The Economic and Environmental Implications of Centralized Stock Keeping

Transcription

1 The Economic and Environmental Implications of Centralized Stock Keeping H. Scott Matthews and Chris T. Hendrickson Keywords e-commerce input-output life-cycle assessment (IO-LCA) inventory management logistics spare parts warehousing Summary Recent changes to the management of inventory and warehousing methods have created significant changes in business processes. These changes have produced economic savings to firms from reduced handling of supplies. The system-wide impacts of this shift in methods on overall cost and the environment are still unclear, however. Reductions in inventories can provide significant environmental savings. In this article, we analyze the changes in inventory control methods and assess the environmental and cost tradeoffs between increased trucking and more efficient centralized warehouses. We consider the case of consolidating the spare-parts inventory at U.S. Department of Defense warehouses and discuss similarities to other existing businesses. The case suggests large economic and environmental benefits due to reductions in warehousing expenses, despite higher transportation costs. Address correspondence to: H. Scott Matthews Carnegie Mellon University Department of Civil and Environmental Engineering Porter Hall 119 Pittsburgh, PA , USA Copyright 2003 by the Massachusetts Institute of Technology and Yale University Volume 6, Number 2 Journal of Industrial Ecology 71

2 Introduction Beyond the much-scrutinized productivity implications (Litan and Rivlin 2001), two fundamentally important questions arise when considering the implications of the new economy. First, if there is an increase in convenience as a result of information-technology-enabled systems (e.g., substituting shopping trips with home deliveries), what will be the effects on environmental quality and sustainability? Second, will the much-heralded economic benefits of the new economy also lead to environmental benefits? One of the hallmarks of the new economy is the substitution of information technology for physical products and processes (Borenstein and Saloner 2001). Examples of this substitution include Internet sites for taking orders, business-to-business procurement, and digital record keeping. This transformation from physical to digital assets has significant business and economic implications, suggesting that there may be environmental and sustainability concerns as well. For example, an early estimate of the electricity burden needed to power the Internet was on the order of 8% of total U.S. electricity demand (Mills 1999). Subsequent studies suggest that the Internet s share of that electricity is considerably lower, on the order of only 1% (Koomey et al. 1999; Romm 1999). Regardless of the size of the actual number, the use of information technology depends on other physical assets. Likewise, all business processes rely on many other goods and services in the economy. Another example is the use of the Internet (and package-delivery companies) rather than traditional retail networks for product delivery to customers (Matthews et al. 2000). This distribution system may result in significantly greater transportation impacts, particularly if airfreight is used. Studies have suggested, however, that Internet deliveries of products pose no significant additional harm to the environment (Caudill et al. 2000). Matthews and colleagues (2001) have shown that the economic and environmental savings from reducing unsold book inventories which are generally 35% of production more than offset the added economic and environmental costs of overnight shipping from centralized warehouses to homes. Benefits for energy use, air pollution, and greenhouse gas (GHG) emissions could be realized. Similarly, Klassen (2000) found that pollution prevention and justin-time (JIT) production strategies were complementary and mutually reinforcing for the furniture industry. The new economy has clearly led to new and innovative solutions to traditional problems. Inventory Management A key piece of any business is its inventory management. Although not a core activity of most businesses, the financial implications of inventory management are critical to profitability. The components of inventory management include organizing incoming supplies, work-inprocess inventory, and final product. The management shift toward lean production, including JIT inventory management, changed a paradigm of maintaining a large inventory of parts and supplies to one where firms accepted deliveries more frequently but maintained lower overall inventory levels (Womack et al. 1990). Another effect of JIT inventory management is that deliveries may be made via faster freight methods (e.g., truck