Friday, August 16, 2019
Postponement
Postponement Strategy Materials Management End Term Project (Term ââ¬âIV) Date: 15th September 2009 Submitted To: Submitted By: Prof. Vivek Kumar Namrata Agarwal(81031) Prof. Kaushik Paul Neha Gupta(81034) Contents Chapter 14 Introduction4 1. 1 What is Postponement? 4 1. 2 A specific example5 1. 3 Postponement in operation7 Chapter 29 Literature Review9 Chapter 314 When is Postponement Appropriate? 14 3. 1 The Postponement/Speculation (P/S) Matrix14 3. 2 Costs & Benefits of Postponement15 3. 2. 1 More variety15 3. 2. 2 Inventory reduction18 3. 2. 3 Better forecast accuracy19 . 2. 4 Inventory cost reduction20 3. 2. 5 Logistics cost reduction22 3. 2. 6 Improved customer service levels22 3. 2. 7 Increased product development cost23 3. 2. 8 Increased manufacturing cost23 Chapter 424 Case Studies24 4. 1 Automobile Manufacturing: GM24 4. 2 Aircraft Manufacturing: Embraer26 4. 3 Clinical Equipments: Dade Behring29 4. 4 Sports Goods Manufacturing: Reebok32 4. 5 Xilinx34 Chapter 536 Concl usion36 Chapter 638 Future of postponement38 6. 1 Services and postponement38 References40 Chapter 1 Introduction Over the past 2 decades, logistics activities have gained increasing strategic importance for most companies. Fixed costs of production have increased, consumer demands have become more complex and are harder to predict, both in time and place. Technology is rapidly changing and product life cycles have shortened while product range has increased. Now more than ever, companies are faced with the challenge of producing an increasingly large variety of products in a responsive manner while keeping materials and inventory to a minimum. These issues represent significant challenges for companies producing and selling in a variety of international markets. Not only does demand vary from country to country, but products need to be altered for different markets in consideration of differences in language, culture and local standards. Increasingly, companies are using a strategy known as postponement or mass customization to improve customer service and minimize the risks associated with making different products in different countries. This paper presents a framework for understanding postponement and how it can be implemented. Also, with the help of successful case studies potential savings as well challenges in implementation will be highlighted. . 1 What is Postponement? The term postponement refers to delayed decision-making about a product. It is beneficial to delay commitment to product-specific characteristics as late as possible in order to avoid a mismatch between orders and inventory on hand. The length of delay is specific to a product but the common strategic motivation is to gain better information about customer demand by w aiting to customize a product for a particular market or customer. At the point of postponement a standardized module or platform starts to acquire customer or market specific characteristics. Figure 1-1 shows the spectrum of opportunities for postponement that extends from procurement to distribution. The point of postponement can occur as early as the design phase and as late as packaging and distribution. Postponement at the manufacturing stage has arguably the most potential for cost savings in inventory due to risk pooling. Other points of differentiation can occur in the assembly, labeling, packaging, or distribution phases. Some postponement can even occur after the point of sale in the form of service offerings. [pic] Figure 1-1: Possible points of differentiation in the supply chain Postponement enables forecasters to make better predictions about end product demand over time since the standard module is built-to-forecast and the finished product is built to a better forecast or even built-to-order. Lee and Whang [20] observe that shorter the time horizon over which predictions are made, the more accurate the forecast. The benefits are better end product forecasts and the ability to respond quickly to demand signals by holding unfinished goods in inventory awaiting final assembly or customization. Postponement also creates opportunities to lower inventory costs due to risk pooling because goods are kept in unfinished or component form and can be used to assemble more than one type of finished goods. The monetary value of an unfinished good is less because it is not committed to becoming a finished product and lacks the added value gained in final assembly. 1. 2 A specific example Consider a common case of postponement involving a fast food restaurant. Burger King started a trend with the ââ¬Å"have it your wayâ⬠marketing jingle as a way of advertising the value of getting a customized sandwich ââ¬â fast! This strategy ensured the customer that each order would be made individually at the time of purchase ââ¬â not taken from a batch of pre-made products. In a restaurant, ingredients are ordered in aggregate because it is not known what the final customer orders will be. Ingredients that are common to all sandwiches, like buns and lettuce, are ordered based on a total forecast of sales for each type of sandwich. Having a bun and lettuce ready and waiting for final assembly is the ââ¬Å"platformâ⬠for the sandwich. The rest of the ingredients, like cheese, meat, and pickles, are components that are specific to each end product. If more of one type of sandwich is ordered or less of another, the total number of buns is not affected by this deviation in demand, however, the amount of cheese would be. It is much less costly to throw out a piece of cheese and use the platform for another order than to throw out an entire sandwich. At Burger King, inventory is managed at the aggregate level. There are four choices of meat and three different types of bun. In addition to buns and meat, there is the choice of cheese, bacon, lettuce, tomato, pickles and onion. In total Burger King can produce 768 different sandwiches as show in Table 1. 1. They know that it is costly to try and predict individual customerââ¬â¢s preferences so they aggregate orders into common platforms which consist of a bun, patty (chicken, beef, fish, or veggie) and lettuce, reducing the options from 768 to 128. Once common components are paired together in a platform, the number of options reduces dramatically because variety is determined by multiplying the number of options together. Once the platform is specified by a customer the rest of the sandwich is made-to-order. Component |Number of Options | |Patty |4 | |Buns |3 | |Cheese |2 | |Bacon |2 | |Lettuce |2 | |Tomato |2 | |Onion |2 | |Pickle |2 | |Total Combinations |768 | Table 1. 1: Sandwich options at Burger King This example illustrates how postponement through platform design and stocking individual components instead of finished goods is able to mitigate the risk associated with producing a wide variety of products. This concept can be applied to more than just food. Examples of products which can benefit from postponement include consumer appliances, automobiles, apparel, and even airplanes. These products have one or more of the following characteristics: high degree of forecast uncertainty, modularity, and high inventory carrying costs. 1. 