Keywords

1 Introduction

Supply chain management is essential for successful business models. Previous studies have proposed several methods to improve supply chain performance. For example, Bowersox et al. [1] presented seven ways to maintain supply chain operating, including integrated behaviors, mutually sharing of information, mutually sharing of risks and rewards, cooperation, same goal and target customers, integrating processes and building of long-term relationships.

Among the mentioned methods, trust is critical to the stability and performance of a supply chain. With trust developing among supply chain partnerships, mutual communications between members could be much easier and more efficient. In this way, it increases the transparency of a supply chain and strengthens the relationships among members. Therefore, many business models are designed based on trust, particularly for the supply chain management. The Toyota Production System (TPS) is one of the best practice examples for lean production as it integrates production systems, and maintains relationships.

In addition to the production processes, the TPS model highlights the importance of integration. In other words, the model could ensure the accessibility of relevant information and develop the trust by the understanding of members. As trust is one of the keys to the performance of supply chain management, effective mutual communications would be a vital approach to establish.

The purpose of this study is to investigate members’ trust and trustworthiness through experiment in a three-echelon supply chain, which includes a supplier, a wholesaler, and a retailer. The bullwhip effect, trust diffusion, and suppliers’ production adjustment are considered in this study to be the performance of the supply chain.

2 Literature Review

2.1 Information Sharing and Trust

Companies invested considerable resources in improving supply chain visibility and collaborating with partners to be beneficial in supply chain management. While focusing on the bullwhip effect, studies showed that developing favorable relationships through partnership could mitigate its adverse impact for all members in the supply chain, such as suppliers and retailers [2, 3].

Comparing forecast information sharing and non-sharing operational systems, Ali et al. [4] found that sharing information was much more effective. Moreover, the trust could be established as a result. They further pointed out that partners’ credit was related to the level that they trust each other, which meant that high information sharing quality guaranteed trust level and strengthened members’ collaboration within the supply chain. Moreover, Simatupang et al. [5] reported that it would be hard to collaborate without a trust-based relationship in the supply chain. Thus, to build trust relationships, minimizing the uncertainty on demand forecasting was helpful, which kept the information quality by information sharing [6].

Because information sharing would be suitable for trust development and collaboration in a supply chain, the reliable information about demand forecasting was associated with both processes and performances. Besides, trustworthiness indicated information sharing quality. Members would distinguish whether the information could be trusted or not based on trustworthiness. Hence, trustworthiness and trust were to be considered in supply chain management.

2.2 Information Sharing and the Bullwhip Effect

A supply chain became unstable and inefficient under the bullwhip effect. Lee et al. [2] proposed four causes of the bullwhip effect, demand forecast updating, order batching, price fluctuation, and rationing and shortage gaming. Accurately forecasting demand by updating information was the most critical for reducing the bullwhip effect among four causes [8].

Therefore, researchers put their effort into lowering the uncertainty via the latest information and then enhancing the operation of information sharing to ensure the demand forecasting quality. Seferlis et al. [9] developed a demand prediction model to eliminate the uncertainty in inventory control strategies with the autoregressive integrated moving average, which increased information quality. Moreover, Li et al. [10] reduced uncertainty probability from the aspect of operational causes, including lead time delays, demand, production processes, and inventory processes. In terms of the simple two-echelon supply chain, Tanweer et al. [11] improved effectiveness by increasing supply chain transparency and demand forecasting accuracy.

Other techniques such as collaborative planning, vendor-managed inventory, sales and operations planning were introduced based on the concept of transparency improvement on information sharing and coordination in a supply chain. Also, immediately updating the information about product availability was useful to correct and adjust demand and ordering policy, reliving the impact due to the bullwhip effect and other adverse outcomes [7].

In addition to determining the reasons for the bullwhip effect, evaluating the bullwhip effect was the way to observe the performance of a supply chain. Dejonckheere et al. [12] measured the bullwhip effect in an order-up-to policy by the variance amplification of orders from a control-theoretic approach. Another evaluation method was to calculate with a coefficient of variance with input (order placing) and output (order receiving) quantity, which provided an early indication of the bullwhip effect [13].

2.3 Experiments About Trust and Supply Chain

Trust game [14] and investment game [15] were simulated to measure the spontaneous trust in several fields, especially in financial and supply chain management. The repeated interaction in games was to examine the pattern of trust increasing and decreasing during the focusing period. Consequently, a long-term relationship could be noticed as the period went on.

