电子商务对动态供应链管理的影响评价(英文译文).doc

Assessing the impact of e-business on supply chain dynamics
S.M. Disney, M.M. Naim*, A. Potter
Logistics Systems Dynamics Group, Cardiff Business School, Cardiff University, Aberconway Building, Colum Drive,
Cardiff, CF 10 3EU, UK
Received 15 April 2002; accepted 18 November 2002
Abstract
The Internet and related information and communication technologies (ICT) have recently enabled the cost-effective dissemination of information between disparate parties in the supply chain. New supply chain strategies, such as vendor managed inventory (VMI), collaborative planning, forecasting and replenishment (CPFR) and efficient consumer response (ECR), have begun to exploit these new communication channels, principally at the retail end of the supply chain. The impact of the e-business enabled supply chain on manufacturers and materials/component suppliers is,however, less well understood and exploited. This paper is aimed at establishing e-business enabled supply chain models for quantifying the impact of ICT, in particular its effect on dynamic behaviour. The paper concludes that simple, yet robust, models enable considerable quantitative insights into the impact of e-business on supply chain dynamic behaviour prior to their implementation.
Keywords
Supplychaindynamics7e;Lommerce7Kullwhip7+ZQS7Vendormanagedinventory7e;Shopping
1. Introduction
While information and communication technologies (ICT) in the form of e-business is advocated as an enabler to the 1–2–1 enterprise (Peppers and Rogers, 1997) by allowing market place information to be shared by all businesses in the supply chain, there is little analytical or quantifiable evidence that it will actually improve the overall performance of the enterprise in delivering customer wants. It is usually proposed that passing information to all businesses in the supply chain via ICT will improve performance. In fact, recent research (Hong-Minh et al., 2000) has shown, via the supply chain ‘‘Beer Game’’ (Sterman, 1989), that simply passing information on to businesses can have a detrimental effect. This is due to the fact that, as well as having more information available, schedulers need to know what to do with it.
There are many ways in which innovative information flows could be used within supply chains. Kiely (1998) provides a good starting point, specifically focusing on using demand data for forecasting purposes. In this paper we analyse the impact of four ICT enabled scenarios by investigating the bullwhip effect (Lee et al.,1997a, b) using two different approaches and comparing them to a traditional supply chain. The first approach is based on an analysis of the results of a management flight simulator, the Beer Game. The second approach is based on a quantitative z-transform analysis using the tools highlighted by Disney and Towill (2002). The aim is to compare and contrast the two approaches qualitatively to assess the implications of their evaluations of e-business scenarios on supply chain dynamics.
Bullwhip is an important measure, being symptomatic of a poorly performing supply chain (Jones and Simons, 2000). It is a surrogate measure of production adaptation costs (Stalk and Hout, 1990) and implies the inclusion of ‘‘just-in-case’’ stock holding to buffer against uncertainties There is considerable empirical evidence of bullwhip including recent examples in the:
* food sector where the supplier orders two tiers further upstream varied 10 times more than the electronic point of sales (EPOS) data (Jones and Simons, 2000).
* automotive sector where the ratio of the variance between incoming orders and order to suppliers at just a single echelon in the supply chain was 1:2 (Naim et al., 2002).
The five supply chain strategies considered are:
* Traditional—in which there are four ‘‘serially linked’’ echelons in the supply chain.
* e-Shopping—where the distribution network is by-passed and information and materials flow directly between the end consumer and the product suppliers.
* Reduced—where an echelon in the supply chain had been removed.
* Vendor managed inventory (VMI)—that is simulated by developing a protocol positioned between two businesses in the supply chain that gives the necessary inventory and sales information, authority and responsibility to the supplier in order to manage the customer’s inventory.
* EPOS—where information from the market place is transmitted to all enterprises in the supply chain.
Although various e-business scenarios are available the above were chosen by four groups of four Masters Programme students based on their review of commonly quoted and/or implemented strategies in both the academic and practitioner literature. It was these Masters students who implemented the scenarios in the Beer Game.
