Inflation and inequality bias in the presence of bulk purchases for food and drinks

Abstract

We study how to improve the estimation of annual food expenditures from household budget surveys with limited information on bulk purchases. We compare three alternative imputation methods by i) estimating the average amount of over or undervaluation according to a Poisson model of the frequency of purchase, ii) measuring the impact on a food share regression of belonging to household types with different bulk purchase habits, and iii) analyzing the outliers exclusively attributable to the shortcomings of each imputation procedure. Finally, we study the implications of different alternatives on the measurement of food price inflation, food expenditure inequality, and household total expenditure inequality.