The implications of mean scaling for the calculation of aggregate consumer elasticities

Mean scaling is a common assumption in the estimation of aggregate consumer elasticities—in particular, expenditure elasticities, but also (implicitly) compensated price elasticities. The assumption is that each household’s income changes in the same proportion as aggregate income. If correct, that implies no bias in the use of aggregate data for estimation of expenditure elasticities. If incorrect, though, there may be substantial bias, especially if there is a high degree of inequality in the underlying income distribution, and regardless of whether one uses micro or aggregate data. We explore this issue, both theoretically and illustratively, using realistic (empirically derived) elasticity estimates coupled with relatively high and low degrees of income inequality.