or air instead of rail or water), which results in a shorter pipeline because of the reduced inventory in transit. The reduced inventory levels lead to less production. Another paradigm shift has been toward using warehouses in the sky, where the airplanes of package-delivery services such as Federal Express and United Parcel Service are considered part of a company s inventory management system. In this case, firms can deliver products to overnight carriers for immediate delivery instead of storing them. More recently, firms have partnered with package-delivery companies to manage inventories and to fulfill and deliver orders. The major package and courier service firms all now have e- logistics subsidiaries to nurture such growth areas. Instead of maintaining their own warehouse and order management systems, firms can electronically forward their orders to the carriers that manage inventory, pick up and ship orders, and deliver them to customers. Because logistical activities are outsourced to more efficient orga- 72 Journal of Industrial Ecology

3 Figure 1 A comparison of estimated energy implications for air, truck, and rail transportation. Total and direct energy use is shown in terajoules (TJ) per $1 million of transportation, and in megajoules (MJ) per (short) ton-mile. Direct values are from the exclusive provision of transport, while total energy values include effects from the modal supply chain (CMU GDI 2002). Note 1 TJ 95 million BTU, average 1997 transportation cost is from U.S. DOT (2002). nizations, retail firms can concentrate resources on areas such as marketing, while the courier services can leverage their experience and resources to manage inventory in bulk at locations near their logistics hubs (Kuby and Gray 1993; Shaft et al. 2001). Another emerging practice is virtual warehousing. In this case, firms may actually hold no physical inventory of supplies or parts. Firms contract with partners or third-party vendors, who manage an overall level of inventory and guarantee that their clients will receive the necessary supplies when needed. Firms save money by outsourcing the activity of inventory management, while the contract firms can centrally manage similar inventories of products for many companies at once (Landers et al. 2000). One of the obvious tradeoffs when substituting transportation for inventory levels is the implications of faster freight modes. In 1997, the average U.S. cost to move 1 ton of freight over the distance of 1 mi was $0.80 by air, $0.26 by truck, and $0.02 by rail (U.S. DOT 2002). Figure 1 shows a comparison of the energy effects for the three freight options using economic inputoutput life-cycle assessment (LCA) (CMU GDI 2002). The energy required per ton-mile is even more skewed than cost. Chemical management and aircraft spare parts provide two interesting examples in this regard. Many companies are entering into chemical service agreements with manufacturers rather than maintaining their own chemical inventories (Reiskin et al. 1999). These agreements allow producers to drastically reduce their on-site chemical storage facilities, which has both financial and regulatory compliance implications. Firms pay their chemical service providers on a per-unit production basis, which aligns incentives on both sides to minimize production and use of chemicals. In the airline industry, carriers need to keep substantial inventories of spare parts on hand to support safety inspection and maintenance functions. Firms need to manage inventories of thousands of spare parts, even for parts not prone to Matthews and Hendrickson, The Implications of Centralized Stock Keeping 73

4 failure. Virtual warehousing allows a third party to perform the majority of the work and for airlines to pool inventories. As a result of these changes, warehousing and inventory control are increasingly centralized, both in organization and in warehouse sites. With centralization, the amount of inventory and warehousing costs are reduced, but transportation costs increase. For those products that have low demand, become obsolete, or deteriorate in storage, the lower warehouse inventory levels achieved by centralization could lead to less production of stock, lowering costs and environmental impact. Warehousing, Airfreight, and Trucking Impacts In this article, we create case study models to show the environmental implications of the transitions occurring relevant to inventory management and warehousing. We use a form of LCA called economic input output-life cycle assessment (EIO-LCA). This software, developed by Carnegie Mellon University s Green Design Initiative in 1995, traces out the various supplychain economic transactions, resource requirements, and environmental emissions required for a particular product or service (Lave