Mutual Diligence Business In Supply Chains - A Dynamics Business System Approach

Abstract In this paper, we have considered the multiple stage supply chain where agents order, produce and distribute goods while considering uncertainties of customers’ behavior. Generally, those operational risks result in undesirable effects of the supply chain, e.g. bullwhip effects. We model the problem in terms of system dynamics which seems adequate for a process-oriented perspective where time is a decisive factor. The original model developed by Forrester (1958) is extended by integrating data warehouse architectures which make the collaborative processes between the agents more handy. Key figures of the analysis are the bullwhip effect, the maximum production rate and imbalance period. Some supply chain models are tested by simulation runs where the parameters are adapted to different scenarios. It is shown that, generally, a reduction of the delivery time has a greater influence on the performance of the supply chain than the improvement of mutual trust between the partners.

Introduction:  Supply Chain Management is one of the hybrid management concepts between market and hierarchy. The big aim of this concept is to achieve the advantages of cooperation without giving up the advantages of autonomy. Hence the reduction of production costs must exceed the increase of transaction costs.
Transaction costs mainly consist of information costs that occur due to a complex transmission of information. Indirect information costs result from inefficient processes caused by information distortion between the supply chain partners.
In this article we first characterize global supply chains and then derive different scenarios of regional and continental supply chains with their typical information systems. Finally, we compare these scenarios by a system dynamics approach.
Global Working Supply Chains: Global supply chains, unlike regional supply chains – can be described by several characters. First of all, far distances between the supply chain partners determine long delivery times that require a consideration of the inventory in transit (supply line). It has to be considered that the supply line is relatively non-volatile as goods remain between two partners of the supply chain for a long period.  All processes in the supply chain are rather inflexible for these reasons. Second, there exists a higher level of information distortion in a global supply chain.
that leads to the bullwhip effect (see e.g. Lee, Padmanabhan, Whang (1997), pp. 95). Two reasons can be stated for the information distortion: On the one hand, the lack of personal contact between the employees of global supply chains can cause a lack of trust. This means that less

information will be exchanged between the supply chain partners and no mature information system that includes collaborative planning will be implemented. On the other hand, the transmission of information may take more time than in regional supply chains.Last but not least, we have to distinguish which supply chain member is situated in which distance. In one case, only the producer is situated far away from the other members and the market. In the other case, all members are situated away from each other and only the retailer is located near the customer.
Data Warehouse Architecture: Without having implemented a data warehouse, the impulse of all activity in a supply chain is given by the order of the customer. This order is fulfilled from the inventory of the retailer, afterwards an order to refill his inventory is made. The quantity of that order is calculated by the order-up-to-policy described by Sterman (1989), p. 333. The other supply chain members act corresponding to this order rule, no other information is exchanged between the supply chain members. As all distortions in the system are accumulated a strong bullwhip effect is expected.
The data warehouse concept was designed by Inmon (1996) as a database that explicitly supports strategic decisions (see Kimball et al (1998), pp. 9). Without a data warehouse data is usually located in different information systems. For that reason, on the one hand, data is unstructured and not integrated, much data is stored several times on different places and much data that is not necessary for any decision is stored (information overload). On the other hand, data that is needed for a decision cannot be found easily (lack of information). This problem can be reduced by the data warehouse concept as all kinds of information can be stored and classified. Direct access is possible to all information if requested.
In the last decade, the concept was extended towards a database that provides all processes of the whole enterprise (or, here, the whole supply line) with timely, detailed and integrated information as described in Inmon, Imhoff, Battas (1996, pp. 12). Complex analysis tools that enable a collaborative business are usually included in a data warehouse to pre calculate all kind of measures that are needed to control the enterprise. Due to the integrated database all measures are free of redundancies and inconsistencies. Figure 1 illustrates the structure of a data warehouse.
A data warehouse includes a central storage of the data as well as a central analysis tool that enables collaborative planning. It is a complex information system that causes a high level of investments and maintenance costs. In a data warehouse the reasons for the delays and distortions are partly eliminated, as the customer’s orders are stored in the data warehouse and the future customer’s demand is forecasted there. These forecasts of the demand are used by every supply chain member to calculate his order quantities or (incase of the producer) production quantities. As the implementation of a data warehouse requires a high level of trust and cooperation it has to be considered which kind of a data warehouse will be implemented in which supply chain.
Scenarios of Different Supply Chains Using Data Warehouses: A typical supply chain that consists of three members (retailer, distributor and producer) is modeled in this article. Orders that are not forwarded by the data warehouse take 1 day to arrive at the next member of the supply chain. The delivery time between two members amounts to 3 days if the members are situated nearby and increases to 5 days if the partners are situated further apart. The production of an order takes 5 days, no capacity constraints are modeled here. According to the considerations above, three different scenarios can be modeled for such a supply chain