3 Postponement in operation Operational postponement can be applied in one of two ways ââ¬â manufacturing and assembly postponement and logistical postponement. Manufacturing and assembly postponement involves the engineering of a product as a module or platform which can take on several different features thereby increasing the variety of end products. The point of postponement can occur as early as the design phase. The intermediate product is stored in inventory and awaits customization. The value added through assembly or manufacturing may be performed at a finishing facility or at a warehouse just before shipping. Manufacturing and assembly postponement involve decisions made while the product is in production. Engineers seek to design a product as a module or platform which can accept different attachments or features in order to transform the appearance and or function to increase product variety. This concept was referred to as a ââ¬Å"vanilla boxâ⬠by Swaminathan and Tayur [42] because the generic platform is one without any customized value and is therefore the common denominator among a family of different products. Logistical postponement takes into account all other types of postponement involving logistical decisions like packaging, labeling, and distribution. Packaging and labeling postponement traditionally applies to small consumer goods products like razors, batteries, compact disks, film, and snack foods. Large retailers like Wal-Mart and Target require different configurations of packages to accommodate their customer demand and shelf space capacity and to differentiate commodity products. Gillette is well known for their packaging postponement operations. In 1996, Gillette decided to outsource the packaging of their health and beauty items to Sonoco. Bulk quantities of products are sent to Sonoco to await final packaging. Once orders are received appropriate packaging configurations are assembled and shipped to retailers. Manufacturers spend a significant amount of capital and labor trying to satisfy the variability in demand for different configurations for their retailers. However, companies like Gillette, that focus on their core competency, innovating and manufacturing razors and razor blades, push the risk onto their packaging supplier. Sonoco assumes the risk of forecasting for the different retailers which allows Gillette to produce to an aggregate forecast. The benefits for Gillette included a reduction in order fulfillment time from six weeks to one, a 15 percent decrease in packaging inventory, a 10 percent improvement in inventory accuracy, and a 15 percent reduction in packaging costs. Not only does this save Gillette from mismatching demand and configurations, it allows them to focus on engineering, design, and manufacturing of new products instead of packaging. Gillette avoided plant expansion, has a focused factory workforce and is winning favor with retailers by being so responsive. Another example of logistical postponement is the postponement of decisions made about the product during its distribution lead time (from finished product to customer delivery). Whirlpool, a popular manufacturer of household appliances, provides a good example. Customers of Whirlpool include retailers like Sears and Home Depot. Holding inventory of large appliances such as refrigerators and washing machines at local stores is costly because of the high product value and the space taken up in a back storage room. For this reason Whirlpool will send finished goods to a central distribution center and ship directly to the home once a customer order is placed. This method saves the retailer in inventory cost and eliminates additional transportation cost by bypassing the retailer. In addition, it reduces the risk that is inherent in sending a dedicated number of products to individual stores and having to transship orders between retailers. Chapter 2 Literature Review Sources date the idea of postponement as far back as the 1920s and the first use of postponement as a manufacturing strategy as early as the 1950s. Early mention of postponement suggested that costs due to risk and uncertainty were a function of variety and that an efficient means of producing a product is to ââ¬Å"postpone changes in form and identity to the latest point in the marketing flow [and] postpone changes in inventory location to the latest point in timeâ⬠. In 1965, Louis Bucklin recognized that little had been done in the area of postponement despite its tremendous potential for cost savings. He defined total cost as the sum of inventory holding cost and delivery cost, both of which are a function of delivery time. He argued that ââ¬Å"a speculative inventory will appear at each point in a distribution channel whenever its costs are less than the net savings to both buyer and seller from postponementâ⬠. In other words, postponement is not cost effective when there is sufficient information about demand to produce finished goods in mass and store them in inventory. For some products it makes sense to postpone the finishing process by introducing a finishing cost and increasing the delivery time because the product is not readily available from stock. Zinn and Bowersox [50] classified postponement into five distinct types; labeling, packaging, assembly, manufacturing, and time. Labeling postponement assumes that products are standardized until they receive a label distinguishing them by brand. Packaging postponement is best suited for products in which variation is determined by package size. Paint, chemicals, medicine, razors, and many food items sold in bulk are good candidates for packaging postponement. Assembly postponement is applied to products in which variety is based on cosmetic features like cars, iPods, t-shirts, and printers. Hewlett-Packard (HP) provides an excellent example of assembly postponement. Printers designed for different global markets are inherently the same product except for country specific power supply modules, power cord plugs, and instruction manuals. HP makes two types of printers in Vancouver: a US version and a generic version that is customized once it reaches a distribution center in Europe, Asia, or the Pacific based on country specific orders. One benefit is decreased transportation cost because printers are shipped in bulk and are considered â⬠vanillaâ⬠until they receive the value-added accessories like language manual and power supply. Manufacturing postponement occurs when parts are shipped to the finishing center from more than one supplier. It has the greatest potential for cost savings in inventory because the value of the product increases through the addition of each successive component. Manufacturing postponement usually results in higher production costs. The increase is due to the capital cost of switching machinery between different types of variety and shipping them to different finishing facilities. Time postponement occurs when finished products are shipped to centralized warehouses closer to the customer than the manufacturing location. The motivation is to increase customer service levels by decreasing customer lead time and to respond quickly to orders by placing inventories closer to the customer without committing to an individual order. |Postponement Type | Potentially Interested Firms | |Labeling |Several brand names | | |High unit value products | | High product sales fluctuations | |Packaging |Variability in package size | | |High unit value products | | |High product sales fluctuations | |Assembly |Selling products with several versions | | |High volume incurred by packaging | | |High unit value products | | |High product sales fluctuations | |Manufacturing |High proportion of ubiquitous material | | |High unit value products | | |High product sales fluctuations | |Time |High unit value products | | |Large number of distribution warehouses | Table 2. 1: Poten tial Utilization of Postponements The final outcome of their research is a framework which serves to assist managers in determining what type of postponement is best for a given product or supply chain structure. Table 2. 