This paper attempts to find out the explanatory power of observed changes in the distribution of current income in explaining the relative change in aggregate consumption using the distributional approach of aggregation by Hildenbrand and Kneip (1999, 2001). The coefficients in the aggregate relation in each period can be estimated from cross-section data as an average derivative of the individual consumption relation without requiring a specific functional form of this relation and the time-invariance property needed in the time-series model of Davidson et al. (1978). We use the nonparametric Direct Average Derivative Estimation technique to estimate these coefficients of the aggregate relation in the Family Expenditure Survey of UK [1974-1993]. It turns out that for most of the commodity groups the inclusion of dispersion in the income distribution, in addition to the mean, improves the predictive power of the model.
Aggregate Consumption Income Distribution Heterogeneity Average Derivative Estimator Representative Agent Model