1、 1 外文翻译 原文 Supply Chain Inventory Management and the Value of Shared Information MaterialSource:http:/mansci.journal.informs.org/cgi/content/abstract/46/8/1032 Author: Ge rard P. Cachon Marshall Fisher In traditional supply chain inventory management, orders are the only information firms exchange,
2、but information technology now allows firms to share demand and inventory date quickly and inexpensively. We study the value of sharing these data in a model with one supplier, N identical retailers,and stationary stochastic consumer demand.There are inventory holding costs and back-order penalty co
3、sts. We compare a traditional information policy that does not use shared information with a full information policy that does exploit shared information. In a numerical study we find that supply chain costs are 2.2% lower on average with the full information policy than with the traditional informa
4、tion policy, and the maximum difference is 12.1%. We also develop a simulation-based lower bound over all feasible policies. The cost difference between the traditional information policy and the lower bound is an upper bound on the value of information sharing: In the same study, that difference is
5、 3.4% on average, and no more than 13.8%. We contrast the value of information sharing with two other benefits of information technology, faster and cheaper order processing, which lead to shorter lead times and smaller batch sizes, respectively. In our sample, cutting lead times nearly in half redu
6、ces costs by 21% on average, and cutting batches in half reduces costs by 22% on average. For the settings we study, we conclude that implementing information technology to accelerate and smooth the physical flow of goods through a supply chain is significantly more valuable than using information t
7、echnology to expand the flow of information. (Supply Chain; Multi-Echelon Inventory Management; Periodic Review Policies; Electronic Data Interchange) 1. Introduction Information technology has had a substantial impact on supply chains. Scanners collect sales data at the point-of-sale, and electroni
8、c data interchange (EDI) 2 allows these data to be shared immediately with all stages of the supply chain. The application of these technologies, especially in the grocery industry, has substantially lowered the time and cost to process an order, leading to impressive improvements in supply chain pe
9、rformance (see Cachon and Fisher 1997, Clark and Hammond 1997, Kurt Salmon Associates 1993). As a result of these success stories, there is now a general belief within industry that capturing and sharing real-time demand information is the key to improved supply chain performance. The purpose of thi
10、s research is to test this belief by rigorously measuring the value of information sharing and comparing this value to two other sources of supply chain improvement: reducing lead times and increasing delivery frequency by reducing shipment batch sizes. Note that the same information technology that
11、 facilitates information sharing also contributes to the reduction of lead times and shipment frequency by reducing the time and cost to process orders. Thus, the question addressed in here is not whether information technology improves supply chain performance, but how. Specifically, does the prima
12、ry gain come from sharing information or from allowing products to flow more quickly and evenly in the supply chain? We address this question within the context of a supply chain with one supplier and N identical retailers that face stationary stochastic consumer demand with a known distribution. Th
13、ere are fixed transportation times between locations, and shipment quantities equal a multiple of a base batch quantity. There are holding costs at all levels and back-order penalty costs at the lowest level. This model provides a reasonable representation of supply chains selling an established pro
14、duct under constant pricing conditions. We consider two levels of information sharing. With traditional information sharing the supplier only observes the retailers orders. With full information sharing the supplier has immediate access to the retailers inventory data. We develop an inventory policy
15、 for each information sharing level. Reorder point policies are used with traditional information sharing. The retailers also use reorder point policies with full information, but the supplier does not. Instead, the supplier uses its additional information to better allocate inventory among the reta
16、ilers and to improve its order decisions (i.e., to better time its own replenishments). The difference between supply chain costs under traditional and full information is one measure of the value of shared information. However, there may exist even better policies for either information level, that
17、 is, optimal policies are 3 unknown for each level. To account for this possible bias, we develop a simulation- based lower bound over all feasible policies, no matter what the level of information sharing is. The cost difference between traditional information and the lower bound is the maximum val
