School of Management is delighted to welcome Professor Mihalis Markakis, IESE to host a research seminar discussing ‘Bayesian Demand Learning and Revenue Management under Limited Capacity’.

Abstract 

Marketplace platforms provide a convenient meeting point between customers and suppliers, and have become an important element of e-commerce. This sales channel is particularly interesting to suppliers that wish to learn quickly the popularity of new products, because a platform provides expanded reach to potential clientele. On the other hand, the associated lower margins (due to the platform’s cut) constitute a disincentive for suppliers to stay with the platform after learning enough about the demand for their products. In this paper, we show that, under limited capacity, the optimal price/quantity decision is primarily driven by the total demand rate during the sales season. When the latter is unknown, the seller may need to adjust its revenue management tactics over time as it learns the demand rate. We formulate the dynamic optimization problem under Bayesian learning. We solve it analytically when there is one unit of capacity for sale, and develop effective heuristics for the multi-unit case. Surprisingly, we find that lack of information pushes the seller to opt for a low-volume, high-margin position for a longer time, and only reduce margins close to the end of the selling season, as opposed to earlier, which would accelerate learning.

Joint work with Victor Martínez de Albéniz and Marcos Serrano.