|
...products volatile and difficult to forecast; the rapid rate of technology innovation causes short product lifecycles, low production yield and, oftentimes, long production lead time, all of which hamper the firm's ability to respond to market changes. Uncertain economic times exacerbate these challenges. Whereas in an environment of sustained demand growth, firms might build inventory or hold excess capacity to buffer against demand variability, most are reluctant, or unable, to assume such financial risks in a downside market. Nevertheless, high-tech companies recognize that in order to sustain their customer base and to seize revenue opportunities, they must be able to manage successive technological innovations effectively, e.g., introducing high-margin innovative products at the right moment while maximizing the return-on-investment for older, more mature products. To do so, firms must structure capacities in their supply chain so that over time it is possible to respond to demand surge from new product introduction and market upside, and to absorb short-term decline due to technological migration and market downside.
The role of capacity management is even more important in industries in which capital equipment cost is high. For example, in the semiconductor industry, manufacturers are faced with astronomical capacity costs, long capacity lead times, high obsolescence rates and high demand volatility. A new semiconductor fab costs $1-4 billion to build, and the price for a single machine may be as high as $4-5 million. Moreover, the rapid technology innovation leads to short product lifecycles and thus to higher obsolescence rates and increased equipment usage costs. To make the situation worse, the demand variability during a particular quarter may peak above 80% of the average sales and, according to the Semiconductor Industry Association, the equipment procurement lead times are usually as long as 6-12 months. This means that the demand beyond the capacity lead times is highly uncertain. The above environment drives semiconductor manufacturers to adopt exceedingly conservative capacity expansion policies (Erkoc & Wu, 2004). However, in a fast-growing global market, the conservative capacity-expansion policy leads to severe shortfalls in service levels. In a recent survey of managers by the Fabless Semiconductor Association (FSA), the respondents emphasize significant concerns about whether the manufacturers (foundries) will be able to supply wafers to meet demand. The most common reason cited in the survey for missing forecasts is the lack of adequate capacity from the foundries rather than internal issues, such as the lack of a specific technology or process (Ridsdale, 2000).
A similar phenomenon is also observed in the biotech industry. Many bio-drugs on the market require such high volumes of manufacturing capacity that capacity is always in shortage. Recent surveys reveal that 50% of the contractors in this sector believe that the general capacity shortage is unlikely to improve soon; this is due to the current business practice that puts most of the capacity expansion risk on the manufacturer's shoulders (Fox et al., 2001). A typical manufacturing facility in the biotech sector costs between $200 million and $400 million to build, a significant risk to bear, especially when a drug has yet to receive regulatory approval. To address these challenges, many drug manufacturers have begun to work with their pharmaceutical and biotechnology customers to forge long-term commitments and co-investment plans. An emerging trend in the industry is that major drug developers begin to co-invest facilities and equipment with their manufacturing partners in exchange for guaranteed (capacity) slots.
In this article, we review literature relevant to high-tech capacity planning and management. Using the competitive operational environment of the high-tech industry as the backdrop, we examine the impact of capacity from strategic as well as tactical and operational perspectives. The papers we review are not limited to applications in the context of high-tech industries but they are, in our opinion, representative of critical research ingredients in this area.
KEY RESEARCH ISSUES FOR CAPACITY PLANNING
We consider research issues for capacity planning from the strategic, tactical and operational levels. At the strategic level, capacity planning involves not only the firm's own capacity investment, but also its supply chain partners' investments. The capacity investment of one firm in the supply chain could have enormous impact on the performances of all upstream and downstream firms; thus, strategic interactions between two or more players need to be taken into account. In addition to monolithic models that employ tools such as expected utility theory and dynamic programming, the literature increasingly considers settings that model independent multiple decision makers in the context of supply chain management. Research in this area utilizes game theoretic models focusing on issues such as contracting, coordination and risk-sharing mechanisms.
