We assess the empirical validity of the life-cycle model using a time series of cross sections and a novel parametrization of preferences. The main findings are as follows: (i) The excess sensitivity of consumption growth to labor income disappears when we control for demographic variables. (ii) The elasticity of intertemporal substitution (EIS) is a function of several variables, including the level of consumption. The EIS increases with the level of consumption. (iii) The variables that change the EIS are also important in explaining excess sensitivity over the business cycle. We are able to reconcile our results with those in the macro and micro literature.