تاثیر تعاملات مکرر بر قرارداد های زنجیره تامین : یک مطالعه آزمایشگاهی

In this laboratory study, we investigate the interactive behaviors that develop over a perceived long-term contractual relationship. Specifically, three types of contracts are examined under a two-echelon supply chain setting with stochastic demand: wholesale price contracts, buyback contracts and revenue-sharing contracts. The supply chain contracting theory has demonstrated that the simple linear price contract is inefficient; whereas the latter two contracts can coordinate the channel through risk-sharing, and they are mathematically equivalent. We propose an experimental design that controls for individual decision biases to isolate the behavioral impact of repeated negotiations. Our experimental results indicate that participants systematically deviate from predictions by the normative model that assumes a one-shot interaction between self-interested players. We find that when future opportunities to punish are available, social preferences for fairness and reciprocity are reinforced; and reputation-building behaviors are motivated to achieve long-term economic benefits. As a result, the performance of the overall supply chain is enhanced. Moreover, we observe that buyback contracts behave differently from revenue-sharing contracts by inducing higher order quantities over time.

Contractual relationships among supply chain partners have attracted considerable interest from both practitioners and academicians. In any channel that consists of individual firms, inefficiency may arise due to decentralized decisions motivated by local interests. Particular research efforts have been devoted to the analytical design of contracting mechanisms to align the economic incentives of multiple parties in a supply chain. Cachon (2003) provides a comprehensive review of the theoretical literature on coordinating contracts and concludes that future empirical research should “challenge the assumptions and analysis of the theory” (Cachon, 2003, pp. 330–331). Most supply chain contracting models assume that decision makers behave in a way that maximizes their own expected payoffs. In other words, agents are (1) fully rational, and (2) have no social preferences for fairness or reciprocity. However, behavioral studies from various fields have provided counter-evidence. For example, a recent experimental study by Katok and Wu (2009) shows the effectiveness of coordinating contracts are reduced by persistent individual decision biases. Kahneman et al. (1986) argue that willingness to enforce fairness is common, and that even profit-maximizing firms can be motivated by preferences for equitable payoffs. Empirical research in marketing has documented cases where fairness plays an important role in the manufacturer–retailer relationships in many industries, including automobiles, consumer packaged goods, semiconductors, computers, and telecommunications ( Kumar et al., 1995, Kumar, 1996 and Scheer et al., 2003). In the basic contracting model, a supplier initially proposes a contract, and a retailer then responds by determining an order quantity as a newsvendor. Most analyses focus on a one-shot interaction between these two players, thereby excluding the possible social and economic impact from repeated exchanges. In practice, however, even if a retailer is selling a perishable product, there may be multiple opportunities to negotiate the contract arrangement with the same supplier. Past research has suggested that the perceived long-term relationship helps foster trust and reinforce reciprocity. For instance, Dyer (1997) conducts surveys and interviews to analyze automotive channel relationships in the U.S. and Japan. The study reports that Japanese auto companies maintain longer transaction relationships with a smaller group of suppliers compared to their counterparts in the U.S. Because of these “repeated games”, Japanese suppliers indicate that they are more likely to trust Japanese auto-makers to treat them fairly. And due to the anticipation of future opportunities to reward and punish, the incentives to behave opportunistically are largely reduced for Japanese automotive partners. The current study experimentally investigates how decision makers interact through various supply chain contracts when repeated negotiations are possible. We choose to study, under a stochastic demand environment, three types of contracts: the wholesale price contract, the buyback contract, and the revenue-sharing contract. These contracts represent distinct characteristics in general: non-risk-sharing versus risk-sharing, and non-coordinating versus coordinating. Although buyback contracts and revenue-sharing contracts have been demonstrated to be mathematically equivalent (Cachon and Lariviere, 2005), we would also like to examine whether there exists any behavioral difference between these two contracts. The focus of our research is on the strategic behaviors developed over a perceived long-term relationship. In our experiments, pairs of human subjects interact with the same anonymous partner repeatedly via computer. The supplier is asked to determine the parameters for a given type of contract. The retailer, on the other hand, can choose to reject the contract, order the minimum possible demand, or place the order quantity that maximizes her own expected profit (q⁎). Our computer program automatically calculates this optimal order given each offer proposed by the supplier. This information is provided to both players before they make any final decisions. Unlike most newsvendor experiments in the literature, in which the decision maker can choose from a wide range of nonnegative integers as the order quantity, we constrain the retailer's options and offer decision analysis tools to compute q⁎ for all players. The purpose of this design is to control for the impact of possible individual decision biases. Our manipulation is motivated by Katok and Wu (2009), who propose a design to eliminate human interactions. In their experiments, human retailers or suppliers deal with computerized partners. Decision makers do not have any incentive to treat a computer fairly or act strategically if the computer is not programmed to respond. Therefore, any deviations from theoretical predictions can only be attributed to an individual's bounded rationality. In this study, we intend to isolate the effect of repeated interpersonal interactions by the current design. If a subject knows what would be the “optimal” decision beforehand and still chooses to behave differently, it must be due to strategic concerns, not cognitive limitations. Since Katok and Wu (2009) have removed social considerations, their results are more in line with theories that assume self-interest, and thus will serve as a benchmark for our study. We observe that the behaviors of both the retailer and the supplier differ systematically from what the normative theory predicts. Repeated interactions improve the supply chain performance in general. Namely, the linear price contract performs better than suggested, even when the efficiency loss from negotiation failure is considered. The channel coordination is approximated under the two risk-sharing contracts, but at the cost of the supplier. Buyback contracts induce higher order quantity than revenue-sharing contracts, and the differences become more apparent over time. The rest of this paper is organized as follows. Section 2 reviews relevant literature. Section 3 provides the analytical background and establishes the research hypotheses for the study. Section 4 describes our experimental design and implementations. Observational results are reported in Section 5. Section 6 concludes the study and discusses our research limitations and opportunities for future work.