At the tactical level, capacity planning focuses on capacity expansion tactics as related to the operational aspects of the firm. A comprehensive survey of the OM literature on the size, type, and timing of capacity investments is given by van Mieghem (2003). In this paper, we focus our examination of tactical capacity expansion literature on three key aspects. First, we review OM literature that incorporates characteristics of the high-tech industry in the traditional production, inventory and demand management models. Second, we review the growing literature on real options as related to high-tech capacity planning. In recent years, real options have become quite popular as a means for modeling capacity investment risks based on endogenous and exogenous factors. Third, we survey papers that examine risk sharing and vertical integration between suppliers and buyers through capacity reservation contracts.
At the operational level, capacity planning typically refers to decision support models developed for a specific operational environment. There is a significant literature for operational capacity planning in the semiconductor industry, which will be our main focus. We classify operational models in semiconductor manufacturing according to the level of detail that they capture and the length of the planning horizon that they consider.
STRATEGIC AND TACTICAL MODELS: GAME THEORETIC AND ECONOMIC ANALYSIS
Capacity Planning with Production, Inventory and Demand Management Perspectives
Demand uncertainty and the short product lifecycles of high-tech products are two key factors that influence capacity expansion models. High-tech manufacturers avoid carrying inventory due to high obsolescence rates. In fact, high-tech products are often treated as perishable goods. Inventory models developed in this context typically use news vendor or news vendor networks settings with single-period and stochastic demand. These models consider capacity investment by a single or multiple independent decision makers in a stationary environment; once capacity is built it stays unchanged during the planning horizon. The profit to the firm is modeled as a function of the capacity level, K, and the state of the world (e.g., realized demand), [xi]. Extensions to news vendor models study multi-period settings with capacity investment adjustments over time. In this setting, capacity at time t - 1, [K.sub.t-1], is adjusted to [K.sub.t] at time t at some cost. If the investment is (fully or partially) reversible, then contraction (i.e., [K.sub.t] < [K.sub.t-1]) is possible. Otherwise, with irreversible investment the firm either expands or maintains its current capacity across periods.
Newsvendor-Style Models
This literature employs aggregate planning for the acquisition and allocation of resources to satisfy customer demand over a specific time period. The short product lifecycles in high-tech make such approach quite appropriate. A few papers in this category investigate capacity expansion and investment strategies jointly with inventory management and/or outsourcing policies. Bradley and Arntzen (1999), observe that firms achieve better financial results by optimizing their capacity and production/inventory decisions simultaneously. They demonstrate their result using a case study performed at an electronics firm. With the increasing pace of technological innovation and the increasing cost of manufacturing equipment, many OEMs are reluctant to respond to economic cycles by adjusting their own in-house capacity. Consequently, capacity outsourcing has become an integral part of capacity investment decisions (Mason et al., 2002). Atamturk and Hochbaum (2001), propose a four-way tradeoff among capacity, production, subcontracting, and inventory levels over a finite horizon. Kouvelis and Milner (2002) consider two-stage supply chains and analyze the impact of supply/demand uncertainty on capacity and outsourcing decisions. They conclude that greater supply uncertainty encourages vertical integration, because the OEMs have incentives to make investments in their suppliers to ensure reliable and continuous supply. In contrast, outsourcing becomes more attractive as uncertainty in demand increases.
Pindyck (1993) shows that demand uncertainty can discourage firms from capacity expansion when there is perfect competition; while Kulatilaka and Perotti (1998) show that higher uncertainty may increase the firm's incentive to invest when there is imperfect competition. Van Mieghem (1999) studies the trade off among capacity investment, production and subcontracting in a two-stage, two-player, two-market setting. He models the interactions of a manufacturer's and a subcontractor's decisions. He observes that the manufacturer subcontracts more (invests less on his own capacity) as the demand uncertainty increases, which induces the the subcontractor to invests more. Under a similar setting, Tan (2004) investigates capacity investment and pricing decisions for a manufacturer and a subcontractor with guaranteed availability. Van Mieghem and Dada (1999) examine the interplay among capacity, inventory and pricing decisions. The authors study the impact of the timing on these decisions and on the firm's profitability; they examine different market settings such as monopoly, oligopoly and perfect competition Other news vendor-like game theoretical models study sizing (Bernstein & DeCroix, 2004), timing (Ferguson, DeCroix, & Zipkin, 2002), and allocation (Karabuk & Wu, 2003; Mallik & Harker, 2004) of capacity under competitive settings.
