A Microeconometric Characterization of Household Consumption Using Quantile Regression

Abstract

The paper uses cross-section micro data from the GfK consumer panel for an econometric demand analysis of households in Germany. Contrary to most research which considered "average" behavior, we analyze consumer behavior for different "intensities" of consumption. Our analytical tool is quantile regression which allows us to describe the conditional distribution for any quantile including the (conditional) median representing "average" behavior. As an illustrative example, we use the demand for beer and wine. The paper shows quite distinct patterns regarding price and income effects for light and heavy consumption, respectively, which leads to a better characterization of household demand.

Keywords: Demand analysis, consumer panel, micro data, quantile regression

JEL Classification: C21, C49, D12