What Happens When a Retailer Drops a Product Category? Investigating the Consequences of Ending Tobacco Sales

We measure the cross-category spillover effects of a retailer changing its assortment at the extensive margin (by dropping an entire category from its portfolio) on the outcomes for its rivals in the industry. By leveraging the quasi-experimental nature of the exit by a national pharmacy chain, hereafter referred to as the exiting chain (EC), from the tobacco category, we measure the effects of this event on the revenue generated by nontobacco products at rival pharmacy chains. We show, using Nielsen store-level aggregate (Retail Measurement Services) data, that for each 1% increase of cigarette sales in non-EC stores that are located near an “exiting” EC store relative to those non-EC stores that are not, the revenue generated by nontobacco products grows by 0.04%. Next, using Nielsen household panel data (Homescan), we try to uncover the mechanism underlying this spillover by looking at how the behavior of smokers and nonsmokers at EC stores is affected by the decision. We find that the frequency of trips to the exiting stores that included some tobacco product was negatively affected, suggesting that tobacco was one of the main drivers of store patronage for those trips. For nonsmokers, we find that they react to EC’s action by increasing the frequency of trips to the exiting chain. However, the gains from nonsmokers do not seem to outweigh losses caused by smokers. To assess the generalizability of our results, we analyze the impact of a set of tobacco bans imposed by municipalities in Massachusetts and find cross-category loss patterns for drugstores in those areas that are similar to the gains to rival stores from EC’s exit.