In this paper we test a rational expectations permanent income model with data drawn from the 1986-1996 Spanish Family Expenditure Survey. Among the arguments of the utility function, our model does not suppose separability among the components of total consumption (food, other nondurables and durables). The results show that Spanish households incorporate expected information into their intertemporal allocation of nondurable consumption. Nevertheless, the rejection of an excess of sensitivity of consumption to expected income is dependent upon the business cycle. Furthermore, the hypothesis of separability among the components of total consumption, which has a remarkable influence upon the acceptance of the model, is rejected.

Key words: consumption, permanent income hypothesis, panel data, consumption separability, Elasticity of Intertemporal Substitution (EIS).
JEL Classification: D12, D91, E21.