关键词: 公平偏好, 双渠道进货, 不完全信息, 行为实验

Abstract: The practice of dual sourcing is common among manufacturers and is considered an effective strategy to deal with the risk of supply disruptions, high costs, capacity limitations, and/or lead time variability. To negotiate better terms, firms often rely more on a primary supplier, resulting in uneven order allocations between suppliers in dual sourcing. However, suppliers may reject the order, that is, supply shortage can be caused by intentional refusal of the suppliers even though their supply capacity is sufficient to meet the order. Fairness concern is one of the major drivers for the observed order rejections. To mitigate this type of behavioral supply risk, we focus on the effect of order information on the fairness concern of suppliers.
To analyze the effect of order information on the fairness concern of dual-sourcing suppliers, this paper considers a supply chain consisting of a manufacturer and two suppliers. Two suppliers offer an incremental quantity discount policy. To benefit from discount, the manufacturer orders more from one supplier, which causes dissatisfaction of another supplier that receives a smaller order quantity. The small-order supplier’s concern of fairness may result in order rejection. We firststudy a normative model of order allocation game analytically and provide a theoretical benchmark. By designing and conducting behavioral experiments in laboratory, we compare the decision behaviors under two treatments. In one treatment, the order quantity is complete information, that is, each supplier can observe the order quantities that manufacturers allocated to both suppliers. In the other, the order quantity is private information, that is, each supplier only observes its own order quantity. We collect a total of 880 decisions for each role in complete information treatment and 1040 decisions in incomplete information treatment. We use the one-sample Wilcoxon signed-rank test to compare experimental results with theoretical predictions and the Wilcoxon-Mann-Whitney test to compare the two treatments’ experimental results. To explain the observations in experiment, we build behavioral models for both treatments. For the complete information treatment, we combine the ERC model and logit choice model. For the incomplete information treatment, we use perfect Bayesian equilibrium to analyze the decisions of manufacturers and suppliers. Maximum likelihood estimation is applied to estimate the parameters of the decision makers’ behavioral preferences.
Experimental data indicate that behavioral decisions of subjects deviate from the normative models’ predictions in both treatments. In experiment, the manufacturer allocates to the small-order supplier higher order quantities than the prediction, but these orders are frequently rejected. Comparing with complete information treatment, manufacturers allocate higher order quantities to the small-order supplier in incomplete information treatment, but the orders are rejected more often.All these experimental results are statistically significant. Identifying fairness concern as the main driver, our behavioral models and parameter estimation demonstrate that the suppliers show stronger fairness concern in the incomplete information treatment than in the complete information one. The behavioral models capture the subject’s decision process, and accurately predict the manufacturer’s order allocation and the supplier’s rejection behavior in experiment. This study shows that manufacturers should focus on the fairness concern of the small-order supplier to reduce order rejection. Keeping order information private as in the incomplete information treatment is not an effective way to mitigate the behavioral supply risk caused by fairness concern, and order information transparency is more effective.

Key words: fairness concern, dual sourcing, incomplete information, behavioral experiment