The twin supply chain management challenge facing retailers, distributors, manufacturers, and suppliers is to maximize customer service while minimizing costs. For most supply chain operations the service challenge can be expressed in terms of availability: having the right product at the right place when the customer wants it. The cost challenge is to make that happen at low cost.
Matching supply with ever-changing and unpredictable demand is what separates Wal-Mart from Sears, Dell from HP’s PC business, and Zara from Marks & Spencer. Supply chain disruptions, at their core, impair a firm’s ability to satisfy demand because of a reduction in the firm’s manufacturing or delivery capacity. But stock-outs and lack of product resulting from wrong orders also represent a failure to satisfy demand, albeit because of a wrong forecast rather than a disruption.
Demand in certain industries is notoriously unpredictable. Consequently, much can be learned from understanding how leading operators in these industries design their supply chains to mitigate such problems. Chapter 6 is focused specifically on supply chain designs aimed at mitigating forecasting difficulties. This chapter, however, first explains the structure and dynamics of supply chains.
In 1927, Ford Motor Company began producing the Model A in its River Rouge plant. Spanning over 2,000 acres, the “Rouge” was the largest industrial complex in the world. In 1929, it employed more than 100,000 people and incorporated all phases of auto manufacturing. Through 1942, it generated its own power and made its own parts, even glass and rubber, in the process of producing 15 million automobiles. The company fed the Rouge from its own iron-ore mines in northern Michigan and Minnesota and from its coal mines in Kentucky and West Virginia. Raw materials were brought in on Ford-owned railroad lines and ships for processing at the Rouge mills.1
Henry Ford clearly did not want to depend on suppliers. But even though Ford ran a vertically integrated factory, where all stages of manufacturing were controlled by Ford, it still had the challenge of coordinating the internal parts of its sprawling system.
The main challenge facing Ford operations managers was deciding what to produce, how much to produce, and when to produce it. For example, if they planned to make 20,000 cars in a given week, then they needed to have the parts for all those cars. To complete the wheel assembly, for instance, Ford would need to have on hand 40,000 axles, 80,000 wheels, and 400,000 lug nuts (five per mounted wheel). They would need 80,000 tires (not including a spare), and the tire plant at the Rouge would need to have enough raw material on hand, which it would have to process weeks and months ahead of the car assembly, in Ford’s South American plant. Even in 1927, each automobile had thousands of parts, each part made of other parts and raw materials. These parts took time to manufacture and thus managers needed to schedule production far in advance.
The first step in managing all the parts of a complex product such as the Ford Model A is to list them all. In manufacturing terminology, the list of all parts, and the quantities involved in manufacturing a product, is known as the bill of materials (BOM). Consider the manufacturing of a toy car.2 The BOM and the processes involved are depicted in the process map in figure 5.1. In this example, two pins for the axles (Part #2) are cut (in Process 3 shown by the circled numbers in the figure) and four drums (Part #3) are drilled (in Process 4) to be welded (in Process 5) into an axle set (Part #6). Metal paint (Part #7) is applied to the axle set which is then assembled with the four tires to produce a wheel set (Part #8). That part is then mated (in Process 12) to the car body (Part #14) to produce the toy car (Part #15).
Production planners use the BOM in the material requirement planning (MRP) process to identify the parts needed, the quantities required, the inventories of parts available, and the time when they should be ordered so that they will be available for the manufacturing process.
Modern supply chain processes involve significantly more complicated BOM and MRP processes. In most cases, the MRP uses special computer applications to control the tens of thousands of parts and processes involved. These applications identify situations in which several finished products rely on the same supplier, or on the same manufacturing process, with an eye toward efficiency—leveling the requirements and the work loads. The same MRP logic, however, can also be used to identify vulnerabilities of suppliers, plants, and processes. Thus, the BOM can be turned into a supply chain vulnerability map, as mentioned in chapter 2.
At the height of the high tech bubble, Lucent’s main supplier of inductors to dozens of Lucent products cut off deliveries to all Lucent plants. The supplier had fixed capacity and a huge demand, and it had decided to divert its constrained output to a large consumer electronics firm. This resulted in an immediate scramble on Lucent’s part using the BOM of all products to identify where the capacitors were used, identify the future demand for those products, use the BOM again to identify the resulting future demand for the inductors, and start looking for new sources of supply. Because of its immediate response, including the involvement of senior management with the supplier, Lucent was able to get the supply restarted, but the event served to highlight the importance of resilience to Lucent’s top management. This case was one of the impetuses for the establishment of the Supply Chain Network organization (described in chapter 11), including a group of supply relationships managers. These actions made Lucent even more resilient, making such surprises less likely and diminishing their potential to affect Lucent.