et al. 1995; Hendrickson et al. 1998). The model captures the various economic manufacturing, transportation, mining, and related requirements to produce a product or service. When estimates of environmental effects as related to production are available, such a model is a useful industrial ecology tool to see supply-chain-wide effects (Joshi 1999; Matthews et al. 2001). The model is based upon the U.S. Department of Commerce s 1997 Annual commodity input-output model of the U.S. economy (Kuhbach and Planting 2001) and is publicly available on the Internet at This model is able to show the implications of a change in production for a particular economic sector and how this change relates to emissions of GHGs and other air pollutants, health and safety effects, toxic releases, and energy use. Air pollutants estimated by EIO- LCA include carbon monoxide, sulfur dioxide, nitrogen oxides, particulate matter, and volatile organic compounds. The more familiar Society for Environmental Toxicology and Chemistry (SETAC)-type process-based LCA models are needed when specific materials or technologies are compared. Input-output-based LCA models are useful when sector-level comparisons are sufficient (as in this article). A primary limitation of input-outputbased models is that any activity must be described as one or more of the 500 existing economic sectors. Further, all of the information about individual sectors applies then to average production in that sector and at a very aggregate level. If a particular firm or process is cleaner than average, the results will not represent that fact. Finally, effects of new or emerging technologies can often not be described by existing sectors. For a more detailed comparison of these LCA models, see the analyses by Matthews and Small (2000) and Lenzen (2000). The model presented here assumes generic location of warehouses. This simplification ignores potential implications that would result from siting warehouses in areas of environmental concern, for example, in ozone nonattainment zones. The additional trucking in such a scenario, although creating the same marginal effects as those expressed here, would create significant additional externalities. In this case, the transfer of trucking from geographically disparate to centralized locations may reduce total emissions, but would likely have severe implications on human health for residents near the new warehouses. As stated above, the general tradeoff from virtual warehousing is less use of warehousing by a corporation, but increased use of logistics services to route and deliver centralized inventory. As a general comparison, we show the estimated economic and environmental implications of equal purchases of warehousing compared to transportation. Using EIO-LCA, the expected effects across the supply chain due to $100 million of warehousing and $100 million of truck transportation expense are summarized in table 1. The difference column shows the positive effect (i.e., reduction generated) as a result of shifting a dollar from warehousing into more truck transportation, as is being done in many modern inventory control systems. Negative values imply that more impact would occur from the transition (e.g., about 1,700 TJ more energy would be 74 Journal of Industrial Ecology

5 Table 1 Total environmental releases and resource consumption across supply chain from $100 million of warehousing and storage and truck transportation in 1997, raw difference and trade-off Effects Warehousing Truck trans. Raw difference Trade-off (unitless) Total supply chain economic purchases (million US$) Electricity used (MkWh) Energy used (TJ) 770 2,500 1, Air pollutant emissions (metric tons) 830 5,700 4, Fatalities Greenhouse gases released (metric tons CO 2 equivalents) 63, , , Ores used (metric tons) 1,600 1, Hazardous waste generated (metric tons) 4,200 2,800 1, Toxic releases and transfers (metric tons) Toxicity-weighted toxic releases and transfers (metric tons) Source: CMU GDI (2002). Note: Warehousing represented by EIO sector , U.S. Standard Industry Classification Code (SIC) 422, truck transportation represented by EIO sector , SIC Codes 421 and 423. Transfers of toxic releases include shipments off-site and to publicly owned treatment facilities. Air pollutant emissions are sum of sulfur dioxide, carbon monoxide, nitrogen dioxide, volatile organics, and particulate matter emissions. Although each gas is in metric tons, gases are not comparable; however, sum is presented to conserve space. used). The various effects summarized in table 1 suggest that a dollar-for-dollar trade-off from warehousing to trucking would lead to significantly higher uses of resources and energy, as well as greenhouse gas (GHG) emissions. Thus, the actual expenses shifted will be important. As