1 shows a list of the postponement types and the firms which would benefit from implementing each type of postponement. Swaminathan and Lee [42] go further and identify the factors which influence the costs and benefits of postponement as market factors, process factors, and product factors. Market factors refer to characteristics of demand and uncertainty. Process factors refer to characteristics of operating policy within the firm as well as the external supply chain, such as managerial support and the location of and relationship with suppliers. Product factors refer to the design and characteristics of an individual product such as integral versus modular and inventory carrying cost. They also highlight enablers of postponement such as process standardization, process resequencing (redesigning the assembly process to move value-added processes closer to the customer), and component standardization. Redesigning products with these characteristics makes postponement possible and reduces the risk to the manufacturer by eliminating redundant processes and designing products to be modular and component interfaces to have standard ports for easy assembly. Alvin Lehnerd and Marc Meyer [21] offer a detailed look at the benefit of engineering products to be platforms for a family of different products. The authors define two terms which are the basis for postponement. â⬠¢ Product platform ââ¬â a set of common components, modules, or parts from which a stream of derivative products can be efficiently created or launched â⬠¢ Product family ââ¬â a set of products that share common technology and address a related set of market applications These are both concepts that Black and Decker (BD) considered when they started to redesign their line of power tools. In the 1970s BD replaced customized parts with standardized components, interfaces, and connections in order to pool the part inventory and save on component inventory costs. Components included common screws, gears, and the motors which powered 122 different power tools. At a cost of $17 million over three years, BD was able to fully integrate its supply chain, reduce scrap rate from six percent to one percent, reduce failure rate from 11 percent to less than five percent, and reduce the selling price by half while still maintaining a 50 percent margin. BD was also able to reduce the number of suppliers and push its competition out of the market. This is one of the first cases of postponement using product platform design. Product platforms are also common in automotive and aircraft design. Lee, Billington, and Carter [20] discuss Hewlett-Packardââ¬â¢s strategy when it created a single platform for its DeskJet Plus, Deskwriter, Deskwriter Appletalk, and the DeskJet 500 series. A major source of variability for HP was the final shipping destination. HP ships its DeskJet Plus printers to North America, Europe, Asia, and the South Pacific. Each one requires a different power supply module and language manual. Under the ââ¬Å"DC-localizationâ⬠initiative printers are shipped from the manufacturing center in Vancouver and arrive at a local distribution center (DC) without language manuals or power supply modules. The DC supplies the remaining country specific features and packages the printers for final sale. This allows HP to pool the risk of stocking inventory by destination. Taking the process a step further, HP realized that Vancouver was close enough to the US where it could act as the local DC and hence two different types of printers are produced; US and non-US versions. This example illustrates how postponement is used in multiple ways for a single product. The designers at HP had to create a printer with a generic power supply port which is a form of assembly/production postponement. The local DCs had the job of assembling a final product complete with instruction manual, power supply, and the appropriate packaging material. Robert Stahl and Thomas Wallace [47] propose a framework for implementing postponement by classifying products according to two factors; product complexity ââ¬â the number of product varieties, and speed ââ¬â the time from customer order to delivery. This results in four levels of differentiation as shown in Figure 2. 1. [pic] Figure 2. 1: Complexity vs. Speed Matrix [47] Companies in each of the four quadrants have different challenges when adopting a postponement strategy. For example, a company in quadrant B produces a product that has very little variety but takes a long time to produce and deliver to the customer. Wallace and Stahl suggest that a company in this quadrant focus on speed by reducing the lead time from suppliers and expedite the delivery to the customer. They can accomplish this by implementing lean manufacturing initiatives, improving the work flow, and reexamining the location of their suppliers in terms of distance to the customer. This dilemma illustrates the trade-off between cost and service level. One way to take advantage of distant suppliers and still achieve fast delivery is to decouple the production process and hold inventory of intermediate product locally. Chapter 3 When is Postponement Appropriate? Postponement has the potential to lower the total delivered cost of a product. However, postponement does come with its own costs to implement and maintain. The benefits outweigh the costs when postponement is implemented correctly for the right type of product. For products with certain characteristics postponement allows companies to offer more variety, improve forecast accuracy, reduce inventories, and improve customer service levels. With these benefits come the costs of implementation and manufacturing. 3. 1 The Postponement/Speculation (P/S) Matrix Pagh and Cooper (1998) developed a simple but very powerful conceptual model to show the range of postponement strategies that could be adopted by companies. Four generic strategies were identified: full speculation, logistics postponement, manufacturing postponement and full postponement. These were presented in the form of a matrix as shown in Figure 3. | |Logistics | | | |Speculation |Postponement | |Manufacturing |Speculation |The full speculation strategy |The logistics postponement strategy | | | |low production and distribution costs |low production costs | | | |high customer service and high inventory |low/medium customer service and inventory | | | |costs |costs | | | | |high dist ribution costs | | |Postponement |The manufacturing postponement strategy |The full postponement strategy | | | |low distribution costs |low inventory costs and customer service | | | |medium to high production costs, |medium/high production costs | | | |inventory costs and customer service|high distribution costs | Figure 3. 1: The P/S Matrix (Source: Pagh & Cooper, 1998) The strategy of full speculation represents a complete reliance on forecasting, where all differentiating manufacturing operations are performed prior to the product being moved to different markets (ââ¬Ëpushââ¬â¢ based system). The strategy of full postponement represents the highest level of delay in the supply chain (ââ¬Ëpullââ¬â¢ based system). As shown by Figure 3. 1, the decision about which strategy to use is essentially a tradeoff between different levels of customer service and inventory, production and distribution costs. 3. 2 Costs & Benefits of Postponement The question arises, when is the postponement strategy appropriate and when it is not? Where should a company position itself on the P/S matrix? In order to determine the most appropriate level of postponement that should be practiced, the benefits and the related costs must be weighed accurately. 