In the second half of the twentieth century, many manufacturers realized that total vertical integration had many disadvantages. A company could not be the best in everything. For example, it was difficult for Ford to be the best tire maker in the world while it focused on building the best cars in the world. Thus, companies started focusing on their core competencies, buying ready-made parts and services from suppliers whose core competency was, presumably, the part or service bought. The suppliers were able to concentrate on making specific parts, enjoying economies of scale by serving several customers, and focusing their expertise. That trend accelerated further in the 1980s and 1990s with the globalization and deregulation of commercial activities worldwide, leading to the widespread practice of outsourcing. In many cases production moved offshore, as certain regions of the world and certain countries, such as China, developed expertise and scale in many stages of manufacturing. It also led to the birth of modern supply chain management, which is focused on the flow of products through the global web of suppliers, manufacturers, distributors, transportation carriers, and retailers from raw material to finished goods in consumers’ hands, and the recycling and disposal of these products.
Figure 5.2 depicts a simplified view of the main stages of garment manufacturing—from raw material to finished product. It starts with the use of cotton, wool, or oil-based raw material to produce fiber. The fiber is used as the main ingredient in the manufacturing of yarn. The yarn is textured, twisted, warped, drawn, woven, scoured, dyed, and finished into fabric. The fabric is rolled up, packed, and sent to apparel manufacturers. Then the fabric is cut, sewn, and finished to make apparel. The apparel is then shipped and sold in retail stores to consumers.
In some cases textile manufacturers control several of these steps in one location; in others they use specialized plants in disparate locations. Chapter 1 outlined the example of the manufacturing of Champion sports bras by Griffin Manufacturing of Fall River, Massachusetts, using material from the Far East that is processed in Massachusetts, Honduras, and Vermont before it is shipped to U.S. retailers.
Another example was described by Victor Fung, chairman of Li & Fung, Hong Kong’s largest export trading company, in an interview with the Harvard Business Review:
Say we get an order from a European retailer to produce 10,000 garments. . . . For this customer we might decide to buy yarn from a Korean producer but have it woven and dyed in Taiwan. The Japanese have the best zippers and buttons, but they manufacture them mostly in China. Okay, so we go to YKK, a big Japanese zipper manufacturer, and we order the right zipper from their Chinese plants. Then we determine that, because of quotas and labor conditions, the best place to make the garment is Thailand. So we ship everything there. And because the customer needs quick delivery, we may divide the order across five factories in Thailand. . . . Five weeks after we have received the order, 10,000 garments arrive on the shelves in Europe, all looking like they came from one factory with colors, for example, perfectly matched.3
The term supply chain is a simplification of the supply web or network of suppliers, manufacturing plants, retailers, and the myriad supporting companies involved in design, procurement, manufacturing, storing, shipping, selling, and servicing goods. Figure 5.3 depicts a stylized supply network and the main flows associated with such a network.
These flows comprise the movement of parts and products downstream (from the suppliers to the retailers) and the flow of money upstream. These two flows define the supplier-customer relationship. Just as important is the flow of information; customers send orders and product requirements upstream and suppliers broadcast their inventory levels and their actions (such as shipping notices and invoices) downstream to their customers. Thus, information flows both ways.4
When Ford managed its network of factories at the Rouge plant, there was little focus on inventory management across all stages in Ford’s internal chain. Because Ford’s revenue materialized only when cars were sold, the company emphasized availability of the finished product. It was a cardinal sin for plant managers not to have enough parts or raw material on hand, causing a production delay and lost revenue. Consequently, Ford’s various subassembly and parts manufacturing units kept a large inventory of parts, subassemblies, and finished products. The same inventory management mentality prevailed though the 1970s until Toyota demonstrated the cost and quality advantages of lean manufacturing systems: relying on little inventory; just-in-time deliveries; strong relationships with suppliers; concurrent processes of product design, manufacturing and distribution; continuous quality improvements; and a culture of employee empowerment.
One of the fundamental tenets of modern supply chain management is the focus on controlling inventory levels. The role of inventory, and how companies manage it, is outlined in the next section.
Firms have no intrinsic desire to have inventory of material, parts, or finished goods. Inventory requires management attention and costs money for each day it is held; so the question is, why hold inventory at all?