mentioned above, the EIO-LCA model estimates the economic, energy, and social effects of a given economic sector across its total supply chain. Thus, the total supply chain economic purchases refer to the total dollars spent in the U.S. economy to support $100 million of increased warehousing output (or an additional $100 million of gross domestic product from warehousing). The warehousing sector has a supply chain dependent on real estate management, construction and maintenance, personnel, energy, wholesale trade, and security services. Purchases of $100 million of output from the sector itself and from the warehousing sector to all of its suppliers total $178 million. The total is more than $100 million because of double counting of intermediate outputs. In other words, there is $178 million of economic activity in the economy, but output or gross domestic product only counts final output production, which is $100 million. The production of intermediates, although not counted as final output, still generates pollution and other social impacts. Thus, the effects of intermediate production should be included when considering the implications of producing final output. EIO-LCA is able to estimate effects across the supply chain for any impact of concern for which sectoral detail is available. One such impact is worker health and safety in terms of fatalities resulting from production. In this example, the number of estimated fatalities across the warehousing supply chain for $100 million of output are estimated to be 0.2 (i.e., it would take $500 million of warehousing output to cause a single expected fatality). The results above allow for the construction of tradeoff functions between warehousing and trucking. Consider the tradeoff of GHGs between the two alternatives. For some amount of warehousing (W, in millions of dollars), the GHGs emitted would be (W)*630 equivalent metric tons of CO 2. Similarly, the amount of GHG emissions from trucking would be (T)*1,850 equivalent metric tons of CO 2.To compare the tradeoff in terms of GHG emissions between warehousing and trucking, the breakeven amount of trucking that would need to be used to offset the GHG reduced from a $1 million decrease in warehousing could be found by creating a ratio of the above equations, that is, T (630/1,850)*(W), or roughly T 0.34W Matthews and Hendrickson, The Implications of Centralized Stock Keeping 75

6 (or T/W 0.34). Thus, for centralized warehousing to achieve a net reduction in GHG emissions, the total cost of truck transportation must not exceed one-third of the value of the warehousing savings. As an example, $100 million of reduced warehousing would reduce 63,000 t of CO 2 equivalent. As long as the amount of increased use of truck transportation was less than $34 million (i.e., emitting 63,000 t), the system-wide GHG emissions would be less overall. The final column of table 1 shows corresponding partial tradeoffs for all of the effects summarized above between warehousing and trucking (the result is unitless, as it divides values with common units from trucking and warehousing effect columns). Tradeoff values near 1 suggest trucking and warehousing effects are relatively equal on a comparable cost basis. High numbers in this partial tradeoff analysis suggest that large amounts of trucking relative to warehousing would be needed to offset the benefits from reduced warehousing. Low numbers suggest that only modest trucking expenses would be needed to offset savings from less warehousing. The primary tradeoff occurring in virtual warehousing is having a third party manage inventory, which represents a significant cost savings to businesses, at the expense of incurring more delivery charges from needing to replenish inventory and supplies as required. From the numbers in table 1, the tradeoff for air pollutants is notable. From a strictly economic perspective, a firm would shift to centralized warehouse operations whenever cost savings can be garnered. Travel time delays associated with trucking, however, generally imply that warehouse savings must be significantly higher than transport cost increases to warrant a change in inventory management. To better understand the sources of the effects summarized above, table 2 shows the top ten sectors causing emissions of GHGs in the supply chain of warehousing, along with the estimated direct and indirect economic purchases. Electric services (the production and distribution of electricity for use in heating, lighting, etc.) is the largest sectoral source of global warming potential emissions, representing 50% of the total, even though purchases of electricity are estimated to be only 2% of the total economic transactions in the supply chain of warehousing. In contrast, the warehousing sector itself