2. Methodology
Research on improving the dynamic behaviour of individual manufacturing businesses and supply chains is well known. Most recent research methodologies may be categorised as:
* Management games: Tools such as the Beer Game that was developed at MIT at the end of the 1950s (Sterman, 1989), are useful to illustrate the benefits of different supply chain strategies. Games are limited in the sense that generally nothing can be rigorously proved from the game in itself, but they do provide a valuable source of anecdotal evidence and are a good learning device. Other authors have extended or computerised the Beer Game including van Ackere et al. (1993), Kaminsky and Simchi-Levi (1998), and Lambrecht and Dejonckheere (1999a, b).
* Empirical studies: A number of authors have investigated the impact of ICT on the supply chain including Holmstr .om (1998), Fransoo and Wouters (2000), and Kaipia et al. (2000). However, this type of contribution looks at quantifying the improvement performance of a known strategy after its implementation; that is, there is no predictive element and the focus of the research is to identify best practices. Unfortunately, it is not always possible to compare ICT implementation strategies directly due to the varying nature of the environments they have been implemented in.
* Statistical: This type of contribution typically provides statistical insights about the impact of demand properties such as standard deviation and correlation, and supply chain properties such as lead-times and information paths on inventory costs and the bullwhip effect orTdemand amplification. Statistical methods are often used to quantify the performance of real situations. These methods however, fail to show how to reduce or eliminate the detrimental dynamic effects, such as ‘‘bullwhip’’, and insights into the causes and effects of system structure on performance are rarely obtained in depth from the technique. Recent significant contributions of this type include Lee et al. (2000) and Chen et al., (2000).
* Simulation and system dynamics: This approach was advocated by Forrester (1961) as a method of investigating the dynamical effects in large non-linear systems without resorting to complicated mathematical control theory based models (Edghill and Towill, 1989). Simulation approaches alone suffer from being cumbersome, time consuming and only provide limited insight (Popplewell and Bonney, 1987), but they
do have the advantage of being able to model non-linearities whilst avoiding complicated mathematics. Previous work using simulation is very prolific and includes (but is no means limited to) Forrester (1961) and Coyle (1982), who studied traditional supply chain structures, Cachon and Fisher (1997) and Waller et al. (1999) who studied VMI.
* Continuous control theory techniques: The Nobel Prize for Economics Winner in 1978 Herbert Simon (for his work on organisational dynamics) (Simon, 1952) was the first to describe how to use linear deterministic control theory for production and inventory control. Axs.ater (1985) presents a useful review paper of early work, summarising the advantages and limitations of the field. He concludes that control theory ‘‘illustrates extremely well dynamical effects and feedback’’, but cannot incorporate sequencing and lot-sizing issues. Much research from the Department of Production Economics at Link.oping University in Sweden has been presented in the literature. They have been applying the Laplace transform and economic techniques such as Net Present Value to MRP systems (Grubbstr .om, 1967) Continuous control theory suffers from the fact that some scheduling and ordering scenarios are inherently discrete and the continuous representation of discrete time delays is mathematically complicated.
* Discrete control theory: is a very powerful way of investigating sampled data systems, that is scheduling and ordering systems or a computer system, all of which are inherently discrete. Vassian (1955), inspired by Simon’s work in the continuous domain, studied a production-scheduling algorithm using discrete control theory. DeWinter (1966), in possibly one of only two contributions that consider novel supply chain structures, looks at a form of centralized inventory control used in naval supply chains. Deziel and Eilon (1967) describe a significant application. Burns and Sivazlian (1978) consider a four level traditional supply chain using z-transforms. Bonney and Popplewell (1988) have investigated MRP systems. Dejonckheere et al. (2003a) have been using z-transforms to investigate the bullwhip performance of common forecasting mechanisms within common control structures. Disney (2001) has been using discrete control theory to investigate VIM supply chains. The disadvantages with discrete control theory are that the mathematics often involves lengthy algebraic manipulation.
