A Behavioral Study of Capacity Allocation in Revenue Management

In this paper we present a set of laboratory experiments that investigate how human subjects solve the two-class capacity allocation revenue management problem. The two-class problem is the simplest possible revenue management problem, that is a fundamental building block for a family of complex and sophisticated RM problems. We study the capacity allocation problem with ordered and unordered arrival, as well as a simplified version of the problem in which the decision-maker makes the single capacity allocation decision at the beginning of the selling season. We find that our laboratory participant generally accept too many low class customers, leave to few units of unused capacity, and as a result do not accept enough high class customers. We also find that making decisions up-front improves performance in the ordered arrivals case, and does not hurt performance in the unordered arrivals case. Lastly, we find that in the unordered arrivals setting, participants accept more low class customers than they should, when there is ample capacity, and reject more low class customers than they should when there is scarce capacity.