Pricing and learning with uncertain demand

Practical policies for the monopolistic pricing problem with uncertain demand are discussed (for discrete time, continuous prices and demand, in a linear and Gaussian setting). With this model, the introduction of price variations is rationally justified, to allow for a better estimate of the elasticity of demand, and increased profits due to better pricing. An approximation of the dynamic programming solution is introduced, exploiting convex optimization methods for computational tractability. Numerical experiments are described. • First draft April 1999 (talk at INFORMS Cincinnati, May 1999), second draft June 2003 (talk at INFORMS Revenue Management Conference, Columbia University,