3. 2. 1 More variety Having variety allows for a closer match between customer preferences and offered products leading to increased sales and (sometimes) increased prices. The build-to-order strategy pioneered by Dell shows how manufacturing a product according to customer specifications is one way to offer a large variety in a cost effective way. Dell offers enough options for their Dimension 4600C desktop to build over 100 million different computers using combinations of the components listed in Table 3. 1. Parts |Options | |Intel Pentium 4 |5 | |Operating Systems |5 | |Productivity Software |6 | |Memory |8 | |Hard Drive |4 | |Floppy/Storage Device |4 | |CD/DVD Drive |6 | |CD/DVD Software |4 | |Storage Devices and Media |2 | |Keyboards |3 | |Mouse |4 | |Monitor |9 | |Total Combinations |100million | Table 3. 1: Component List and Options for Dell 4600C Just like Burger King, Dell does not stock each of the 100 million varieties. Instead, they wait for customers to place an order before they build a machine. They have perfected this strategy so well that they are able to shape demand and produce popular combinations to forecast. Dell can offer discounts on combinations that are popular because of economies of scale and can carefully encourage customers to choose components that are in-stock using discounts. This strategy allows them to offer a quick turnaround and ensures that customers will not have to wait more than a week for a new product. Figure 3-2 shows a system dynamic loop measuring different factors that affect the number of product variety offerings. There are seven loops in the figure. The reinforcing loops (denoted by a positive arrow) show factors which increase the growth of product variety. The balancing loops (denoted by a negative arrow) show factors which inhibit the growth of variety. [pic] Figure 3-2: Systems dynamic loop showing product variety proliferation Loop one is a reinforcing loop that shows how variety grows because of the need to satisfy individual customersââ¬â¢ needs. The more customers see that their needs can be met, the greater their satisfaction in finding a unique product. This can force their expectations to be greater which narrows down markets even further. Loop two is a balancing loop that shows how a company reacts when it has captured most or all of the market, suppressing the need for innovation and excess product variety. Loop three is a reinforcing loop that shows what happens when there are multiple firms competing for market share. As a companyââ¬â¢s customer base increases it continues to innovate and offer more variety as a competitive advantage. Loop four is a reinforcing loop that shows the effect of technology on product variety. Loop five is a balancing loop that suggests that customers will become saturated with information and buy the product which offers them the best value given their search costs (time and information processing). As the number of choices keeps growing, negative aspects of having a multitude of options begin to appearâ⬠¦ the negatives escalate until we become overloadedâ⬠[38]. When too much variety exists, companies must tradeoff between offering variety and holding inventory. Loop six is a balancing loop which shows how high variety is traditionally associated with higher unit costs. When the unit cost increases, the customerââ¬â¢s willingness to pay for that variety goes down unless the extra cost adds value to the customer, which is the goal of customization. Similarly, in loop seven, as production lead time increases, customer service levels drop and customers are less willing to wait for variety without some compensation in terms of added value. Loops six and seven are opportunities where postponement can change the direction of the loops from balancing to reinforcing negating the traditional trade-off that exists between higher costs and variety with poorer levels of service. Postponement allows for more variety through standardization and holding intermediate product inventory and better customer service though relocating final assembly closer to the customer. 3. 2. 2 Inventory reduction Reduction in inventory under a fixed level of service is another benefit of postponement. When companies increase variety they increase the number of SKUs they must maintain which translates into higher inventory costs. Each SKU is subject to different forecasts and therefore require different levels of safety stock. Safety stock buffers against sudden increases in demand. Holding safety stock ensures better customer service but is also expensive because of inventory holding costs. In a study of the effect of product variety on production-inventory systems, Benjaafar and Kim [8] found that inventory levels increased linearly with variety. They also found that cost was most sensitive to demand variability, capacity constraints, and set-up costs (assuming a fixed cost to switch the production line between products). This highlights the risk associated with having too much variety for products, especially those with high demand variability. Companies can mitigate this risk by standardizing parts, holding more work in process (WIP) inventory, and postponing customization. 3. 2. 3 Better forecast accuracy Delaying the final customization of a product until more information is available allows forecasters to make better predictions of finished product demand. In order to delay customization, however, it is necessary to define what features or components make a product unique. Figure 3-3 shows how postponement reduces the variability of end product demand and saves on total inventory cost. [pic] Figure 3-3: Demand accuracy of postponed and non-postponed operations over time Using Figure 3-3, suppose that coffee mugs come in five different colors. The demand for each color is an independent random variable normally distributed with mean ? i and standard deviation ? i where i = 1â⬠¦ 5 for each of the different colors and ? i = ? ij and ? i = ? ij for all i and j. Total demand for mugs is N( i, v i2). The standard deviation for the demand of white mugs, v i2 , is less than the sum of the standard deviations of the individual demand, v i2, which explains why aggregate forecasts are less volatile. Additionally, forecasts generally improve over time therefore, ? i,T > ? i,t where T > t and ? i,t is the standard deviation in demand of mug i at time t. In this example, assume information about demand gained in the period up until time L/2 reduced the standard deviation of demand for each individual mug by half. Also, assume that at time L/2 the finishing time is equal to the customerââ¬â¢s willingness to wait. The producer is then forced to start painting the mugs at time L/2 to meet the customer demand on time. The variability of demand for mug color is more accurate at this point than it was at the start of the manufacturing process. It makes sense, then, to produce i or 5? uncolored mugs at time zero and then paint them at time L/2 assuming there are no additional switching costs incurred in this two-stage model. 3. 2. 4 Inventory cost reduction The amount of variety also affects inventory levels and hence, cost. The appropriate inventory level for a single SKU during a period of time consists of stocking the expected demand plus safety stock. Safety stock acts as a buffer to avoid stock-outs. Holding more safety stock improves customer service levels, but it comes at a cost. There are many formulas and practices for determining safety stock, however, this simple ââ¬Å"fixed safety factorâ⬠approach assumes demand is normally distributed and is commonly used to determine the appropriate level of safety stock, ssi , given a certain level of customer service, ssi = k ? i (3. 