A Stochastic Dominance Approach to Dynamic Pricing Problem

We consider a discrete time dynamic pricing model for perishables with a binary per-period demand. A standard approach to such problems lies in deriving solution under a specific optimality criterion, which is usually risk-neutral revenue maximization. Instead, we approach the problem from stochastic dominance prospective and look for elimination of dominated strategies. We establish conditions that all efficient (i.e. first order stochastically non-dominated) strategies must satisfy. These properties substantially simplify funding an exact distribution of seller’s payoff. It is also shown that all strategies that satisfy these properties are mutually non-dominated, which might explain a variety of pricing patterns observed in real industries. We then derive and compare policies optimal under several criteria common in stochastic optimization and discuss techniques for finding payoff distribution.