This paper examines the effects of supermarket loyalty programs on the demand for private labels (PLs). Using transaction level data on grocery purchases and individual level information on the membership of loyalty programs, I estimate a model of demand in which membership may affect the consumers' valuation for PLs, their sensitivity to price changes and have spillover effects on both named brands (NBs) and rivals' PLs. My identification strategy of the membership effect exploits observed variation in shopping patterns at the consumer level over time and across customer types (i.e., members and non-members) in each period to control for as much exogenous variation as possible, and includes a control function using characteristics of loyalty programs as instrumental variables to account for a potential selection bias related to unobserved factors of the membership decision. I find a significant effect of loyalty programs on consumer preferences for PLs. Compared to non-members, membership reduces consumers' price sensitivity for the products sold by the supermarket they are members of, but increases it for products sold by supermarkets they are not members of. These effects are weaker for households that are members of the loyalty programs of multiple supermarkets. Counterfactual simulations show that when a supermarket modifies its loyalty program while competitors keep their own unchanged, it loses about 19% of customers to its rivals, on average. Furthermore, if loyalty programs were changed altogether, the demand for PLs would considerably decrease, while the demand for NBs would increase.