18、ue of shared information. In a numerical study with a wide range of parameter values we find that information sharing reduces supply chain costs by 2.2% on average, and the gap between traditional information policy cost and the lower bound is 3.4% on average. Cutting lead time by nearly half (from
19、five to three periods) reduces costs by 21% on average, and cutting batch size in half reduces supply chain costs by 22%. We recognize that this comparison is meaningful only if those lead time and batch size reductions can be reasonably expected from the implementation of information technology. In
20、 fact, we did observe comparable reductions at Campbell Soup Company when it implemented information technology to improve its supply chain.1 There has also been other documentation on the impact of information technology in the grocery industry: Barilla, the worlds largest pasta producer, reduced i
21、ts lead time from over one week to two days (Harvard Business School case 9-694-046); and H.E.B., a large grocery chain based in Texas, eliminated 6 to 10 days from its lead time (Harvard Business School case9-195-125). We conclude that while information sharing does reduce costs, simply flowing goo
22、ds through the supply chain more quickly and more evenly produces an order of magnitude greater improvement. 2. Literature Review The following papers show how sharing demand and inventory data can improve the suppliers order quantity decisions in models with known and stationary retailer demand: Bo
23、urland et al. (1996), Chen (1998), Gavirneni et al. (1999), and Aviv and Federgruen (1998). Lee et al. (2000) use shared information to improve the suppliers order quantity decisions in a serial system with a known autoregressive demand process. Liljenberg (1996) studies how to use shared informatio
24、n to improve the suppliers allocation of inventory among the retailers. In our model shared information is exploited for both uses: better supplier replenishments and better allocations to the retailers. We focus on sharing demand and inventory data, but there are other data that can be shared in a
25、supply chain. Gavirneni et al. (1999) measure the benefit of sharing the parameters of the retailers ordering policy with the supplier. Aviv (1998) explores the benefits of sharing forecasts for future demand. 4 In our model, as in the other studies mentioned, it is assumed that information is alway
26、s shared truthfully. Cachon and Lariviere (1997) study forecast sharing when the forecast provider has an incentive to provide an overly optimistic forecast of demand. Both Lee et al. (2000) and Gavirneni et al. (1999) assume there exists a perfectly reliable exogenous source of inventory; informati
27、on sharing has no impact on the retailer because its orders are always received in full after a fixed number of periods. In the other papers, as in our model, the supplier is the only source of inventory. Therefore, information sharing may impact the retailers by changing the suppliers order quantit
28、ies or allocations. Gavirneni et al. (1999) and Aviv and Federgruen (1998) allow for limited supplier capacity, whereas capacity is unrestricted in our model and in the other papers. The reported benefits of information sharing vary considerably. Liljenberg (1996) finds that better allocation lowers
29、 supply chain costs by 0% to 3.9%. Chen (1998) finds that supply chain costs are lowered up to 9%, and on average by 1.8%. Aviv and Federgruen (1998) report benefits of 0%5%. In contrast, Lee et al. (2000) find that information sharing lowered supply chain costs by about 23% in their scenario with t
30、he highest demand nonstationarity. However, Graves (1999) studies a similar model, with the exception that there is no outside inventory source, and concludes that information sharing provides no benefit to the supply chain. Gavirneni et al. (1999) report that sharing the retailers demand data reduc
31、ed the suppliers cost by 1%35%. The impact on the supply chains cost would be lower because information sharing in their model has no impact on the retailers costs. There is other research related to our work. Lee et al. (1997) find that sharing information reduces the suppliers demand variance, whi
32、ch should benefit the supply chain, but they do not quantitatively measure this benefit. There are many studies that investigate a supply chain model with one supplier, N retailers, stochastic consumer demand, and batch ordering. Some of them assume traditional information (e.g., Axester 1993, Cacho
33、n 1995, Chen and Samroengraja 1996, Lee and Moinzadeh 1986, Svoronos and Zipkin 1988), while others assume full information (e.g., Chen and Zheng 1997, Graves 1996, Mc Gavin et al. 1993). Because of different assumptions and test problems, it is not possible to meaningfully compare supply chain cost