We present a laboratory study to examine strategic behaviors developed over repeated contract negotiations in the supply chain. We propose an experimental design that controls for individual decision biases and allows direct comparison with results reported in the literature. The supply chain contracts examined are the wholesale price contract, the buyback contract and the revenue-sharing contract. We find that the behaviors of both the retailer and the supplier deviate from the predictions of the traditional contracting model, which assumes a one-shot interaction between self-interested agents. Moreover, we observe strategic and reciprocal behaviors developed dynamically in the experiments. Motivations for these actions can be two-fold. The decision maker may have a generic preference for equitable payoffs. Meanwhile, due to future opportunities to punish in our experiments, a self-interested player may benefit economically in the long run by imitating behaviors that indicate concerns for fairness and reciprocity. Both types of rationales produce behaviors that enhance the supply chain performance. The simple linear contract, despite its high rejection rate, produces higher channel efficiency than predicted. This finding may help explain its wide applications in many business settings, including the software, pharmaceutical and consumer goods industries (Cui et al., 2007). The buyback contract and the revenue-sharing contract approximate coordination by balancing both the benefit and associated risk. Between these two mathematically-equivalent contracts, the buyback contract outperforms with fewer cases of disagreement and, on average, higher order quantities induced over time. We speculate that this difference has something to do with how the risk-sharing effect is “framed” under each contract. Although a few recent behavioral studies (Cui et al., 2007 and Katok and Pavlov, 2009) have embarked on extending contracting theories to include concerns for fairness, the analysis has focused on relatively simple contracts with deterministic demands. Our research points out several directions for future analytical work. For example, models can be developed to include fairness preferences over relative demand risk or profit variability under a stochastic environment, to incorporate strategic behaviors such as reputation building or reciprocity that arise from repeated exchanges, and to generalize behavioral discussions to more complicated supply chain contracts. We believe that the current research has provided several useful new insights. However, it is also important to point out our limitations. First, our experimental design eliminates the possibility that decisions are motivated by individuals’ bounded rationalities. However, individual biases have been demonstrated to be relevant to contract decisions, and decision support technologies may not be readily available. We have shown, through the comparison with Katok and Wu (2009), that bounded rationality and social preference may not always affect contracting behavior in the same direction. Thus, it will be interesting to further investigate how these two categories of behavioral motivations interact when they are both present. Another limitation of our design is that we only offer punishment options to the retailer. Future research into the channel relationship may also include opportunities to reward. Second, we assume perfect information by displaying the profit divisions to both parties in the experiments. In reality, a retailer may not know a supplier's cost, and therefore could not compare the relative payoffs. Future study could investigate how robust our conclusions are if information asymmetry exists in the supply chain. However, there are some industries where transactions closely approximate a world of perfect information. For instance, an executive of a Japanese auto supplier reported in an interview by Dyer (1997): “It would be virtually impossible for us to get away with inaccurate cost estimates. Nissan has much data; they have very good information on our operations and they can analyze our cost position. They can visit our plants and gather information” (Dyer, 1997, p. 546). Thus, repeated exchanges may make payoff information more transparent among supply chain partners. Lastly, we analyze a simple structure of supply chain with one retailer and one supplier. Our results imply that various contracts can perform differently based upon how the bargaining power is distributed within a channel. One way to directly test this conjecture is to introduce competitions. For example, experiments can be designed with multiple retailers and one supplier (or vice versa) in the same channel. In addition, while we observe behavioral differences between the two mathematically equivalent contracts in the laboratory, these contracts are implemented in quite different industries. The buyback contract is often adopted by the publishing and pharmaceutical industries (Padmanabhan and Png, 1995); whereas revenue-sharing contracts have been largely applied in the video rental industry (Mortimer, 2004). Whether or not the behavioral difference will persist in and account for field practices remains an open question. Future experiments may consider other operational factors such as the administrative requirements of different contracts. Nevertheless, we would like to note that the ultimate validation of the laboratory results will demand field studies. We hope that this research has contributed by moving a step further to bridge the gap between theory and practice.