Loss Leader Pricing and Complementarity: Empirical Evidence from Gas Stations

We study the effectiveness of loss-leader pricing using cash discounts for fuel at more than 2,700 U.S. gas stations that change their discounting behavior. Cash discounts reduce fuel margins and increase cash-paid fuel sales, but they also increase convenience-store sales and, surprisingly, card-paid fuel sales. These gains appear to come primarily from attracting additional customers rather than changing the purchasing behavior of existing customers. Our estimates imply that, absent these spillovers, cash discounts would be profitable for only 3% of stations. Once spillovers to card-paid fuel and convenience-store purchases are included, however, roughly 80% of stations in our sample would benefit from discounting. Yet stations discount far less often than our estimates predict. We explore several explanations for this gap. Although the returns to discounting vary substantially and predictably across stations, expected profitability does not meaningfully predict actual discounting behavior. Learning may be difficult: using only a station’s or chain’s own data, identifying whether discounts are profitable is challenging. At the same time, nearby competitors’ discounting strongly predicts a station’s own behavior, consistent with the use of simple heuristics. Overall, we provide empirical evidence that loss-leader pricing can be profitable through cross-product and payment-method spillovers, but is substantially underused.