Demand Estimation in the Presence of Revenue Management

Yield management has become a standard tool in several industries to increase the profits of firms facing demand uncertainty or consumers heterogeneity. But this technique also raises econometric problems in the estimation of demand models. Quantity-based management, in particular, is the source of both an endogeneity and a right-censoring problem. Disposing of macro data only and ignoring these issues leads to an aggregation bias. We develop a structural model of demand in the presence of quantity-based management. We show that the price elasticity is identified in this model provided that (i) we observe two subpopulations that face different prices but are not separated in the yield management policy, (ii) the highest prices and quantities sold at these prices are observed, (iii) the highest prices vary with time or across markets. We apply our method to the French railroad industry, using disaggregated data on trips between Paris and big cities on the period 2007-2009. Our estimates of the price-elasticity are consistent with a rather responsive demand, from 1.7 to 2 in economy class and from 1.3 to 1.5 in business class.