Most of the above papers consider a single product and resource type. In reality, however, different products may share the same resource, and different types of resources may be needed to process a particular product. A line of literature considers the presence of multiple resources where decisions must be made for the configuration and selection of optimal resource types. This results in a network design problem. Few researchers tackle the capacity expansion problem with multiple products and multiple resources; however, there is growing interests in models that consider more general product/resource settings. In an early paper, Dixit (1997) discusses optimal investment policies in a two-resource setting. Harrison and van Mieghem (1999) propose a product-mix linear programming model and use its optimal shadow prices to extend the classical news vendor model to a so-called multidimensional news vendor solution with multiple resources and multiple product types, which leads to the well-known critical fractile solution. The critical fractile values balance overage costs with underage costs and are computed using shadow prices. Using the multidimensional news vendor approach, Van Mieghem (1998) studies a two-product setting in which the firm has the option to invest in two product-dedicated resources or one flexible resource that can process both products. The paper examines the impact of price, cost, demand uncertainty and demand correlations on the investment decisions. Later, van Mieghem and Rudi (2002) extend these models to news vendor networks that incorporate multiple products, multiple resources and multiple storage points. They observe that when demand is normally distributed, the optimal expected investment value is an increasing function of the demand vector and a decreasing function of any variance term. Bish and Wang (2004) also use news vendor networks to investigate investment policies for product-flexible versus dedicated resources in a two-product setting with correlated demand.
Multi-Period Models with Capacity Adjustments
Multi-period or dynamic capacity expansion models determine policies that specify the timing and scope of capacity adjustment so as to maximize the expected net present value of the firm's investment. These models seek answers for when and how much capacity to build in a dynamically changing environment. Capacity decisions are strongly influenced by the length of the planning horizon and the associated rate of depreciation, as well as by the cost of investment and the demand uncertainties during the aggregated planning period. Most papers in this area utilize stochastic dynamic programming models. Van Mieghem (2003) points out that there are three main challenges in capacity cost modeling: indivisibility, irreversibility and nonconvexity. Indivisibility implies lumpy capacity expansions, which are common in high-tech applications. Due to rapid technological innovation, almost all capacity investment in high-tech manufacturing is irreversible (in the sense that capacity expansion cannot be undone without significant cost). Irreversibility contributes to nonlinearity in capacity expansion decisions since it might prevent firms from making downward capacity adjustments to demand changes from one period to another, leading to 'no action' policies in some periods. Moreover, one may not be able to assume convex capacity costs due to fixed-cost, economies-of-scale, etc., adding another layer of complexity to the problem.
Research on dynamic capacity models with stochastic demand goes back to the seminal work of Manne (1961), in which he models demand growth using Brownian motion with positive drift for a single-resource system. The resulting regenerative process leads to uniform capacity increments that take place whenever the demand backlog goes beyond a threshold value. The timing of the expansion is modeled by the 'hitting time' of the Brownian motion. The author shows that by appropriately choosing the discount rate based on demand variance, a deterministic equivalent model can be built to solve the stochastic problem. In Eberly and van Mieghem (1997), the authors propose a more general continuous-time model that considers...
NOTE: All illustrations and photos
have been removed from this article.
More articles from Engineering Economist
On the validity of the geometric Brownian motion assumption., June 22, 2005
A virtual classroom experiment for teaching engineering economy., December 22, 2004
Technical note: comparative static analysis of information value in a ..., December 22, 2004
Technical note: approximation solution for the inventory model with ra..., December 22, 2004
Minutes of the 2004 IIE engineering economy division business meeting., December 22, 2004
Looking for additional articles?
Search our database of over 3 million articles.
Looking for more in-depth information on this industry?
Search our complete database of Industry & Market reports by text, subject, publication
name or publication date.
About Goliath
Whether you're looking for sales prospects, competitive information, company
analysis or best practices in managing your organization,
Goliath can help you meet your business needs.
Our extensive business information databases empower business
professionals with both the breadth and depth of credible,
authoritative information they need to support their business
goals. Whether it be strategic planning, sales prospecting,
company research or defining management best practices -
Goliath is your leading source for accurate information.
|