Since the trust was the key to maintain relationships and performances [16], conducting experiments with trust and supply chain could well simulate the real situation. Özer et al. [17] investigated the information sharing of demand forecasting in a two-echelon supply chain, which involved a retailer and a supplier. Their experiments showed the demand forecast and production decision and measured trust and trustworthiness according to subjects’ decision behaviors.

The beer game was introduced to demonstrate the bullwhip effect by considering market demand, ordering policy, and cost in trials. Except for downstream retailers, other members in the supply chain had little market knowledge. Nevertheless, the real supply chain was more dynamic and complex, making the simulation’s performance weak and unrealistic; other implicit issues were adopted to improve the investigation [18]. For example, increasing supply chain members’ motivations and incentives could facilitate excellent performance and make simulation realistic.

In light of the previous studies, this study combined trust and supply chain management in the experiment to learn trust between members and the bullwhip effect in a supply chain.

2.4 Hypotheses

This study aimed to observe trust and trustworthiness in a supply chain from three aspects. First, in terms of information sharing, the comprehension of other members could be an effective way of sharing information and enhancing performance by mitigating the bullwhip effect. Second, for trust diffusion, the trust relationships between partners could indicate a change of trust in different situations. Third, this study also interested in the suppliers’ behaviors for production adjustment since they lacked market information. Experiments would be designed to highlight people’s behaviors, such as trust and trustworthiness in a supply chain. And there were three hypotheses proposed below.

  • Hypothesis 1. With adequate information sharing on the trustworthiness of partners, the bullwhip effect could be reduced in both high and low trustworthiness cases.

  • Hypothesis 2. Trust diffusion is significant when partners in a supply chain have high trustworthiness.

  • Hypothesis 3. Suppliers tend to adjust their production strategy when cooperating with low trustworthiness partners to minimize risk.

3 Methodology

3.1 Experimental Design

The experiment in this study was based on researches about information sharing in the supply chain [17, 19] and the beer game. This study designed a three-echelon supply chain, which included a retailer, a wholesaler, and a supplier. There were two transactions in the experiment: one deal occurred between retailer and wholesaler, and the other was between wholesaler and supplier. Participants were assigned as retailers and suppliers randomly, while the computer program was set to be wholesalers.

This study divided the experiment into two tasks. The first one was the baseline of the initial trustworthiness and trust without the understanding of others, showing behaviors when there was no information sharing in a supply chain. In the second task, information sharing existed for learning about partners. Wholesalers with high or low trustworthiness scenarios were given in this case to control the uncertainty in the experiment.

For the high trustworthiness, the wholesaler in the experiment always delivered an unmodified retailer’s demand forecast report to the supplier. For the low trustworthiness levels, the wholesaler would inflate the number of reports received from the retailer, which was set as twenty in the experiment.

In terms of numerical setting, a retailer could forecast demand with firsthand market demand information ξ, while a wholesaler and supplier could merely know the demand distribution followed by uniform distribution on the interval [100, 400] and the market uncertainty also followed by uniform distribution on the interval [−75, 75].

Moreover, the retailer had to report a forecast demand \( \widehat{{\upxi_{1} }} \) to the wholesaler, and then the wholesaler would place the demand forecasting report \( \widehat{{\upxi_{2} }} \) to the supplier after receiving the retailer’s report. For the number of physical goods, \( {\text{Q}}_{1} \) units meant the wholesaler’s selling quantity, and \( {\text{Q}}_{2} \) units denoted the suppliers’ production quantity. Market demand was formulated as D = ξ + ∈, and ∈ symbolized the market uncertainty. The actual amount of retailer sales was indicated with \( { \hbox{min} }\left( {{\text{D}},{\text{Q}}_{1} } \right) \), and the actual amount of supplier sales was shown with \( { \hbox{min} }\left( {{\text{Q}}_{1} ,{\text{Q}}_{2} } \right) \).

3.2 Experiment

Four participants were in a group for an experiment, and they were paired randomly to avoid partners having preconceived impressions on others. The experiment was realized with z-tree and z-leaf [20]. Additionally, two tasks were involved in an experiment.

Researchers would explain the experiment in the beginning, and then have participants sign consent forms. In the first part of the experiment, participants were asked to complete task 1 by being suppliers and entering production quantity according to the given report from wholesalers for ten trials. The second part of the experiment was for task 2, involving thirty trials. Participants were asked to be either suppliers or retailers. As suppliers, they had to design strategy and decide on production quantity. For retailers, they had to input the demand forecasting amount to the upstream member, the wholesaler, by considering their firsthand market demand information and wholesalers’ behaviors.