1) In equation 3. 1 k is the safety stock factor which is based on a given level of service desired by the producer and _i is the standard deviation of the errors of forecasts over a given period of time. The amount of inventory, hi , to have at the beginning of an order cycle for a single SKU is given by hi = ? i + ssi (3. 2) Assuming that all colors of mugs have the same mean, ? , and standard deviation, ? , of forecast errors, total inventory, H, is a function of the number of varieties, n, H = n(? + ss) (3. 3) Without postponement, inventory cost increases exponentially, not linearly, withn. However, as mentioned above, if orders are aggregated and produced in unfinished form, the total overall variation decreases. For example, assume each mug has the same mean forecast, ? i = 50 and standard deviation or forecast error, ? i = 2 for all i. The company wants to maintain a customer service level of 98 percent which equates to a safety factor of k = 2. 05. A comparison of the amount of inventory required to satisfy the variability in demand at the beginning of the production cycle with and without postponement as variety increases is shown in Figure 3-4. [pic] Figure 3-4: FGI under postponed and non-postponed operations Not only is the amount of inventory less under postponement, the cost to hold a single SKU is also lower because the product is unfinished. There is still the cost of stocking components for the finishing process (paint) but it is less expensive to keep the mug in an uncommitted state and hold the paint in component form. 3. 2. 5 Logistics cost reduction The above mentioned case of postponement illustrates delayed customization involving painting the exterior of a pre-produced standardized good, a coffee mug. Many examples of postponement exist where points of differentiation occur as early as the design phase and as late as product labeling and packaging. A modular product design offers more opportunities for outsourcing non-core processes, like packaging and distribution, to third parties. This can happen both onshore and offshore depending on the location and distance of the end customer. In either case, the manufacturer can save money by shipping products in bulk instead of in packaged form which usually adds extra weight and volume. 3. 2. 6 Improved customer service levels Customer service levels are defined in terms of lead time ââ¬â how long it takes an order to arrive, and item fill rate ââ¬â how often orders are filled from inventory on hand. Providing customers with orders quickly can be the result of improvements in manufacturing processes or by repositioning inventory closer to the customer. Customer willingness to wait is a key factor when assessing a product for postponement and determining the location of the postponement point within the supply chain. If customers are willing to wait a long time for a product then there is no benefit from expediting orders or sourcing components or processes closer to the customer even if they can be done cheaper overseas. On the other hand, if customers are only willing to wait, for example, one week, then the supply chain must be structured so that the finishing lead time and delivery time is less than or equal to one week. This breakpoint between initial and finishing lead times is called the decoupling point and separates production into two stages. The length of time for the first stage is not visible to the customer and therefore all options for achieving lower manufacturing costs can be exhausted. The second stage of the supply chain (from intermediate product to delivery) must be structured in a way that offers the customer the highest level of service without sacrificing cost. 3. 2. 7 Increased product development cost Another cost of postponement is the cost of design. If a product does not already have a modular design but meets all of the necessary market characteristics then it is worth researching the cost of redesigning the product for postponement. The benefit of a modular design is the flexibility it creates for other products within a family. However, there is a balance between too much modularity and its effect on product variety. The risk of too much modularity is a lack of differentiation between products. In addition, the cost to switch manufacturing operations between varieties is sometimes responsible for reducing economies of scale that could otherwise result. In terms of cost, product redesign can take engineers months translating into increased research and development costs. 3. 2. 8 Increased manufacturing cost There is a considerable amount of financial investment and commitment required to reconstruct the supply chain to support postponement. Manufacturing cost per unit may increase due to a restructuring of the production process into two or more stages. There should be dedicated areas for postponed activities in a warehouse and easy access to loading docks. If all manufacturing is not done in-house (which is more likely than not) implementation may require additional facilities to support final assembly and distribution. This also requires more labor at a higher skill level to complete kitting, final assembly, and packaging as opposed to the lower skilled labor required for loading, storing, and sorting. Chapter 4 Case Studies The following case studies give detailed information about several companies that have adopted postponement in some capacity. It is worth understanding the motivations and risks that they incurred in order to understand how companies can determine whether their product is a candidate for postponement. Each case provides background on the company and product that is postponed, a description of the supply chain before and after postponement was adopted, the decoupling point between intermediate product and finished good, costs and benefits, and discusses how the supply chain is structured to take advantage of offshore manufacturing and local final assembly. 4. 1 Automobile Manufacturing: GM The auto industry is a prime candidate for postponement for many reasons. First, a car is defined as a modular system of components. This creates opportunity for commonality by producing a platform and adding modular subassemblies customized according to the make and model and ultimately the end user of the vehicle. Second, individually customized vehicles have high forecast variability. As this case points out there are far too many varieties to accurately forecast each combination and there is typically disagreement on the forecast within the different divisions of a company. Third, cars depreciate as soon as they are driven off the lot. New models come out each year which new features, technologies and capabilities. Lastly, high inventory holding cost. It is much riskier to hold a finished vehicle on the showroom floor than to have a partly finished good waiting for final customization because of the high forecast variability for end products and high product obsolescence cost. General Motors (GM) offers a unique look into customization during manufacturing and after the point of sale. By 2004, GM produced 68 different models in North America. There were over 200 facilities constituting 52 percent of their revenues. There were over 600 million combinations when all the different component variations and customer specific preferences (color, interior