Do differences in reported expenditures between household scanner data and expenditure surveys matter in health policy research?

Household scanner data are increasingly used to inform health policy such as sugar‐sweetened beverage taxes. This article examines whether differences in the level of reported expenditures between IRI Consumer Network scanner panel and the Consumer Expenditure Survey (CES) lead to important differences in demand elasticities and policy simulation outcomes. Using each dataset, we estimated a structural consumer demand system with seven food groups and a numéraire good. To compare the two datasets on a level playing field, we went to great lengths to ensure that the explanatory variables in the two demand models were comparably constructed. Results indicate that scanner data households are not consistently more price responsive than the general population and underreported Consumer Network expenditures do not seem to result in systematic differences in price elasticities. The income elasticities are uniformly lower in Consumer Network than in CES for higher income households because of the positive association between income and the degree of underreporting. This, however, has limited effects on uncompensated price elasticities and policy simulations because food budget shares are small for higher income households. Overall, these findings support continued use of household scanner data in health policy research related to effects of price (dis)incentives.