Transportation Research Part E: Logistics and Transportation Review
The impact of electronic data interchange on reducing bullwhip effect and supply chain inventory costs
Introduction
The aggregated behavior of supply chains has been investigated by many authors (e.g., Mitchell, 1923; Hansen, 1951; Forrester, 1958, Forrester, 1961; Mitchell, 1971; Zarnowitz, 1973; Blanchard, 1983; Sterman, 1989; Krane and Brawn, 1991). Their research shows three typical behavioral patterns which characterize the distortion of demand moving upstream in the chain (retailers/wholesalers/distributors/factory): oscillation, amplification and phase lag. This distortion in demand is known as the bullwhip effect (see Senge, 1990; Blackburn, 1991; Hammond, 1994; Lee et al., 1997a; Holmström, 1997; Frangoo and Wooters, 2000). This distortion produces the same patterns on the inventories throughout all the elements of the supply chain, as illustrated in Fig. 1, Fig. 2: (a) the orders and inventory demonstrate large amplitude-fluctuations at the different nodes in the supply chain (oscillation); (b) a gradual increase in variance across all the elements in the chain (amplification1); and (c) after a certain delay, the peak of orders placed, which commences at the retailer, extends to the rest of the components further upstream (phase lag). The same occurs with the appearance and later disappearance of inventory shortage (Fig. 2).
A small variance in actual consumer demand can result in the various companies operating at different stages of a supply chain being subject to the bullwhip effect, which gives rise to a number of well-known problems: the accumulation of excess inventories at certain times, followed by serious inventory shortages; poor customer service at other times; excess or insufficient capacity, depending on the case; and unstable production and inefficient production planning and scheduling, leading to higher costs resulting from the corrective actions that have to be taken (investments in new capacity, overtime working, deliveries by urgent transport, etc.).2
A number of studies have investigated the various causes of the bullwhip effect (see Forrester (1961); Kahn, 1987; Sterman, 1989; Eichenbaum, 1989; Blackburn, 1991; Naish, 1994; Diehl and Sterman, 1995; Towill, 1996; Lee et al., 1997b). A number of these authors highlight time delays between supply chain links (Sterman, 1989; Blackburn, 1991; Diehl and Sterman, 1995; Metters, 1997; Chen, 1999). For example, Metters (1997, p. 99) stresses that “a lack of inter-company communications combined with large time-lags between receipt and transmittal of information are at the root of the problem”. He also stresses that possible solutions, such as Electronic Data Interchange (EDI) for cutting lead times, are expensive, and managers need to justify these expenses. Although time delays can be susbstantial,3 they are frequently misperceived (Sterman, 1989). On the other hand, the benefits of EDI are hard to perceive (Jiménez-Martı́nez and Polo Redondo, in press) and unevenly distributed amongst trading partners (Nakayama, 2003). Furthermore, purchasing and supply management literature reveals an inconsistent and unclear picture of the impact EDI has, producing a controversy which needs further research (Larson and Kulchitsky, 2000).
This study investigates via simulation the impact of EDI on the supply chain both as a whole, and also on each of the chain’s individual nodes. The working hypothesis is that: As the use of EDI drastically reduces information delays, it should substantially reduce the bullwhip effect. This should also involve a reduction in related costs, both in the chain as a whole, and at each of the chain’s individual nodes.
In the following Sections we firstly discuss the use of simulators for performing supply chain experiments and justify their benefits in comparison with alternatives. We also comment on the role they can play in illustrating the possible benefits to be had from the use of EDI in the supply chain. Secondly, we describe the experiment and comment on its findings. We continue with a review of other articles that have used simulators in related studies, pointing out any differences between them and our own. We conclude with a section devoted to Final Remarks, including some limitations of the study and ideas for further research.
Section snippets
Methodology
The difficulty of empirically assessing the cost effects of the bullwhip effect has been stressed by Metters (1997), who points out that the relative contribution of the bullwhip effect versus other factors is unclear. He also comments on the difficulty of assessing these costs with analytical closed-form expressions. Apart from this, supply chain management will typically involve managers from three or more organizations, and getting them to work together and participate in a research project
Benefits from EDI on the supply chain
Using EDI for information transmission allows companies to save considerable amounts of time and money. In the EU, variable cost savings derived from the fact that exchanged documents are no longer manually written are estimated at 3.5–15% of the value of the product (Chip, 1993). Procter and Gamble have estimated that their order processing cost ranges between US$35–US$75 per invoice due to manual intervention (Lee et al., 1997a), which explains why companies often accumulate demand in batches
The supply chain simulator
For the reasons discussed in Section 2, we decided to create a computerized version of the Beer Game which minimizes the game’s deficiencies whilst preserving all its advantages, particularly the on-line interaction between the players. Thus was born the very first computerized network version of a supply chain via the Internet, programmed in C++ and finished in mid-1996. The two fundamental differences from the original game were: (a) the possibility of simulating the operation of the chain
Other related examples and experiments
There are other differences between our study and those of the other authors cited, in addition to the differences commented on in Section 4 regarding the tool that was used.
The only study that can be considered an experiment with measurement of the statistical significance of the results is Anderson and Morrice (2000), but this is much less exhaustive and, moreover, contains no reference to either EDI or the reduction of information delays. The authors’ basic objective was to determine how
Final remarks
In the preceding pages we have seen how computer simulation is a feasible and promising way to conduct statistically significant tests in supply chain management research. In most cases, such tests are difficult to carry out in real-life situations. We have shown the value of laboratory research and simulations in supply chain strategies that are difficult to implement through traditional methods. Proper control enables trials to be repeated under effectively constant conditions and also
Acknowledgements
We would like to thank the Editor, W.K. Talley, and two anonymous reviewers for the sterling work they have done in helping to improve this paper.
References (49)
- et al.
Electronic data interchange: how much competitive advantage?
Long Range Planning
(1990) - et al.
Effects of feedback complexity on dynamic decision making
Organizational Behavior and Human Decision Processes
(1995) - et al.
Strategic internet application trends in supply chain management
Industrial Marketing Management
(2003) Quantifying the bullwhip effect in supply chains
Journal of Operations Management
(1997)An assessment of EDI use and other channel communications on trading partner knowledge
Information and Management
(2003)Second coming for electronic data interchange
World Trade
(1997)- et al.
A simulation game for teaching service-oriented supply chain management: does information sharing help managers with service capacity decisions?
Production and Operations Management
(2000) - Bartholomeu, D., 1997. Clinging to EDI. Industry Week, June 23, pp....
The quick response movement in the apparel industry: a case study in time-compressing supply chains
The production and inventory behavior of the American automobile industry
Journal of Political Economy
(1983)