Aggregate Consumption Function with Coefficients Random across Countries

Abstract

The problem of estimating an economic relationship from the combined time series and cross-country data poses a serious challenge to econometricians, cf. Houthakker (1961, 1962, 1965). Houthakker (1961, 1962) estimated an aggregate inter-country savings function to find out the determinants of savings in developed and underdeveloped countries. He computed, among other things, a weighted regression of per capita personal savings on per capita disposable income with weights inversely proportional to the square of personal income per capita. These weights were determined on an ad hoc basis. Such studies, useful as they are, suffer from serious limitations imposed by the procedure of pooling data on countries with different levels of development and economic activity. A simple weighting scheme like the one adopted by Houthakker may not eliminate all the heterogeneity among countries. To understand the limitations of these studies it is always desirable to subject the same body of data to different types of analyses based on different sets of assumptions. Zellner and Sankar (1967) analyzed Houthakker’s data from the standpoint of errors-in-variables model in an effort to. determine whether the average propensity to consume depends on the level of income. In this chapter we make an attempt to analyze cross-country data on aggregate consumption expenditures starting from a different set of assumptions. We first set up an aggregate consumption model on the basis of some existing theories of consumption and assume that such a function is defined for each country appearing in our cross-section sample.