3.3 Measurement

Due to retailers’ proximity to the market, they could evaluate market demand more precisely based on their understanding of customer and market trends. With the information in mind, retailers then created a forecast report for wholesalers. To ensure sufficient supply for customers, retailers tended to inflate the data on their forecast reports. The retailers’ inflation depended on two main factors: trust between retailers and wholesalers and the trustworthiness of wholesalers. In terms of suppliers, since they had minimal information about market demand and retailers’ forecast reports, they could only manage production with the wholesalers’ forecast report. Thus, trust and trustworthiness mattered for suppliers. Accordingly, adverse effects, such as the bullwhip effect, would occur under different levels of trust and trustworthiness.

The information sharing between downstream retailers and upstream suppliers was related to the level of trustworthiness and trust [7, 19]. For retailers, low trustworthiness referred to the amount of inflation in the forecast report. On the other hand, retailers had high trustworthiness when they reported the same amount of their firsthand forecast demand. Thus, retailers’ trustworthiness was measured by \( \widehat{{\upxi_{1} }} -\upxi \). The trust between retailers and wholesalers was determined by \( {\text{Q}}_{1} - \widehat{{{\upxi }_{1} }} \). The value would approach zero when the high trust relationship existed and would be more than zero otherwise.

For suppliers, the trustworthiness was examined with the difference in production and the forecast demand from wholesalers, \( {\text{Q}}_{2} - \widehat{{{\upxi }_{2} }} \). Because suppliers had limited information on the downstream, their production strategies relied on the forecast demand report from wholesalers. High trustworthiness meant that their behaviors were consistent with the forecast demand, although some adjustments might be necessary, while low trustworthiness was the opposite. As for the trust between suppliers and wholesalers, the differences between actual supplier sales and forecast demand reports from wholesalers were calculated. A high-trust relationship was shown by the value equal to zero. In contrast, the low-trust relationship was represented by the positive value, and trust would become lower as the value higher.

As suppliers could only receive little market information, observing impacts of the information sharing of partners’ trustworthiness provided methods to improve the performance of a supply chain. Therefore, this study focused on production adjustment in two tasks, which was evaluated with the differences in trust values in both tasks.

Besides, this study investigated how the bullwhip effect was influenced by trustworthiness and trust in a supply chain. In order to measure the bullwhip effect, \( \upomega_{i} \) was the notation of the bullwhip effect ratio. The mean and standard deviation of outgoing orders (\( \widehat{{\upxi_{i} }} \)), and incoming demands (\( {\text{Q}}_{i} \)) for \( {\text{i}} = 1,2 \), in a specific period (from time \( {\text{t}} \) to \( {\text{t }} + {\text{T}} \)) were used to calculate the coefficient of variation. Then the coefficient of variation of outgoing orders was divided by the coefficient of variation of incoming demands to get \( \upomega_{i} \). The formula was shown below:

$$ \omega_{i} = \frac{{c_{out} }}{{c_{in} }}\;{\text{where}}\;c_{in} = \frac{{\sigma \left( {\widehat{{{\upxi }_{i} }}\left( {t,t + T} \right)} \right)}}{{\mu \left( {\widehat{{{\upxi }_{i} }}\left( {t,t + T} \right)} \right)}}\;\text{and}\;c_{out} = \frac{{\sigma \left( {Q_{i} \left( {t,t + T} \right)} \right)}}{{\mu \left( {Q_{i} \left( {t,t + T} \right)} \right)}} $$
(1)

3.4 Participants

There were 36 participants aged from 20 to 26 involved in this study (mean = 22.89, standard deviation = 1.26). All of them were either undergraduate or graduated students. They were paired in the high or low trustworthiness group and randomly assigned as retailers or suppliers by the researchers.

4 Results

4.1 Trustworthiness and the Bullwhip Effect

Fig. 1.
figure 1

The bullwhip effect in high and low trustworthiness

Full size image

Hypothesis 1 stated that the trustworthiness of partners could reduce the bullwhip effect whether their trustworthiness level was high or low. This study calculated the bullwhip effect based on the input forecast report and output sales to display the fluctuation of information sharing biases in a supply chain, which could lead to the bullwhip effect and other negative influences in a supply chain.

According to the bullwhip effect indicator in this study, the value 1 was the threshold and implied that the bullwhip effect was slight when the value was approaching 1. Comparing both trustworthiness levels, the bullwhip effect was lower in a high-level case (mean = 1.06, standard deviation = 0.06) than in a low-level case (mean = 1.11, standard deviation = 0.10). In other words, the results pointed out that the bullwhip effect was worse in low trustworthiness case despite members being given their partners’ trustworthiness levels. Thus, Hypothesis 1 was not supported by the results.

