Overbooking can help service providers improve revenues, but it comes with costs, such as the compensation paid to “bumped” customers. The existing operations management (OM) literature focused on booking limits implicitly assumes that, given pre-defined price
and bumping compensation, those booking limits do not affect customer demand. Empirical research in marketing however, suggests that bumping negatively affects demand, an important effect that is the focus of this paper. With airlines as a backdrop, we formulate a model in which a service provider jointly determines price, booking limit, and bumping compensation, and customer reactions to the three induce an equilibrium demand distribution. For the traditional setting in which price and bumping compensation are exogenously fixed, we provide sufficient conditions under which bumping sensitivity leads the airline to reduce overbooking. These conditions suggest that airlines can decouple the value customers obtain from flying from their cost of being bumped. We demonstrate that effective overbooking policies must nevertheless jointly determine bumping compensation and booking limits. We then consider auction-based compensation schemes and show that, for the airline, they dominate fixed-compensation schemes. Numerical experiments demonstrate that: the equilibrium bumping rates obtained from our model are consistent with those observed in practice; fixed-compensation policies that account for bumping-sensitive demand can significantly outperform those that do not; and auction-based policies can bring smaller but still significant additional gains over those provided by fixed-compensation schemes. Our results support the empirical observation that bumping negatively affects customer demand and should be carefully managed, and they suggest that, contrary to traditional booking-limit controls, effective overbooking policies must jointly determine booking limits and bumping compensation. Finally, they demonstrate that recently adopted auction-based compensation schemes are a particularly effective means of managing bumping-sensitive demand.