generates only 25% of the global warming potential emissions, but 58% of the supply-chain purchases. Case Study of Military Spare-Parts Management As an example of the benefits of centralized warehousing, we consider the case of spare-part inventory management for the U.S. Department of Defense. Although this example is not purely a result of new economy business processes, the benefits are comparable and are discussed at the end of this section. The U.S. General Accounting Office (U.S. GAO) estimated the total inventory of spare parts in the Army, Navy, and Air Force to be 632,000 distinct line items (or 108 million total parts), with a value of $83 billion (U.S. GAO 1997). We focus on the operations of the U.S. Defense Logistics Agency (U.S. DLA), which oversees inventory management for the military services. In 1997, the U.S. DLA had a total of 286 spare-part storage locations, holding more than 700,000 line items and 100 million total parts. The total value of these parts was roughly $83 billion. Table 3 summarizes the inventory held by the U.S. DLA. As seen above, the vast majority of line items, parts quantity, and inventory value is contained in only a subset of the storage locations. Only 19 of the 286 storage locations hold more than 80% of the total parts and 90% of inventory value. Based on the inventory management discussion above, significant benefits might accrue if the inventory were completely centralized into the 19 major locations. Benefits would come from reduced inventory handling (warehousing) costs, less production of unused inventory, and so on. Additional costs would be incurred, however, because of transportation of parts and increases in repair delays while waiting for deliveries. Above and beyond any national defense budgeting benefits, environmental benefits might also be realized. We follow the U.S. GAO report and focus on the potential for savings by better management 76 Journal of Industrial Ecology

7 Table 2 Total and top ten sources of greenhouse gas emissions across the supply chain from $1 million of warehousing and storage (EIO sector and U.S. Standard Industry Classification Code [SIC] 422) Sector Economic purchases (thousand $) Global warming potential (metric tons) Total (all 500 sectors) 1, Warehousing and storage 1, Real estate agents, managers, operators, and lessors Personnel supply services 57 1 Trucking and courier services, except air Other repair and maintenance construction 46 1 Management and public relations services 33 1 Electric services (utilities) Wholesale trade 27 4 Petroleum refining Detective and protective services 23 2 Source: CMU GDI (2002). Table 3 Summary of Defense Logistics Agency spare parts inventory Storage locations Number Line items Quantity Value (1996 million US$) Major ,500 86,700,000 75,165 Other ,325 21,300,000 8,315 Total , ,000,000 83,480 Excess ,700 13,140,000 2,744 Other non-excess ,625 8,160,000 5,570 Source: U.S. GAO (1997, tables 1 4) 1 There is some double counting here as the Army, Navy, and Air Force have colocated storage sites as well as identical line items. For simplicity, because the overlap is small, we show a total as the sum of major and other. of inventories of seldom-issued spare parts at the nonmajor locations. As indicated in table 3, the U.S. GAO estimated that $2.7 billion of the U.S. DLA s inventories are in excess of the amounts needed to meet current operating and war reserve requirements (about 33% of the total nonmajor locations inventory level) (U.S. GAO 1997). Because this inventory is unlikely to be used, it could be sold as surplus and would thus reduce inventory handling costs. This would leave only 80,625 line items, or 8.16 million inventoried units, totaling $5.57 billion in the nonmajor locations. Further, the U.S. GAO analyzed dormant inventory as items not requested in two years. Although the U.S. GAO did not specify a value for all dormant inventory, the Army had more than 50% of nonmajor inventory classified as dormant (or 25% of the value at these locations). Thus, the amount of inventory actually needed from the nonexcess items is likely low. We assume that only half of the nonexcess inventory would need to be shipped back to major locations for management. This would include about 40,000 line items, or 4 million units, to be transported by truck. The remainder of the analysis considers the impact of centralizing the remaining nonexcess inventory contained in the nonmajor locations, a total of about 4 million parts. To consider the overall impacts of centralizing inventory, we need to consider the following components: (1) the benefits associated with reduced inventory management expenses at nonmajor locations, (2) the costs of moving the remaining inventory from nonmajor back to major locations, and (3) the costs of increased inventory handling expenses at major locations. Details of these three components are given below and are summarized in table 4. Matthews and Hendrickson, The Implications of Centralized Stock Keeping 77