In a regional supply chain, all members are situated in the same area. Thus, the delivery time between two chain members is only 3 days short and the information distortion remains low. The high level of mutual trust allows a data warehouse in which all members of the supply chain are integrated and the forecasts of the customer’s demand are immediately transferred to all members. All members are able to plan their order and production quantities, respectively, on the basis of timely and accurate information.

In a continental supply chain, the producer is situated in a further distance from the other supply chain members. The delivery of goods from the producer to the distributor will take 5 days for that reason. The delivery time between the distributor and the retailer is still 3 days. Another effect of the distance between the producer and the other supply chain members is a lower level of mutual trust. Hence the producer will not be integrated in the data warehouse architecture unlike in the regional supply chain and he gets the orders by a non standardized and conventional way of data exchange. As mentioned above, orders will take 1 day to be forwarded to the producer. As a consequence, the producer is not supported with useful information that well in this architecture as he has no access to the data warehouse. The transaction costs in this scenario are lower than in the first scenario because less company has to be integrated in the data warehouse architecture. Another advantage is that collaborative planning stays more manageable and flexible for the same reason.
In a global supply chain, all members of the supply chain are situated in different areas. The delivery time between the producer and the distributor and between the distributor and the retailer will increase to 5 days in this scenario. The lack of personal contact between all members effects that no data warehouse will be implemented at all. All order information is forwarded conventionally with a delay of 1 day.
According to the different scenarios, we will describe the corresponding cost effects. Additionally, we will test how the negative effects can be lowered especially in case of the continental and the global supply chains. Therefore two basic strategies of improving the original results are regarded in both scenarios:
•    Reduction of the delivery time between the partners that are situated further apart from 5 to 3 days or
•    Improvement of mutual trust between the supply chain partners that enables the implementation of a more mature data warehouse (i.e. a data Warehouse between retailer and distributor in a global supply chain and a data warehouse where all partners are collaborating in the continental supply chain).
Altogether, seven scenarios are interpreted: three basic scenarios and four improved scenarios. Transaction costs of improvement are not calculated.
Complex systems like the scenarios of an entire supply chain can be modeled with the system dynamics approach (see Forrester (1958)). A set of equations describes all processes of the supply chain so that all variables of all partners can be illustrated at any time. This approach is especially adequate for modeling a global supply chain as information delays and delivery times can explicitly be considered.
The described architectures are simulated with this system dynamics approach in which the system of equations is developed corresponding to Reese, Waage(2007).
Results: In the beginning of the simulation period, the customer orders 1000 units of a product per day, all the system is in balance. After one day, the orders of the customer increase by 10 % up to 1100 units per day. That quantity of orders remains unchanged until the end of the simulation period after 500 days (i. e.nearly 1.5 years).
•    The described shock leads to an imbalance that causes a bullwhip effect. According to Sterman (1989) the bullwhip effect is calculated as the ratio between the variance of production rate of the producer and the customer’s orders.
•    Another important measure to demonstrate the negative consequences of the system shock is the maximum production rate because it determines the necessary capacity at the producer’s in order to avoid stock outs. Capacity is one of the largest cost factors in each production system.

•    A third measure is the period of imbalance (tolerance +/- 10 %). This measure excellently shows how a system can handle disturbances and how it can counteract against these disturbances. Every imbalance of a system requires a special degree of flexibility and thus more capacity than a balanced system.


    Reference
scenario    Delivery time reduction    Improvement of mutual trust
regional    27.91    -    -
continental    86.67    47.97    50.44
global    217.78    113.26    184.39
Table 1: Bullwhip effect
Table 1 visualizes the increase of the bullwhip effect when the supply chain partners move apart. In a regional supply chain, the bullwhip effect is nearly 28, i.e. the variance of the production rate is 28 times higher than the variance of the customer’s rate of orders. In the continental supply chain, the bullwhip effect increases up to nearly 87. Thereby it is 2.5 times as high as in the regional supply chain. In a global supply chain, the bullwhip effect is even 8 times as high as in the regional supply chain. Hence increasing the delivery time as well as establishing a less mature information system both have a clear influence on the bullwhip effect.
Table 1 also reveals that the bullwhip effect in a continental supply chain can be equally decreased by nearly 50 % by reducing the delivery time or extending the data warehouse architecture towards the producer. In the case of a global supply, chain the delivery time reduction again shows a large influence on the bullwhip effect (reduction of about 50 %). However, the implementation of a more mature data warehouse in this scenario has a much smaller effect (decrease by about 15 %). Consequently, a data warehouse architecture comprising of the distributor and the retailer is less promising when considering a global supply chain. More efforts have to be taken into account though it is very costly to also integrate the producer in the architecture.

