Estimation of Quadratic Expenditure Systems Using German Household Budget Data

Drawing on household budget data from the German Income and Consumption Surveys of 1988 and 1993, we estimate demographicaily structured Quadratic Expenditure Systems. Both Full Information Maximum Likelihood methods and a Limited Information approach as proposed recently by Ding and Hadri (1996) are employed. The quadratic specification is found to significantly improve the model fit as compared to the linear one used in previous demand studies for Germany. Price elasticities of demand for different household types are analyzed at distinct income levels, and ratios of estimated subsistence expenditures are compared to both institutional and empirical equivalence scales.