options) were considered. Forecasting was extremely difficult, considering these many combinations. Different divisions within GM used different methods of forecasting which further complicated the problem and led to excess inventory on the field. Searching for a way to create variety and mass customize beyond the idea of platforms, GM looked at software configuration, entertainment, and aesthetic features as a different way to use postponement. From a software standpoint, each of the systems within a vehicle can also be considered a unique central processing unit (CPU) made up of several electronic control units (ECUs). These include safety systems, engine, and transmission controls. In the 1990s there were only one or two ECUs in a vehicle. Now there are as many as 30-35 per vehicle because software is becoming increasingly essential in automobiles for voice recognition, global positioning systems, and entertainment. Before postponement, GM experienced the effects of product variety proliferation and high inventory costs of stocking ECUs for individual models. The ECUs came to GM in finished form with all of the software pre-loaded. Suppliers charged GM a premium for custom software installation which not only raised the price but also created problems with repair and maintenance. GM decided that they would assume the responsibility for software configuration and postpone the installation until the latest possible point in the assembly process. In order to accomplish this, GM had to redesign both the assembly process and the ECU hardware. In the mid-1990s GM achieved the capability to install custom software for individual orders towards the end of the vehicle assembly process. The ECU now comes from suppliers to GM in a generic form. The hardware is a common platform which can receive customized software in just 81 seconds. GM dealers also had to acquire the capability for flash programming for individual cars at the point of sale as well as after-market upgrades. After realizing that software could be postponed, GM looked at other systems that could be delayed until purchase. They recognized the emergence of the accessory market for vehicles as another way to differentiate and increase revenues. Entertainment systems have become far more sophisticated over the years and offer key differential options on a vehicle. Because of the plug-and-play capability, entertainment systems can be uploaded into the vehicle at the dealer. Another key differentiator is the wheel set. Dealers are very involved in putting specialized wheels on a car to make it more desirable. Through the use of the internet, GM introduced an on-line purchasing website. Customers can log on to GMbuypower. com and point and click their way to the car of their dreams. GM offers a 99 percent guarantee that they will deliver the vehicle within one day of the projected delivery day to a dealer close to the customer. By 2004, about 18 percent of the cars in assembly at GM were custom made and 82 percent were made-to-stock for dealers and showrooms. The goal is to move to 60-80 percent custom orders but the shift is happening in different markets at different rates. GM is experiencing the benefits of postponement through delayed software configuration and customization. In a study to estimate the benefits of postponement, GM, along with MIT and Stanford University, developed a cost model which projected inventory cost savings to be 10-15 percent. Other benefits included maintenance cost savings due to the highly communized ECU hardware and having GM software engineers solve repair issues instead of sending parts back to suppliers. GMââ¬â¢s main goal, however, is to create a more flexible supply chain that can handle higher throughput and is more responsive to immediate demand. 4. 2 Aircraft Manufacturing: Embraer The commercial aircraft production at Embraer provides an example of production and assembly postponement in the airline industry. The motivation for postponement was to focus on ââ¬Å"optimizing cash flowâ⬠by creating a flexible supply chain that can provide the right airplane to the right airline company. In other words, the goal is to give customers the ability to change their decision regarding customizable features, or to cancel an order completely, by designing the aircraft to accept these changes as late in production as possible. In response to the changing dynamics within the aircraft industry Embraer differentiates its new family of regional jets based on the number of seats. The new family of regional jets, the Embraer 170, 175, 190 and 195, focuses on a high degree of parts commonality as all four jets have exactly the same cockpit and fly-by-wire systems. Embraer decided to implement postponement in order to make its supply chain more flexible and able to respond quickly to changes in demand. This was evident when a customer, US Air, had to cancel an order for six ERJ 170 aircraft because of financial constraints in October 2004. With the majority of the production complete it was too costly to go back and change any of the customized features and reconfigure it for another airline. Embraer developed a strategy for postponing as much of the high value features, like engine type, software, radar devices, and interior specifications as possible. Not only did it save on costs, the flexibility to change order specifications became an attractive alternative to backing out of an order or having to pay for costly reconfigurations. The current supply chain at Embraer is structured to allow for two postponement points throughout the production cycle as illustrated in Figure 4. 1. The first point occurs roughly one year before delivery to the customer where the platform is differentiated based on product family (170 versus 190 family of aircraft). Six to eight months later it will assume the configurations, engine, software and hardware which distinguish it as a 170 versus a 175 or 190 versus a 195 aircraft. After this point the customer specific features such as seating arrangements, galley configurations, and tail art are added. [pic] Figure 4. 1: Lead time break down of value added components and features Embraer still builds-to-order because of the high cost to hold a finished airplane in inventory. The white tail concept (analogous to a ââ¬Å"vanilla boxâ⬠) allows the production processes to begin and run in parallel with some of the steps that usually take a long time to complete such as certification for safety, avionics, and entertainment systems. Total lead time for production is usually 24-36 months because of the long lead time for suppliers. Production begins 12 months before delivery and the order is considered 90 percent ââ¬Å"frozenâ⬠or unchanging. However, some customers change their mind within the final month of production. Embraer is committed to developing the idea of postponement further within the company. Any flexibility that can be gained through delaying the customization makes Embraer jets more attractive to a customer facing the uncertainties of the aircraft industry. Engine, avionics, interior and galley layout are some of the hardest subassemblies to change and also have the highest value. The white tail concept allows Embraer to have flexible production in its new family of 170/190 aircraft. They do hold some inventory of semi-finished aircraft that await orders from larger companies in the corporate jet market because the orders are more predictable. Embraer represents a company that is practicing postponement and is not seeing huge savings in inventory. Instead they redesigned their process to accommodate the addition of components based on value to the customer and degree of customization. Better service levels and customer satisfaction give Embraer a competitive advantage in a very competitive market. 