8 1. In total, the U.S. GAO found that $382 million in annual inventory holding costs could be avoided by eliminating the excess spare-part inventories from 101,000 items (U.S. GAO 1997, table 6). We will reduce all inventory from these locations, however. Thus, we scale up the $382 million by the ratio of other to excess line items (184,325/103,700) and estimate $693 million in annual benefits from removing all inventory from these locations (an additional $311 million). 2. The U.S. GAO did not provide estimates of inventory movement costs. The U.S. DLA, however, noted that the average delivery cost (to purchasers) was $20 (U.S. DLA 1997). This is assumed to be the cost for individually shipped items, so we assume a 50% reduction in cost for bulk handling and transportation of the remaining nonexcess inventory items between locations with 4 million items at $10/item. This would be $40 million in trucking expenses. 3. U.S. DLA inventory expenses are calculated annually per square foot. Items are charged according to the area occupied by their container s footprint, however, not the actual part s footprint. For example, the U.S. GAO notes the case of a $3 plastic bumper stored as the lone item in a plastic bin at a nonmajor location. Because the bin occupied 2 ft 2 of space and inventory was charged at $5.15 per square foot, the annual storage cost was about $10 for the $3 part. In short, even though inventory is being centralized with savings from shutting down the nonmajor locations, there will be some increased inventory handling costs from dealing with the newly centralized inventory. With no specific information available, we assume this to be 25% of the $311 million of inventory costs associated with the remaining nonexcess inventory, or $78 million. Although not included in the analysis above, the sale of surplus equipment would also generate revenues to be used toward buying new parts or to pay for inventory consolidation. Table 5 shows the results from estimating the supply-chain implications of centralizing inventory using EIO- LCA and the three analytical components described in table 4. Although the energy, environmental, and social impacts of warehousing might not be considered important, it turns out that reducing warehouse usage has economic, energy, and environmental benefits, even when additional truck transportation is needed. Model Limitations and Sensitivity Analysis As stated above, use of EIO-LCA depends on sectoral comparisons. The use of warehousing and trucking by the U.S. DLA may or may not be comparable to the average production in the United States of those sectors. The U.S. DLA may be more or less efficient in economic, energy, or environmental categories. The order-ofmagnitude assumptions used are probably relevant, however. Also, the estimates above apply only to the given assumptions. Although large benefits are realized, different cases or assumptions could lead to different results. Table 6 shows the costs of shipping the inventory to be centralized (by truck or air) that would be needed to outweigh the benefits obtained from reduced warehousing, similar to the break-even analysis suggested above. The scenario of net increased inventory costs in major locations is not considered; however, a second sensitivity analysis case is included when the increased warehouse costs at major locations are double the assumed levels (i.e., $156 million). In the worst-case scenario for truck Table 4 Summary of costs and benefits from centralizing spare parts inventory Category Cost (savings) EIO-LCA sector used Nonmajor location inventory reductions ($693 million) Warehousing and storage Shipping back to major warehouses $40 million Trucking and courier services, nonair Major locations increased inventory $78 million Warehousing and storage 78 Journal of Industrial Ecology

9 Table 5 Summary of supply-chain implications from centralizing inventory Effects Reduced nonmajor warehouse (1) Truck (2) Increased central warehouse (3) Net reduction Electricity used (MkWh) Energy used (TJ) 5, ,800 Air pollutant emissions (metric tons) 5,700 2, ,800 Fatalities Greenhouse gases released (metric tons CO 2 equivalents) 440,000 74,000 49, ,000 Ores used (metric tons) 11, ,200 9,000 Hazardous waste generated (metric tons) 29,000 1,100 3,200 25,000 Toxic releases and transfers (metric tons) Toxicity-weighted toxic releases and transfers (metric tons) Source: CMU GDI (2002). Table 6 Sensitivity analysis of truck transportation costs needed to offset centralized warehousing benefits (in millions of dollars of transportation