Supplier’s opportunistic behavior and the quality-efficiency tradeoff with conventional supply chain contracts

Abstract

This paper presents a supply chain game with a manufacturer and its supplier, where each firm seeks to allocate its own resources between improving design quality and reducing the production cost of a finished product over finite contract duration. The firms agree on a linear contract where the supplier either periodically updates the transfer price, i.e., cost-plus contract (CPC), or sets a definitive transfer price at the beginning of the contract, i.e., wholesale price contract (WPC). Assuming a committed manufacturer, we account for the possibility that the supplier is either committed or non-committed, and derive homogeneous and heterogeneous Nash equilibrium strategies under a CPC and a WPC. We then compare the impact of the supplier’s strategy on the tradeoff between quality and efficiency and the firms’ payoffs, and shed light on the relative merits of a CPC and a WPC. We notably show that a CPC is more robust to the supplier’s strategy type than a WPC in terms of efficiency, quality, and profits. Contrary to the literature, we conclude that a variable transfer price is preferable to a constant transfer price.

Keywords:

• cost-reducing R&D
• design quality
• linear contracts
• supply chain management

Notes

1 It has been shown that the adoption of QFD within many US companies resulted in significant reductions in overall project costs and cycle time and major increases in productivity (e.g., Guinta and Praizler, 1992). In contrast, Vonderembse and Ragunathan (1997) conclude from a survey of 80 QFD projects undertaken by 40 firms that QFD’s implementation showed only modest improvements in terms of product costs. In this regard, we choose to make a prudent assumption that lets us avoid treating QFD as a source of efficiency.

2 An autonomous (i.e., cumulative output-based) learning effect (e.g., Kogan, El Ouardighi, & Chernonog, 2016; Kogan, El Ouardighi, & Herbon, 2017) could have been assumed as an additional source of efficiency in our model. However, this assumption would have changed the focus of the research and significantly reduced the tractability of our problem. Future research could indeed consider adding autonomous learning as a state variable.