34、s across those two sets of studies. Several researchers study allocation rules, but none addresses the issue of information sharing (see Cachon 5 1995, Chen and Samroengraja 1996, and Graves 1996). Anand and Mendelson (1997) study a one-period model in which retailers possess some local information
35、that cannot be shared with either a central agent or other retailers. In our full information model all relevant information can be shared with the central agent (i.e., the supplier).Chen and Zheng (1994) develop a lower bound over all feasible policies for a multiple retailer model. They show that
36、a full information policy is reasonably close to optimal, but they do not compare this policy with a traditional information policy. Graves (1996) also shows that his full information policies close to optimal. 3. Discussion Our results are surprising. Indeed, we undertook this research with the str
37、ong expectation that we would be able to demonstrate significant benefits to information sharing in these models. So why do the data speak the opposite conclusion? We conjecture the following answer: The retailers orders convey to the supplier a substantial portion of the information the supplier ne
38、eds to perform its ordering and allocation functions. When a retailer is flush with inventory, its demand information provides little value to the supplier because the retailer has no short-term need for an additional batch. A retailers demand information is most valuable when the retailers inventor
39、y approaches a level that should trigger the supplier to order additional inventory, but this is also precisely when the retailer is likely to submit an order. Hence, just as the retailers demand information becomes most valuable to the supplier, the retailer is likely to submit an order, thereby co
40、nveying the necessary information without explicitly sharing demand data. How can we reconcile our results with the clear trend across many industries to apply information technology to logistics and inventory management? In fact, our results are quite consistent with that trend: We do find substant
41、ial savings from lead time and batch size reductions, both of which are facilitated by the implementation of information technology. We only conclude that the observed benefits of information technology in practice are due more to the impact of information technology on lead time and batch size than
42、 in facilitating information sharing. Further, because we measure the value of information sharing using a lower bound on any feasible inventory policy that uses shared information, this finding will not be changed by the formulation of more sophisticated algorithms for sharing information. Although
43、 our model is representative of many actual supply chains, we recognize that our conclusion is limited to the setting we consider. In particular, in 6 our model demand is known, the retailers are identical, there exists only a single source for inventory, there are no capacity constraints, there are
44、 no incentive conflicts among the supply chains firms, and firms choose rational ordering policies. We anticipate that information sharing can have a significantly greater value in environments with unknown demand, for example, early sales of new products or established products on promotion. In tho
45、se settings information sharing would improve the suppliers ability to detect shifts in the demand process. We study a model with identical retailers for two reasons. First, no identical retailer models are far more complex to analyze. With no identical retailers it is not even known how to evaluate
46、 reorder point policies with traditional information. Second, in our setting we anticipate that information has the highest value with identical retailers. Information sharing allows the supplier to identify which retailers have the highest need for replenishments. This is most important when the re
47、tailers are not distinguishable, i.e., when they have identical demand processes. In fact, Aviv and Federgruen (1998) obtain comparable results to ours in a model with no identical retailers. Our justifications not withstanding, additional research is needed to assess information sharing with no ide
48、ntical retailers. Although a single inventory source is reasonable for many settings, in some supply chains the firms may have access to a second inventory source, albeit at a higher cost. In those settings information sharing may allow the supply chain to better decide when it should utilize its al
49、ternative sources. Gavirneni et al. (1999) found that information sharing is most valuable when capacity is not restrictive; information is valuable only if the system has the flexibility to respond to the information. Hence, imposing a capacity constraint on the supplier would probably lower the value of information in our model. We have assumed that a benevolent dictator decides all inventory shipments. This is reasonable when the sole objective is minimizing total supply chain cost, in other words, it doesnt matter which firm makes the decision. However, in