needed) Trucking Effects Case 1 Case 2 Case 1 Case 2 Economic purchases (million $) Electricity used (MkWh) 1,327 1,159 1,489 1,301 Energy used (TJ) Air pollutant emissions (metric tons) Fatalities ,147 1,006 Greenhouse gases released (metric tons CO 2 equivalents) Ores used (metric tons) Hazardous waste generated (metric tons) Toxic releases and transfers (metric tons) Weighted toxic releases and transfers (metric tons) Source: Data from CMU GDI (2002). Case 1/trucking: Values as estimated in table 5, solving for break even amount of trucking. Case 2/trucking: Same as above, except increase in central inventory doubled to $156 million. Case 1/air: Values as estimated in table 5, solving for break even amount of air transport. Case 2/air: Same as above, except increase in central inventory doubled to $156 million. Air transportation (i.e., where the smallest value for trucking is needed to outweigh benefits from reduced warehousing), we see that air pollutant emissions are most important. In this case, costs would need to double to cause increased overall air emissions. All other scenarios would require even higher amounts of truck transportation to outweigh the benefits. As seen in figure 1, air transport is roughly 3 times as expensive as truck on a per ton-mile basis. Thus, about $120 million of air transport would likely be needed to centralize the inventory. For the air transport worstcase scenario this time, total energy use more than $222 million (almost double) would be required to outweigh the benefits of inventory centralization. Comparison and Linkage to Other Businesses The example above may represent a special case and may not be representative of all centralized warehouse problems; however, many existing business models would show similar bene- Matthews and Hendrickson, The Implications of Centralized Stock Keeping 79

10 fits. For example, siting several large national warehouses for book inventory for an e- commerce company such as Amazon, or even virtual parts warehouses, would be similar. As a comparison with private industry, more than $7 billion of spare parts are consumed per year for Boeing aircraft, and the industry overall holds about $25 billion in inventory. Overall, 26 million parts are changed on airplanes per year, and $150 billion is spent on spare parts per year worldwide (U.S. FAA 1995). Boeing is creating a virtual warehousing system to manage inventory of expendable spare parts in the airline industry (Boeing Company 1999). Inventory management of spare parts is critical in the industry; a Boeing 747 has 6 million parts. In the virtual warehouse, Boeing manages the parts inventory, and airlines only pay for parts as they are used. Boeing has seven regional distribution warehouses worldwide. Generally, individual airlines hold their own local inventories of parts, many of which are seldom used (as in the U.S. DLA example above). Centralizing these inventories at separate geographical sites would have similar effects as those listed above. One notable difference would be that parts would need to be delivered via highly expedited methods (e.g., on the next plane out) because the maintenance delays associated with waiting for a part to be delivered are very high. The opportunity cost from lost revenue due to a maintenance delay on a 747 is $40,000/min. The final columns of table 6, although not related directly to this aircraft example, show how much could be spent on air transportation to expedite delivery of parts. As seen in the example above, the amount of air transportation needed to outweigh centralized inventory benefits is greater than the relative truck/air cost shown in figure 1. Conclusions Recent innovations in distribution and inventory control have led to significant economic savings for firms. In general, these changes have not been studied in terms of their implications for environmental quality, energy use, and sustainability. This article discusses some of the issues and estimated effects associated with substituting truck transportation for warehousing and considers a case study of centralizing military warehouse inventory as a proxy for the warehousing activities associated with modern supply-chain management and logistics systems. Although warehousing may not normally be considered a highly polluting activity, its supply chain depends on many other polluting activities. We estimate that centralizing inventory leads to large reductions in energy and resource use, as well as environmental emissions. In the case of consolidating military spare parts, roughly twice as much trucking or air shipping could be used than currently required and still have net benefits. As the new economy continues to streamline inventory management methods and information technologies facilitate new ordertaking and