Some types of inventory cannot be avoided. Others serve a useful role that should be balanced against the inventory holding costs. Work-in-process (WIP) inventory cannot be avoided. WIP is the inventory of material while it is undergoing any process that adds value to it. Thus, the material on the production line, while undergoing the manufacturing process, is WIP inventory.
Typically, however, the time that materials and goods spend in any value-adding process is dwarfed by the time they spend between processes. For example, DuPont Inc. estimated that in 1993 it took about 168 days to get from raw material to a finished product in the supply chain for yarn. Unfortunately, value was actually added to the material only during eight hours of this time. The rest of the time the material was waiting for the next process to take place.
Inventory, however, does have an important function. Inventory’s role is to decouple processes, disconnecting the various supply chain processes from each other. Decoupling means that processes can operate independently of each other; a tire factory can make tires when it is most efficient and the car factory can use the tires when it is most convenient. Without inventory, the two processes would need to operate synchronously; four tires (of the right size) would roll off the tire-making line just in time to be mounted onto cars in the assembly line.
Decoupling is required because the processes that constitute the supply chain have different characteristics and economies. Thus, to operate well, they need to run independently from the processes that precede or follow them. In the last example, the tire and automobile manufacturers may in fact synchronize their manufacturing processes. The difficulty for the tire manufacturer is that it must synchronize with many automobile manufacturers and also produce for the aftermarket. Thus, a complete synchronization with a single customer may increase its overall costs.
The problem is more obvious when selling to consumers, where it is impossible to coordinate the supply and the demand perfectly. In order to minimize inventory, just as a consumer would come to the store to buy a single package of detergent, a shipment of a single package would be delivered to the supermarket. But consumers do not notify the supermarket when they intend to go shopping and for what product. Furthermore, even if they did, it would be uneconomical to ship detergent in single packages from the factory to the supermarket. So, supermarkets keep inventory on the shelves just in case consumers may require it, thus decoupling the process of procurement and receipt of goods from the process of selling them to consumers. Similarly, a manufacturer may keep an inventory of raw material on hand just in case it gets an unexpected order and needs to use it in the manufacturing process. Given estimates of demand randomness and the costs of ordering and holding inventory, classical inventory theory gives procurement managers the tools to optimize the orders so as to minimize their total expected costs, including the cost of lost sales.
In the mid-1990s, the Swedish car manufacturer Volvo had excessive stocks of green-colored cars. To move them along, the sales and marketing department began offering attractive special deals, even selling at a loss, just to clear the inventory. So green cars started selling, but nobody had told the manufacturing department about the promotions. The latter noted the increase in sales, read it as a sign that consumers had started to like green, and ramped up production of green cars, exacerbating the problem and hurting profitability.5
In many cases, coordination is even more challenging when two companies are involved rather than two departments of the same company. Since it takes time from receipt of an order to delivery, companies need to forecast future demand at the time of placing the order. They need to communicate their material requirements to their suppliers, who, in turn, need to communicate and order from their suppliers further “upstream” in the supply chain. The forecasting challenge is discussed further in chapter 6.
Forecasting demand, however, is not the only challenge facing supply chain operations. Some of the challenges are structural.
Most babies, oblivious to supply chain management issues, use diapers at a steady rate. Week to week, the number of diapers purchased by parents at Wal-Mart, K-Mart, Carrefour, or El Corte Inglés S.A. is, essentially, constant. In the early 1990s, however, Procter & Gamble, the manufacturer of Pampers, Luvs, and Hipoglos diapers, detected some puzzling sales patterns. Despite the stable birth rate in the United States, and the steady rate of babies’ usage of diapers, P&G experienced significant fluctuations of orders from distributors to its factories. In fact, P&G’s own orders to its material suppliers, like 3M, fluctuated even more.6 It soon became clear that orders and inventory levels were subject to greater and greater variations as one moves further and further up the supply chain—from consumers to retailers to distributors to manufacturers to suppliers and their suppliers.
