A Simple Newsvendor Bounds Heuristic for Production-Distribution Systems

Erik Lystad and Mark Ferguson

Georgia Tech College of Management

We consider the classic arboreal multiechelon distribution system under periodic review. In the absence of ordering costs, we show that a base stock policy heuristic based on simple upper and lower newsvendor bounds provides solutions that are reasonably close to optimal and outperform previous approximations over a range of settings. In comparison to previous methods, our approach is intuitive, easily implementable, and lends itself to clean observations on the effect of parameter changes on the stocking levels and inventory costs. We show that this scheme may be useful in a number of supply chain contexts; for instance we investigate the proper extent of delayed differentiation without resorting to the common use of a decoupling assumption.

Back to Presenters’ Page