4. 3 Clinical Equipments: Dade Behring Dade Behring (DB) is an industry leader in clinical diagnostic equipment and reagents. Their customers include over 25,000 hospitals and reference laboratories which require instruments that analyze human fluids such as blood and urine. They have global operations in more than 34 countries and currently deliver products in six main areas: Chemistry, Immunochemistry, Hemostasis, Plasma Protein, Microbiology, and Infectious Disease Diagnostics. DB diagnostic instruments are high value with a retail price ranging from $20,000 to over $200,000. Demand forecasting is a challenge due to long buying cycles ranging anywhere from six months to two years. Forecasts are generally compiled from sales representativesââ¬â¢ predictions. Because of the high cost of the products, the decision making process and financial constraints of the customers, it is somewhat difficult to know when products will be ordered. Additionally, instruments were designed/ configured to local country power requirements which exasperated the forecasting impact. As a result, DB was plagued with less than optimum service levels for some instruments and higher than planned inventories for others. All of these conditions were catalysts for a postponement strategy, which became even more important as a result of an industry-wide European directive. The first postponement strategy involved designing flexible power capability into the Dimension Chemistry/Immunochemistry analyzers that Dade Behring designed and produced. Originally Dimension was offered in either a 110 V or 220 V power versions. To optimally manage inventories of these instruments, DB collaborated with an external supplier to replace the power supply module with a universal power supply. During the redesign phase engineers were able to develop the universal module at a lower cost because of advanced technology which was previously unavailable. The cost to produce the universal module was actually less expensive than supplying two different versions. Then, a second postponement strategy was put into place due to the European IVDD initiative. In 1998, the In Vitro Diagnostics Directive (IVDD) was ublished as the third of three European directives which required medical and diagnostic equipment to come packaged with local language manuals and labeling. The regulation gave 17 countries the right to specify the national language that would come available with each instrument for which they contracted. In total 12 different language manual s were eventually required. The instrument manuals are approximately 350 pages in length and therefore it did not make sense to create a single manual with all 12 languages included nor package 12 different manuals with each instrument. DB initiated the switch to language specific packaging in the industry through the postponement of packaging materials at distribution centers and flexible language capability within the operating software. This is a straight forward process accomplished by marrying a language specific accessory box to the instrument during the shipment process. Shortly after achieving successful packaging operations, DB initiated another postponement strategy in their Chemistry product line. This next strategy was to redesign the product so that it could be configured-to-order at the end of the assembly process. There are currently four variations of the Dimension Chemistry/Immunochemistry Analyzer Series. Dimension is offered as RxL Max Basic and RxL Max HM (heterogeneous model), or as an Xpand Plus Basic and Xpand Plus HM. Through a carefully designed manufacturing process, Dade Behring is able to manufacture a specific model as soon as that specific model is shipped to fill a customer order. This strategy involved the redesign of the manufacturing process so that the analyzer could be configured-to-order at the end of the assembly process. This meant that all of the commonalities between the two different variations of each model would be combined into an intermediate product that would be produced to a forecast, stored as intermediate inventory, and configured-to-order once an order was received. The redesign phase took a team of engineers six months to make changes and train workers on the assembly line. The supply chain as shown in Figure 4. 2 became vastly more efficient and service levels increased dramatically. [pic] Figure 4. 2: Dade Behring supply chain Customer service levels improved and inventory was significantly reduced by eliminating the need to store high value finished goods. Inventory across the supply chain was reduced through a 50 percent reduction in ââ¬Å"bufferâ⬠or safety stock. Service levels went from oscillating between 70-100 percent to greater than 98 percent. Once DB was able to improve service time to customers they started looking at their distribution centers and found opportunities to improve distribution strategies, given the improved flow of instruments through the manufacturing process. Because the opportunity cost of a lost sale in this industry is very high, distribution centers would store finished goods as a way to mitigate the risk of instrument shipment delays. However, when service levels improved, DB found that they could eliminate 50 percent of their global buffer inventory by eliminating the stocking of instruments in distribution centers in Asia and Canada, and reducing inventory levels in Latin America. Their primary instrument warehouses in the US and Europe service their global instrument distribution needs. The make-to-order and inventory management strategy provides DB with a decisive advantage in the industry. This is a classic example of the benefits of the successful implementation of postponement. Because of this success, DB was able to continue developing postponement in other lines of instruments. Today, more than 85 percent of instrument production at DB involves some form of postponement compared to less than five percent five years ago. By redesigning the Dimension instruments to be easily adaptable for configuration, DB realized that the product could also be easily de-configured back to the intermediate stage to support the secondary market for instruments. 4. 4 Sports Goods Manufacturing: Reebok As a licensed supplier for the NBA and NHL and principle supplier for the NFL, Reebok knows the difficulties that come with satisfying the demand of a very ââ¬Å"fair weatherâ⬠crowd. When teams do well more team apparel is demanded. The demand for a player specific jersey is inherently more volatile than for a given team. Meeting customer requirements within a short period of time is a major challenge in the sporting goods industry. Sales of t-shirts and jerseys are not too predictable because Reebok does not know which teams will be ââ¬Å"hotâ⬠at the beginning of the season. Demand for jerseys averages 30,000 per week or 1. 