Article ID: | iaor201113133 |
Volume: | 20 |
Issue: | 6 |
Start Page Number: | 905 |
End Page Number: | 920 |
Publication Date: | Nov 2011 |
Journal: | Production and Operations Management |
Authors: | Soic Greys |
Keywords: | management, economics, supply & supply chains, manufacturing industries, simulation, simulation: applications, decision, stochastic processes, probability, demand |
In their recent paper, Nagarajan and Sošić study an assembly supply chain in which n suppliers sell complementary components to a downstream assembler, who faces a price-sensitive deterministic demand. Suppliers may form alliances, and each alliance then sells a kit of components to the assembler and determines the price for that kit. The assembler buys the components (kits) from the alliances and sets the selling price of the product. Nagarajan and Sošić consider three modes of competition–supplier Stackelberg, vertical Nash (VN), and assembler Stackelberg models–which correspond to different power structures in the market, and study stable supplier alliances when the assembler faces linear and isoelastic demand. In this paper, we study the impact that demand uncertainty has on stability results obtained in Nagarajan and Sošić. We first analyze models in which all decisions are made before the uncertainty is resolved, and show that the alliance of all suppliers remains stable when demand is isoelastic, or under Stackelberg models when demand is linear. However, demand uncertainty may change stability results when both parties make decisions simultaneously (VN model) and demand is linear. We then extend our results by considering scenarios in which some decisions may be postponed and made after the actual demand is known. When the ordering quantity can be determined after observing the true demand, we show that stable outcomes correspond to those obtained in the deterministic case and uncertainty has no impact on coalition stability; if only the assembler's pricing decision is postponed, we need additional conditions for stability results to carry over in the additive demand model.