The methodology utilised in this paper is to analyse the results from the playing of the Beer Game to determine the extent of bullwhip in the various ICT scenarios described. This allows direct comparison with previous published results (Hong-Minh et al., 2000) that yielded counterintuitive results. The Beer Game results encompass both structural and human behavioural aspects. The latter characteristics may include game players’ poor understanding of the game, nondeterministic decision-making and errors in transcribing orders from customers to suppliers. An analytical z-transform approach is then utilised for comparative purposes and to deduce the impact of deterministic feedback system structures on supply chain bullwhip. The analytical approach is particularly important where there is an expectation that ICT systems will handle the bulk of information transactions in the supply chain and human interference will be limited to managing exceptions.
3. Description of the five supply chain scenarios
The five supply chain scenarios researched are summarised by Fig 1. A short description of each scenario also follows.
3.1. Traditional
A traditional supply chain may be characterized by four ‘‘serially linked’’ echelons in a supply chain. Each echelon only receives information on local stock levels and sales. Each echelon then places an order onto its supplier based on local stock, sales and previous ‘‘orders placed but not yet received’’ (Sterman, 1989).
3.2. Reduced
By a reduced supply chain we mean a supply chain with a reduced number of echelons. This is representative of, say, the A supply chain, where the retailer echelon is by-passed in the information and material flow through the use of ICT. Echelon removal has been identified by Wikner et al. (1991) as an effective mechanism for improving supply chain dynamics.

3.3. e-Shopping
By the term e-shopping we refer to the scenario where the manufacturer receives orders directly from end consumers (possibly via the Internet like Dell) and ships the product directly to them after the production and distribution lead-time. Thus this supply chain strategy has exactly the same fundamental structure as a single-echelon traditional supply chain.
3.4. EPOS enabled
In the EPOS enabled scenario, the end consumer sales (CONS) are made visible to all members of the supply chain. This is equivalent to the situation in many grocery supply chains, where the data is available electronically via the Internet, either directly from the retailer or via a third party, and can be used by supply chain members to generate their own forecasts. Specifically, in this strategy the end CONS may be used by each echelon for their own planning purposes, but each echelon still has to deliver (if possible) what was ordered by their customer. A full-scale investigation of this strategy has been conducted using z-transforms by Dejonckheere et al. (2003b), inspired by the simulation approach of Mason-Jones (1998).
3.5. Vendor managed inventory
The specific VMI scenario that we consider is as follows. The distributor in a two-echelon VMI relationship manages the retailer’s stock. The distributor is given information on the retailer’s sales and stock levels. In this scenario the retailer does not place orders on the distributor, but instead trusts the distributor the dispatch adequate amounts of stock to ensure that there is enough (but too much) stock at the retailer. We use the VMI strategy highlighted by Disney et al. (2001) for the VMI echelons in the supply chain. The other echelons in this scenario (the Warehouse and Factory) were run in traditional mode.
4. Impact of ICTon supply chain dynamics
4.1. Beer Game results
Hong-Minh et al. (2000) analysed the results from four different teams playing different supply chain management strategies, one of which was the EPOS scenario previously described. Even though the research literature implies great benefits for information sharing (Mason-Jones and Towill, 1997), surprisingly the EPOS strategy yielded the worst result. Whilst the EPOS strategy limited the degree of bullwhip in the supply chain, this was at the expense of long periods of backlogs (negative net stock). It was concluded that, although market information was shared with all echelons without any delays, each player of the supply chain had their own ordering rule. That is, there was no collaboration between the different players. To test the hypothesis that although sharing market information is potentially a good thing it will only yield benefits as part of an agreed overall supply chain decision making strategy (Mason-Jones, 1998), the EPOS strategy was re-run but with the added characteristic that all players were involved in collaborative planning, forecasting and replenishment (CPFR).
The Beer Game was also run twice by the current authors as part of aMasters Programme in International Transport involving 16 students. The first time the game was played it was run as described by Sterman (1989). It was then played for a second time in with the students working in syndicates. Different syndicates re-enacted the four different ICT supply chain strategies. The demand pattern (which all syndicates responded to) was randomly drawn from an 8-sided dice