    Reference
scenario    Delivery time reduction    Improvement of mutual trust
regional    1890.10    -    -
continental    2896.66    2330.87    2193.64
global    5184.93    3617.26    4620.11
Table 2: Maximum Production Rate
In the reference scenario of the regional supply chain, the maximum production rate climbs up to nearly 1900 items per day (see Table 2) although the customer’s demand is only raised by 10 % from 1000 to 1100 items. This high amplitude is already done a few days after the external order shock occurs.

In the order policy the full consequences of the order shock are only realized after some days so that high adjustments are inevitable. The maximum production Rate in the continental supply chain still exceeds those values by more than 50%.

In a global supply chain, the maximum production rate reaches its worst value after another increase by 1.8 as compared with the continental chain. Nevertheless, the impact of globalization on this measure is not as high as the impact on the bullwhip effect. As the costs of capacity are directly concerned the maximum production rate can be easily interpreted.

The four improved scenarios show similar results like the analysis of the bullwhip effect. In a continental supply chain, a delivery time reduction and an improvement of mutual trust that leads to a more mature data warehouse have nearly the same impact on the maximum production rate (decrease of around 20%). In a global supply chain, the impact of a delivery time reduction is again much higher than the impact of a more comprehensive data warehouse (30 % vs. 10 %). One may argue that is a simple, robust criterion to explain the bullwhip effect.
 


    Reference
scenario    Delivery time reduction    Improvement of mutual trust
regional    52    -    -
continental    155    64    124
global    348    184    305
Table 3: Period of Imbalance
Finally, regarding the period of imbalance, the results of the reference scenarios again show that the physical distance of the supply chain members has a great influence on the performance of the supply chain (see Table 3). For a continental supply chain it takes 3 times as long as for the regional supply chain to re-establish the balance and for the global supply chain the corresponding value is 7 times as large. In the latter case, a shock of 10 % more orders thus keeps the whole system for approximately 1 year unbalanced.
In contrast to the other measures, the period of imbalance generally produces a higher effect of a delivery time reduction in the continental supply chain as well as in the global supply chain. In the continental supply chain, the period of imbalance can be decreased by nearly 60% when reducing the delivery time. A more mature data warehouse only reduces the measure by 20 %. Similar results can be achieved regarding the global supply chain. As a result, the delivery time reduction has a larger positive influence on the duration of a disturbance than on its amplitude.
Conclusion: The simulation study showed that globalization may have a considerable negative impact on the performance of a supply chain. The expected benefits of globalization must be carefully weighed against the costs which are caused by appropriate efforts. The scenarios in which additional measures were taken to mitigate the negative effects of global supply chains lead to the overall result that a reduction of the delivery time may improve the efficiency of supply chains better than a more mature data warehouse. The difficulty about this comparison is that the corresponding costs of these measures can hardly be compared because they were determined by many technical factors. For that reason, no statement can be given if a delivery time reduction will always lead to higher profits than measures of improving mutual trust in a supply chain.
References:
1.    Du T C, Wong J, Lee M (2004), Designing Data Warehouses for Supply Chain Management, Proceedings of the IEEE International Conference on E-Commerce Technology, San Diego, 170-177.
2.    Forrester J W (1958), Industrial Dynamics - A Major Breakthrough for Decision Makers, Harvard Business Review, 36, 37-66.
3.    Inmon W H, Imhoff C; Battas G (1996), Building the Operational Data Store, New York et al.
4.    Inmon, W H (1996), Building the Data Warehouse, 2nd edition, Indianapolis.
5.    Kimball R et. al. (1998), The Data Warehouse Lifecycle Toolkit, New York et. al.
6.    Lee H L, Padmanabhan V, Whang S (1997), The Bullwhip Effect in Supply Chains, Sloan Management Review, 38, 93-102.