Despite that, results indicated the importance of the information sharing quality. High trustworthiness could lead to a lower bullwhip effect since the provided information was reliable. On the other hand, it inferred that information delivered to or provided by low trustworthiness partners was suspected by the majority of participants, which caused the unstable information sharing quality, and then worsen the bullwhip effect (see Fig. 1).

4.2 Trust Diffusion

Fig. 2.
figure 2

Trust in each period

Full size image

Hypothesis 2 could be observed by comparing trust in high and low trustworthiness cases. This study calculated the average trust of each period to examine trust diffusion. The trust decreased as the value lower. The pattern in Fig. 2 showed how trust varied from period to period.

The results had three implications. First, it indicated that both patterns in the high trustworthiness case were consistent. Namely, the trust between one of them was related to the trust of the other. Second, Fig. 2 presented that the average trust between retailers and wholesalers (mean = −15.22, standard deviation = 7.54) was higher than suppliers and wholesalers (mean = −25.93, standard deviation = 8.70) in a low trustworthiness case. This result proved that trust levels were lower while interacting with upstream, who were suppliers and wholesalers. Third, results implied similar trends among periods in both cases, which might imply that trust between supplier and wholesaler changed with the retailer and wholesaler.

Based on three implications, trust diffusion was in both high and low trustworthiness cases, especially for high trustworthiness. Therefore, Hypothesis 2 was supported. However, the results further indicated that trust in upstream was smaller and more varied, making long-term and trust relationships challenging to develop and maintain even though trust diffusion might help increase trust level. Thus, this study reported that trust diffusion in a supply chain with low trustworthiness partners might lead to trust fluctuation, particularly for upstream partners who lack market knowledge.

4.3 Suppliers’ Production Adjustment

Fig. 3.
figure 3

Production adjustment of supplier and wholesaler

Full size image

Hypothesis 3 examined the suppliers’ production adjustment before and after understanding the trustworthiness level of partners. Since production strategies were made with a wholesalers’ forecast report, trust should be taken into consideration. Changes in suppliers’ production strategies between task 1 and task 2 would measure with trust adjustment in the present study.

Because suppliers could receive less market information, making an accurate demand forecast was difficult. More details from downstream partners would be helpful for scheduling production strategies and enhancing overall performances. Figure 3 displayed the trust adjustment of supplier and wholesaler. The higher the production adjustment, the more trust was between suppliers and wholesalers.

In terms of high trustworthiness, 6 out of 9 groups of suppliers increased their trust in wholesalers (mean = 19.30, standard deviation = 29.1). As for low trustworthiness, 8 out of 9 groups of suppliers decreased trust in wholesalers (mean = −19.22, standard deviation = 19.73). Their differences reported that participants would distrust partners with low trustworthiness and be more trust in those with high trustworthiness. It indicated from two points, one was risk tolerance, and the other was the development of trust relationship.

For the risk tolerance, participants, who were suppliers, tended to be risk-averse while cooperating with low trustworthiness partners. Since they believed that wholesalers would modify the demand report to a higher number and ensure an ample supply of goods, they could protect themselves from the loss of uncertainty in this way. For the trust relationship, participants were likely to turn to distrust partners in low trustworthiness that trust in partners with high trustworthiness. Namely, trust became fragile and hard to be mended in a short time in supply chain management with low trustworthiness.

5 Conclusions

This study investigates three aspects related to trust and trustworthiness in supply chain management, including the bullwhip effect, trust diffusion, and suppliers’ production adjustment. There are three findings: first, the bullwhip effect would be mitigated to a lower level when partners in a supply chain had high trustworthiness. Since the bullwhip effect directly influences the performance of a supply chain, cooperating with high trustworthiness partners enhances the information sharing quality. Second, the trust diffuses in both high and low levels of trustworthiness supply chain, which helps build long-term and trust relationships. However, trust diffusion in the upstream takes time due to a lack of market information. Third, suppliers are more likely to adjust their production strategies to cooperate with low trustworthiness partners with minimum risks, which potentially leads to a break in trust relationships.

To sum up, this study suggests that trustworthiness be vital information in a supply chain for enhancing performances, especially for upstream members. Additionally, suppliers behave more cautiously in low trustworthiness rather than high trustworthiness supply chains, which presents risk aversion and the tendency of uncertainty avoidance. Since participants in this study have similar cultural backgrounds, their decision-making procedures, and tolerance of risk and uncertainty may be different from those of other cultures. This study recommends that cross-cultural studies be conducted in future research and that international supply chain management will benefit from it.