Supply Chain Structures

The structure of a supply chain is fundamentally a reflection of a firm’s business model. For example, in the personal computer (PC) industry, the supply chain often takes a hybrid form: the components — processor, memory, hard disk, monitors, and other peripherals — are built to stock, whereas the end product, the “box”, is assembled to order. In such a system, the time it takes to assemble the end product is quite negligible (provided all the components are available), while the production / procurement leadtime for each component is more substantial. Hence, by keeping inventory at the component level, customer orders can be filled quickly. On the other hand, postponing the final assembly until order arrival provides a high level of flexibility, in terms of both product variety and risk pooling. Indeed, this hybrid of make-to-stock and assemble-to-order appears to be an ideal business model in providing both mass customization and quick response to order fulfillment.

As computing technologies are revolutionizing business efficiency and productivity and improving quality of life, the personal computer (PC) industry that supports such changes has been struggling through the difficult process of improving its supply chain efficiency.

Companies in many industries have begun to realize that conflicts of interest among the various parties in a supply chain can engender operationally inefficient behavior. Consequently, many researchers have become interested in identifying and evaluating methods of coordinating supply chains in which multiple decision makers pursue individual agendas (cf. [32]). The typical approach in the OM literature is to partition a traditional inventory model into a number of subproblems, each representing the decisions and objectives of a distinct organization. Most commonly, the supply chain is assumed to contain just two firms, e.g., a manufacturer and a retailer. The analysis then proceeds to pinpoint the root causes of inefficiency, and recommend mechanisms for appropriately adjusting individual incentives.

The basis of competition in the automotive industry is changing. While product innovation and styling remain the most important areas of competition, an almost equally fierce battle is now developing in the areas of customization and order fulfillment (Stalk, Stephenson and King [35]). Currently, all models of vehicle distribution are fundamentally inventory-driven and do not promote customized ordering. However, several vehicle manufacturers — most notably Ford and General Motors — have recently launched initiatives to transform their companies from predominantly make-to-stock to predominantly make-to-order producers. This will enable vehicle manufacturers and their dealers not only to dramatically reduce their finished goods inventory but also respond to challenges and threats from third party Internet companies.

The subject of information sharing has regained the interests of both academics and practitioners. Today a huge amount of information is being interchanged between manufacturers and retailers, between retailers and consumers, between companies and investors, and also among the parties in the same level of a vertical chain. How to measure the gains and losses of such activities to the parties involved is an important issue in supply chain management. This chapter studies the incentives for firms to share information vertically in the presence of horizontal competition. We do so in a setting with one manufacturer and many competing retailers where each retailer possesses some private information about the downstream market demand or about its own cost. Horizontal competition in the downstream brings about new effects of information sharing. In general, vertical information sharing, e.g., transmission of pointof-sales data between a retailer and a manufacturer, has two effects, the “direct effect” on the payoffs between the parties engaged in information sharing, and the “indirect effect” of information sharing on other competing firms. For example, knowing that the manufacturer receives some information from a retailer, other retailers may respond to the fact by changing their strategies, and such reaction may cause additional gains or losses to the parties directly engaged in information sharing.

Spicer Off -Highway Products Division (a division of Dana Corporation) offers their customers a full range of powershift transmissions and torque convertors. The structure of the manufacturing process is typically a hybrid of make-to-stock and make/assemble-to-order. The main problem is that the production lead time for the make parts is substantial due to the large variety of components and the complexity of the production process. Consequently, a large number of manufacturing operations are performed in a highly uncertain environment inevitably resulting in mismatches at the time of the actual customer commitment.

The confluence of several trends in manufacturing organizations has created many opportunities for supply chain management practitioners to apply their skills. These trends include the increasingly ubiquitous implementation of Advanced Planning Systems, the growing complexity of supply chains, and the recognition among managers of the importance of supply chain costs to profitability.

Properly managing inventory is a challenging but critical problem for high-technology manufacturers. Due to rapidly declining manufacturing costs and high obsolescence risk, holding inventory is very expensive and risky. Customers often require near immediate availability of a large variety of products with highly uncertain demand and long lead times. Thus, manufacturers must hold high levels of inventory despite the expense and risk. Xilinx, a firm that produces programmable logic integrated circuits (ICs) for sale to original equipment manufacturers, faces this situation. In the programmable logic industry, manufacturing costs decline at rates of anywhere from 20% to 75% annually, and obsolescence costs are on the order of 5–10% of the gross inventory. As of 1999, Xilinx had upwards of 10,000 different finished products, only about 5% of which would be classified as having stable demand; total manufacturing lead times were 2–3 months; and a significant fraction of customer orders were requests to be filled same-day. A high level of serviceability is critical since although an IC may only be one component of hundreds in a customer’s product, delays in delivery may shutdown a production line. Thus, service is a critical characteristic in supplier selection and retention. To meet this high level of service with long lead times and uncertain demands, leading programmable logic manufacturers in 1998 were holding very large inventories, 80 to 150 dollar days-of-inventory, despite the risk and expense. (The inventory measured in dollar days is the net inventory divided by the average cost of goods sold per day.)

Supply chain design and management has recently gained unprecedented recognition as a vital factor in the success of a business. Among the most fundamental characteristics that determine the performance of a supply chain are capacity, inventory, delivery leadtime, and customer service level. The first two quantities are under direct management control and are associated with the cost in a supply chain; the last two are the effects of managerial decisions and are associated with customer expectation or requirement. While the qualitative relationship among them is often obvious, understanding their quantitative relationship is much more difficult yet increasingly important in today’s business environment. This chapter intends to introduce a set of tools and results to help better understand the quantitative relationship among these fundamental measures and make supply chain decisions more effectively.

In this chapter we study the dependence properties of multi-product, assemble-to-order (ATO) systems with either capacitated or uncapacitated suppliers. The aim is to gain insights into the impact of demand correlation on system performance and to develop easily computable, tight bounds for key performance measures.

We consider a multi-item inventory system of the following sort. Demands arrive randomly in the form of requests for subsets of the multiple items. A demand is met only when its entire subset of items is available. The key difference between this system and many typical inventory models is that it is the number of demands met (or not met) that is important, rather than the number of items supplied (or short).