up to eight consecutive quarters. This length of the temporal dimension is crucial, since it allows both to account for the dynamics of consumption in the preferences as well as to control for time invariant unobserved heterogeneity across households. Our results confirm the importance of accounting for fixed effects when analyzing intertemporal consumption decisions allowing for time non-separabilities. Once we control for fixed effects and use the adequate set of instruments we do not find evidence of misspecification and the results yield supporting evidence of habit formation.