Inventory Management Under Heavy-Tailed Demand

Researchers: Adrian Perez (with Benjamin P. Harris and Canan Corlu)

Inventory management literature was mainly built on the assumption that product demands are normally distributed.  However, in recent years, this simplifying normally distributed demand assumption has been questioned by researchers. In fact, there is growing evidence in the literature that some item demands including book demand at Amazon, movie demand at Netflix, and spare part demand of a European automobile manufacturer follow a heavy-tailed distribution (i.e., Pareto distribution). It is crucial to investigate how the inventory management models that are built under the normal demand assumption would perform in the presence of heavy-tailed demand.

In this research study, we focus on a multi-item budget-constrained inventory optimization setting where the goal is to maximize customer service level subject to a budget constraint on the total inventory investment. We formulate the inventory optimization problem assuming Pareto distributed demand and conduct extensive numerical experiments to investigate the impact of several parameters on the optimal inventory positions.