Abstract
We measure price elasticity across 19 grocery categories over 24 quarters. For each category we estimate a separate random coefficients logit model with quarter-specific price covariates and control functions to address endogeneity. Our specification yields a novel set of 456 elasticities across categories and time that were generated using the same method and therefore can be directly compared. The majority of categories are countercyclical — price sensitivity rises when the macroeconomy weakens. However, there is substantial variation, with a handful of categories exhibiting procyclical price sensitivity and another group showing no relationship. We show that the relationship between price sensitivity and macroeconomic growth depends strongly on the average level of price sensitivity in a category. We explore whether variation in market power and substitution across consumption channels could explain this pattern, finding some support for each argument. We also consider a variety of alternative specifications to ensure our main results are robust.