P&G called this phenomenon, depicted in figure 5.4, the bullwhip effect, denoting the increased amplitude of orders and the increased fluctuations of inventory levels the further one moves upstream in the supply chain. Hewlett-Packard (HP) uncovered a similar pattern when it examined the orders and sales rates of its printers at Office Depot. While sales at the office supplies store exhibited only mild variations over time, the orders that Office Depot placed on HP fluctuated much more. Furthermore, the orders from HP’s printer division to its integrated circuit division had even greater fluctuations. Even Barilla, the Italian pasta manufacturer, has experienced widely fluctuating orders from its distributors in the 1980s, despite the constant level at which Italian households consume pasta. This led to random fluctuations in the distributors’ inventory levels and 6 to 7 percent stock-outs.7
Several factors contribute to the bullwhip effect. The main factor is the lack of coordination between buyers and sellers along the supply chain. For example, a distributor may perceive a small increase in a retailer’s order as an indication of future demand growth, rather than a temporary fluctuation. Anticipating that future orders may be even higher, the distributor orders too much from the manufacturer—not only enough to fulfill the order at hand but also to be ready for furture growth in the retailer’s orders. The manufacturer now has a magnified order signal and, forecasting even higher future orders, may order an even larger amount from its suppliers. When the higher demand does not materialize at the retail level, the distributor ends up with too much inventory and then cuts back on orders, and even stops ordering completely, and the process repeats up the chain. Orders and inventory then fluctuate with ever-greater amplitude upstream in the supply chain as each echelon misinterprets the order signal.
Even worse, in many cases the sparks that trigger the fluctuations are set on purpose. For example, promotions and discounts cause customers to buy more during the promotions and buy less later. In fact, in the 1980s, Wal-Mart was known for taking advantage of local promotions that P&G was running. When P&G offered, say, a 10-cents-off Tide promotion in the New York region, Wal-Mart would buy all of the Tide it needed nationally in New York and then truck it to its stores in other regions.
The batching of orders and shipments can also start the fluctuations. Many companies order only once a month or once every two weeks, creating an order “spike” for their suppliers. In addition, suppliers’ salespeople, motivated by quarterly bonuses, typically create incentives for their customers to order more than they need toward the end of each quarter, causing over-supply in the last month of the quarter followed by low order volume in the next month.
A somewhat different mechanism starts the bullwhip when products are in short supply, for example during the introduction of a new “hot” product such as the Sony Play Station or the Volkswagen Beetle. In these cases, manufacturers may put retailers “on allocation,” meaning that each retailer gets only a fraction of what it orders. Cognizant of the practice, the retailers double and triple their orders; even though they know that they do not need and cannot sell the additional products (known in the trade as “phantom” orders). The manufacturer, facing even greater demand for the product than it anticipated may invest further to increase production. Unfortunately, it is bound to be disappointed when supply catches up with the real demand and the phantom orders disappear, leaving the manufacturer with extra product that needs to be sold at lower margins.
Overcoming the bullwhip effect requires careful coordination and continuous communication between companies and within departments and business units of the same company. Some of the programs used by industry to enhance communication and otherwise mitigate the bullwhip effect are described later in this chapter.
The bullwhip effect underscores the importance of continuous communication along the supply chain in case of a disruption. Companies may order more from one supplier when another supplier is disrupted, or they may order more in anticipation of a labor strike or a port closure. If the suppliers do not get information about the temporary nature of these orders, they may assume that such orders indicate an increase in demand. They may, in turn, increase capacity or order even more from their own suppliers in anticipation, thus generating a strong bullwhip effect. The same phenomenon can take place when a company reduces its orders for certain materials or parts because of a temporary disruption. Such disruptions include plant closures for refurbishing (or new model introduction), industrial accidents, or work stoppages caused by strikes. If the temporary nature of such disruption is not communicated to suppliers, they may over-react and shift capacity to other uses or stop ordering parts from their own suppliers.
Most large manufacturers did not awaken to the potential cost savings associated with inventory reductions until Toyota Motor Company demonstrated, in the 1980s, the magnitude of these savings and the range of other benefits associated with lean manufacturing. With its vaunted just-in-time system, Toyota coordinated activities throughout the supply chain, reducing both inventories and the bullwhip effect. Even more important, Toyota demonstrated that inventories mask quality problems and that lower inventories lead to higher product quality, thus exposing the hidden cost of carrying inventory.
The realization that inventory costs are much higher than initially thought pushed many other companies to integrate their supply chain more tightly, minimizing the inventory between the various stages. A tight supply chain relies on coordinated connections and synchronized processes between trading partners along the supply chain and it operates with less inventory.