5 million each year. The different choices of team name, player name, color scheme, and size makes it extremely difficult to predict demand of an individual item during the pre-season. The idea of postponement in this industry is not new. Images of silk-screen companies working overtime minutes after an NCAA basketball championship game, illustrates the idea of postponement. These manufacturers know that it is better to wait until there is certainty about the outcome of a game before producing apparel with the losing teamââ¬â¢s name on it. As a result they keep white or blank shirts on hand ready for printing. At this point in the supply chain it would not make sense to put in an order for finished shirts from scratch to an overseas manufacturer (even if it costs less to make the shirt). The long lead time would mean missing the increase in sales generated within two weeks after a big win. This can be anything from an important mid-season upset, a new player entering the roster, players becoming ââ¬Å"hotâ⬠, or the end of season championships. Reebok recognized this as an opportunity to restructure the supply chain to cater to both stable items ââ¬â finished apparel that is produced to a forecast much earlier in the season, and customized apparel. The difference in the lead time for both of these items is significant. Retailers expect lead time to be 3-12 weeks for the stable items and as little as one week for the ââ¬Å"hotâ⬠items. Reebok outsource the cutting and sewing of fabric to contract manufacturers in Central America. Some of the jerseys sent to Reebok are finished meaning that there is a customized team and player name already on the garment. Other jerseys, called ââ¬Å"team finishedâ⬠jerseys are sent with everything but a playerââ¬â¢s name. These go straight to a distribution center that Reebok owns and operates in Indianapolis. The blank or team finished jerseys help satisfy two different types of demand. The first is for the hot players or players who sign with a team late in the pre-season and the second is for the players who have a small, but somewhat predictable demand. [pic] Figure 4. 3: Reebok Supply Chain According to Figure 5. the blank jerseys arrive in the US and are ready for screen printing and embroidering. The decision to have a separate facility in the US is a result of the end customerââ¬â¢s unwillingness to wait. Fans expect to find the jersey they are looking for in a store. There is a chance they will be less likely to want one if they have to wait weeks to get it ââ¬â especially when an NFL team only plays 16 games per season. At a price of $25 for a long-sleeve t-shirt or $250 for an authentic jersey, the cost of lost sales is greater than the cost to ship, unpack, finish and reship a jersey from a local finishing center. Reebok is a classic example of two-stage production with postponement. They are able to take advantage of lower labor costs for the production of blank jerseys and optimize service levels by souring the final assembly in the US. This also creates local jobs in the areas of textile and silk-screen printing. 4. 5 Xilinx Xilinx is a semi-conductor manufacturer with headquarters in San Jose, CA. The semi-conductor industry is very volatile due to the wide variety of products and short product life cycle. Semi-conductors manufacturers are supplied to OEMs in the telecom, small electronics, and aerospace industries. However, they have a supply chain of their own which requires assembling and configuring wafers of silicon into programmable dies which later become integrated circuits. Their position in this multi-echelon supply chain makes forecasting for specific end product demand costly, impractical, and very inaccurate. In addition, semiconductor manufacturing is quickly becoming a commoditized process. Comparative intellectual and technological benefits that leaders in this industry were accustomed to are now becoming less of a competitive advantage. The focus has shifted from intellectual advantage to supply chain efficiency as a means of differentiation. The life cycle for an integrated circuit is anywhere from six months to two years. During that time new technology will make existing products obsolete. Having long manufacturing lead times cripples a companyââ¬â¢s ability to quickly respond to these changes as well as changes in customer specific orders. Having a generic product and creating a postponement point separating a die with generic qualities and one with a specific logic configuration allow them to respond quickly and offer flexibility to their customers. Xilinx began with a combination of both process and product postponement. Product postponement was implemented by redesigning the dies to a certain range of parameters for the different characteristics. For example, there are four major sources of variety in an integrated circuit; speed, number of logic gates, package types, and voltage. Customers can specify generic capabilities and can customize the chip to their specific specifications after the fabrication stage. The amount of variety makes postponement very beneficial. Xilinx can manufacture 200 different dies that can proliferate into over 4,000 different end product combinations. That makes the ratio of generic dies to end products roughly 1:20. The manufacturing process is broken up into two stages. Suppose a certain generic die, A, can be configured to take on 20 different configurations, {A1, A2, A3, â⬠¦ , A20). When a customer requests the specifications, they only need to specify the generic die. Once it is pulled from ââ¬Å"Aâ⬠inventory, it is customized to a certain degree depending on customer order specifications. This specification can take place at Xilinx for high volume orders or it can be delayed even further so that the point of customization occurs at the customer. Approximately 20 percent fall into the high volume category and the remaining 80 percent are left for customization at the customer. The final customization is a matter of programming the software within the chip. By eliminating this process from the front end (manufacturing) process, Xilinx cut manufacturing lead time from three months to three weeks. Manufacturing usually takes place in Taiwan or Japan and then product sits in inventory at Xilinx awaiting testing. Testing facilities are located in Korea, Taiwan, and Japan. While postponement has reduced inventory and helped Xilinx meet customer requirements with more accuracy and on-time deliveries, it is just the beginning. Postponement within the semiconductor industry will extend far beyond customer configuration capabilities. Chapter 5 Conclusion The case studies presented in this paper come from a wide variety of industries. Each company was successful in implementing postponement for similar reasons, but have seen a wide variety of results. The most common strategic motivation for starting postponement were to improve service level and to reduce inventory holding cost as a result of an increase in product variety. One of the key factors in successful implementation is product modularity. If a product is not inherently modular, a successful postponement strategy requires a redesign of the product or a rethinking of product definition. In the cases of small consumables, the end product is not a razor blade or a disk, but rather a finished configured package destined for a particular retail outlet. The relationship between forecast variability and the decision between a make-to- stock or a build-to-order strategy is also a common factor. Products with stable demand stand to gain little from a postponement strategy because there is little benefit for delaying production when sales are committed. On the other hand, products with high variability gain from postponement because there is no commitment to final configuration until the order is placed. A company should determine the location of variability when deciding to implement postponement. Variability can be caused by product variety, unreliability of customer orders, seasonality, trends, promotional activities, or it can be a result of the supply chain itself. Varia
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.