| 肖勇波
清华大学经管学院助理教授
|
|
【主题】Managing the Portfolio of Probabilistic Goods
【时间】2011-10-268：30-10：30
【地点】清华经管学院 伟伦453
【语言】中文
【内容摘要】
By offering consumers a choice involving buyer uncertainty, probabilistic selling (PS) provides a new dimension for retailers to segment the market and to benefit from price discrimination. It is shown in the literature the creation of "probabilistic goods" (PGs) has great potential in improving revenue. This paper extends the idea behind PS to a more general situation, by developing a menu of PGs based on limited number of physical items. Each specific PG is characterized by a combination of selling price and fulfillment probability. We consider an online retailer selling two similar and seasonal products, A and B, within a finite period. To maximize revenue, the retailer dynamically controls the offering of the PG portfolio, which is a subset of the $m$ available PGs, based on the selling state of the retailer. Starting from the choice behavior of customers, we model the problem using continuous-time, discrete-state, finite-horizon dynamic programming. We study the optimal admission d
ecisions for the retailer and provide a solution using a constructive method. Due to the complexities of the optimal policy for this MDP model, we propose to study a corresponding fluid control problem, in which the customer arrival and fulfillment portion are assumed to be deterministic. We develop a fluid policy based on the solution of the fluid control problem, which has a nice structure and thus is easily implementable. Furthermore, when implemented for the MDP model, the fluid policy is shown to be asymptotically optimal for fast-moving items. Numerical experiments are conducted to show the revenue impact by managing the portfolio of PGs.
Keywords: Probabilistic Selling; Customer Choice Behavior; Admission Control; Fluid Policy, Asymptotic Optimality. |