distribution methods, further benefits may be realized. Acknowledgments The authors are grateful to Lester Lave and to two anonymous referees for constructive comments. The authors acknowledge support from the National Science Foundation/Lucent Technologies Fellowship grant no. BES , National Science Foundation/U.S. Environmental Protection Agency grant no. R826740, and the AT&T Industrial Ecology Faculty Fellowship Program. References Boeing Company Boeing and British Airways launch global airline inventory network. Press release, Seattle, WA, 22 September. Borenstein, S. and G. Saloner Economics and electronic commerce. Journal of Economic Perspectives 15(1): Caudill, R. J., Y. Luo, P. Wirojanagud, and M. Zhou Exploring the environmental impact of ecommerce on electronic products: An application of fuzzy logic and lifecycle studies. Paper presented at Electronics Goes Green 2000, September, Berlin. CMU GDI (Carnegie Mellon University Green Design Initiative) Economic input-output life cycle assessment (EIO-LCA) model. Accessed 25 July Journal of Industrial Ecology

11 Hendrickson, C. T., A. Horvath, S. Joshi, and L. B. Lave Economic input-output models for environmental life cycle assessment. Environmental Science and Technology 32(4): 184A 191A. Joshi, S Product environmental life cycle assessment using input-output techniques. Journal of Industrial Ecology 3(2/3): Klassen, R. D Just-in-time manufacturing and pollution prevention generate mutual benefits in the furniture industry. Interfaces 30(3): Koomey, J., K. Kawamoto, B. Nordman, M.A. Piette, and R. E. Brown Initial comments on The Internet begins with coal. Memo to Skip Laitner of U.S. EPA, LBNL Berkeley, CA: Lawrence Berkeley National Laboratories. Kuby, M. and R. Gray The hub network design problem with stopovers and feeders: The case of Federal Express. Transportation Research A: Policy and Practice 27A: Kuhbach, P. D. and M. A. Planting Annual input-output accounts of the U.S. economy, Survey of Current Business. January: Landers, T. L., M. H. Cole, B. Walker, and R. W. Kirk The virtual warehousing concept. Transportation Research Part E, Logistics and Transportation Review 36(2): Lave, L. B., E. Cobas, C. T. Hendrickson, and F. McMichael Using input-output analysis to estimate economy-wide discharges. Environmental Science and Technology 29: 420A 426A. Lenzen, M Errors in conventional and inputoutput based life-cycle inventories. Journal of Industrial Ecology 4(4): Litan, R. E. and A. M. Rivlin Beyond the dot.coms: The economic promise of the Internet. Washington, DC: Brookings Institution Press. Matthews, H. S. and M. Small Extending the boundaries of LCA through environmental economic input-output models. Journal of Industrial Ecology 4(3): Matthews, H. S., C. T. Hendrickson, and L. B. Lave Harry Potter and the hole in the ozone layer. IEEE Spectrum November: Matthews, H. S., C. T. Hendrickson, and D. Soh Environmental and economic effects of e- commerce: A case study of book publishing and retail logistics. Transportation Research Record 1763: Mills, M. P The Internet begins with coal: A preliminary exploration of the impact of the Internet on electricity consumption. Technical report. Alexandria, VA: Greening Earth Society. Reiskin, E. D., A. L. White, J. Kauffman Johnson, and T. J. Votta Servicizing the chemical supply chain. Journal of Industrial Ecology 3(2): Romm, J The Internet economy and global warming. Technical report. Annandale, VA: Center for Energy and Climate Solutions, Global Environmental and Technology Foundation. Shaft, T. M., M. P. Sharfman, and M. Swahn Using interorganizational information systems to support environmental management efforts at ASG. Journal of Industrial Ecology 5(4): U.S. DLA (U.S. Defense Logistics Agency) Summary of costs. pr97summary.htm. Accessed 15 December U.S. DOT (U.S. Department of Transportation) National transportation statistics BTS Washington, DC: U.S. Department of Transportation. U.S. FAA (U.S. Federal Aviation Administration) FAA unveils parts broker accreditation plan. Press release, 24 May. U.S. GAO (U.S. General Accounting Office) Defense inventory: Spare and repair parts inventory costs can be reduced. Letter report, 17 January, GAO/NSIAD Washington, DC: U.S. General Accounting Office. Womack, J. P., D. T. Jones, and D. Roos The machine that changed the world. New York: Rawson Associates. About the Authors H. Scott Matthews is an assistant professor in the Departments of Civil and Environmental Engineering and Engineering and Public Policy at Carnegie Mellon University (CMU), in Pittsburgh, PA, USA. Chris T. Hendrickson is the Duquesne Light Professor of Engineering and department head, Department of Civil and Environmental Engineering, CMU. Matthews and Hendrickson, The Implications of Centralized Stock Keeping 81