Some approaches to channel coordination—getting all participants to work together to eliminate the costly fluctuations of orders and inventories—try to eliminate some of the middleman functions. For example, in Vendor-Managed-Inventory (VMI) programs, the retailers give manufacturers the point-of-sale data and the manufacturer, in turn, is responsible for replenishing the retailer’s shelves. This eliminates the need for the retailer to forecast and order, reducing the chance of order amplification and eliminating a whole echelon of inventory. Barilla used a similar approach in its just-in-time delivery program. Rather than respond to distributors’ orders, the company shipped to the distributors just the amount they shipped to the retailers, preventing the distributors from accumulating inventories and amplifying orders. The program successfully reduced distributors’ inventories by 50 percent and, at the same time, practically eliminated stock-outs.8
Wal-Mart developed cross-docking processes in which goods come from the suppliers to Wal-Mart’s distribution centers (DCs) and are moved immediately to the stores. This process reduced DC warehousing costs and, more important, eliminated an echelon in the distribution chain. Rather than ordering from the store to the DC and from the DC to the suppliers, the stores were, in fact, ordering directly from the suppliers.
Bose Inc., the Massachusetts manufacturer of high fidelity audio systems, achieved a similar end result by bringing key suppliers’ representatives in house and giving them authority to function as an integral part of the Bose material and purchasing systems. The process replaced traditional buyers, planners, and salespeople with “in-plant” supplier personnel responsible for ordering material, parts, and services from their parent companies, according to Bose’s needs. The elimination of suppliers’ sales people and buyers’ procurement personnel resulted in lower inventory of parts and raw material, as well as more responsive suppliers.9
The Japanese concept of kanban cards—cards signaling that a particular part is needed—helps control inventory by eliminating the bullwhip effect through control of the ordering process; each stage in the manufacturing process can get the part that it needs only by releasing a kanban card upstream. The total number of kanban cards in the system is controlled, thus ensuring that forecasts are not inflated; each stage in the manufacturing process can order only as many parts as it uses and only when it needs them. This is the underpinning of the “lean” manufacturing systems. Material arrives just in time to be used in the next stage of production, rather than wait just in case it is needed.
There are many other ways in which supply chain participants have cooperated to tighten their supply chains, creating just-intime flow of material and better information sharing in their quest to reduce inventory without reducing service levels. All these processes are based on similar ideas: Reduction in the number of decision-making echelons and tight coordination and synchronization, leading to lower inventory levels and improved service. The specific programs include Continuous Replenishment Programs (CRP) and Efficient Consumer Response (ECR) used in the grocery industry; Quick Response (QR) programs used in the apparel industry; and Collaborative Planning Forecasting and Replenishment (CPFR) used in the consumer-packaged goods industry.
Many modern-day manufacturing plants operate with only a few hours’ worth of safety stock, relying on constant communication with their parts suppliers, transportation carriers, and warehousing operators to avoid parts’ stock-out. But by removing inventory from the supply chain, every participant becomes more dependent on others. Furthermore, the system as a whole becomes less resilient. This became painfully clear to manufacturers such as Toyota who, in 1995, had to close most of its manufacturing plant immediately following the Kobe earthquake because of a shortage of break shoes.
To avoid costly delays in a lean manufacturing environment in which little safety stock is kept, many manufacturers levy a significant penalty on transportation and logistics providers for late deliveries. For example, Ryder Integrated Logistics provides justin-time transportation services to the Saturn plant in Spring Hill, Tennessee. It collects information from Saturn regarding each day’s production schedule and delivers exactly the right parts from all the area’s suppliers to the Saturn plant, just before the parts are needed on the production line. The contract between Saturn and Ryder Integrated Logistics specified a $5,000 penalty for each 15 minutes of a delayed delivery. This contract clause was designed as a deterrent; the actual costs Saturn would incur from a plant shutdown are actually higher.
Thus, lean manufacturing systems, with their low inventory levels, imply less redundancy and increased vulnerability. Although companies reduce their everyday costs, they may be increasing certain long-term vulnerabilities.
Some industries, such as fashion apparel and consumer electronics, face large demand fluctuations and uncertainties as part of their everyday business. Demand forecasting is especially challenging in such industries, yet the companies involved are also extremely cost conscious because they face unrelenting global competition. Thus, on the one hand, they cannot reduce their inventories because these inventories are necessary to buffer against demand uncertainty. On the other hand, they cannot build up inventories because the higher costs will render them uncompetitive. The next chapter examines how such firms deal with the need to reduce inventories and costs yet not increase vulnerabilities to demand disruptions. It offers some lessons in managing challenging supply chains, balancing demand and supply—lessons that can be used in preparing for other disruptions.