A Dynamic Approach to Bid-Price-Based Revenue Management in Make-to-order Production

Capacity control problems in make-to-order revenue management typically are solved by applying bid-prices to approximate the opportunity costs of accepting a customer order. However, in the face of stochastic demand, this approximation becomes less accurate, deteriorating the bid price performance. To address this problem, we propose a dynamic bid-price policy, which exploits the informational dynamics inherent to capacity control problems in make-to-order production. Neural networks are applied to incorporate updated demand information. Computational experiments illustrate the superior performance compared to traditional revenue management methods like randomized linear programming. In addition to that, the dynamic bid-price approach features interesting characteristics for risk adverse decision makers. It dominates traditional methods in risk, as well as in average expected contribution margin.