The poor do not pay more: Evidence from Tanzanian consumer food expenditures controlling for the food environment

Abstract

We revisit the question of whether the poor pay more for food using household food expenditure data from Tanzania. We control for spatial factors that could affect food prices, namely, whether a rural household is in a peri-urban, intermediate, or hinterland rural zone (distinguished by distance to urban areas) and whether an urban household is in a primary city, secondary city, or a town. Our results differ from conventional wisdom. First, we find that the rich and the poor pay about the same price per kilogram on average for certain key Tanzanian food products like rice, maize flour, and cooking oil. Second, we find that poor households do not buy meaningfully smaller quantities per transaction than do richer households. Third, the rich and the poor also make roughly the same number of purchases per month for most food products studied. Lastly, we find that bulk discounts (a decrease in price per kilogram with an increase in quantity purchased) do exist, but only up to a certain quantity (with exceptions in some animal proteins), below which few households purchase. Hence, our evidence suggests that poorer Tanzanian households do not pay significantly more for key food products.