The Economic Cost and Pain of the Bullwhip Effect - Essay Example

Bullwhip Effect in Retail Supply Chains This paper is a discussion of the bullwhip effect phenomenon as a real problem in retail supply chains, and to what extent improved customer demand forecasting can solve it. The first part is a brief discussion of supply chains and their management. The second part describes the bullwhip effect in supply chains and the breakthrough studies on the topic. The third part shows briefly the economic cost and pain of the bullwhip effect. Then, the causes of the bullwhip effect are enumerated and summarised. The paper concludes with a critical discussion of the solutions and best practices on minimising or eliminating bullwhip. Supply Chain and its Management The supply chain is an extended enterprise with different upstream and downstream companies and various service providers that need to operate closely together. These companies are directly linked by the flow of products, services, money, and information from various sources to the customer (Simchi-Levi et al., 2004). The supply chain puts together different products from primary supplies for a particular industry or company all the way through to final consumption, which can include the return of part of the product (for example, bottles and containers) for recycling. Supply chain management (SCM) then includes all the methods, systems, people, and resources that help improve processes and organisations throughout the supply chain. This also includes design of the product or service, purchasing, inventory management, planning and scheduling, logistics, distribution, and customer satisfaction. SCM involves close working together of all companies in the chain. Simchi Levi et al. (2004) classify SCM into two categories: configuration, which is related to basic infrastructure (hardware, software, transport, etc.); and coordination, related to the way the supply chain operates. Configuration issues include decisions on choosing suppliers, outsourcing activities, and policies for purchasing, decisions on production, site location, capacity; distribution channels, retail locations, and transportation costs and issues. Coordination issues include decisions on material flows throughout the chain, how information is exchanged, and payment systems. This shows how complex supply chain management is because it involves many functions and geographic areas. Design and execution are therefore difficult and need to be managed for the supply chain to move with efficiency. An example is shown in Figure 1 (Gereffi, 2002) for retail apparel which links cotton and synthetic fibre manufacturers, textile mills, apparel manufacturers, and retail outlets from all the five continents. If one link in this chain breaks, e.g., the container ship with the raw material supply of African cotton gets lost at sea, the whole supply chain can break down and thousands of clients of Marks & Spencer will have to party using last season's fashion. Supply Chain, Meet Bullwhip What is known as the "bull-whip effect" can be described as follows: the farther away from the customer a supplier is along the supply chain, the higher would be the difference between what is really needed from what is ordered. The term was coined by Lee et al. (1997) based on observations and descriptions made by supply chain professionals at Procter & Gamble (P&G). They noticed that whilst the number of babies and the demand for nappies were stable, the orders coming from retailers and wholesalers for Pampers deliveries, P&G's best-selling nappies brand, fluctuated dramatically. And as they went further down the supply chain, starting with the orders made by P&G from the suppliers of the components that went into a nappy - plastic, cotton, and so on - they noticed even wilder fluctuations. On a graph, they noticed that these fluctuations were similar to the way the amplitude of a whiplash increases down the length of the whip once it is cracked. Thus was born the phenomenon known as the Bullwhip Effect. Figure 2 shows this phenomenon in action, using data from Lee et al. (1997) as a basis for the explanation. The basic premise of the demand driven collaborative supply chain model is that to achieve the highest possible in-stock and simultaneously minimise waste, all parties within the supply chain must have timely access to actual sales demand data and all parties must have a means by which they can work together to coordinate promotions and business planning. A forecast by definition is an estimate made in advance of an event occurring and is therefore an educated guess. Unfortunately, even sophisticated forecasting software can have an error rate of 50% on promoted items. Most troubling of all, forecast accuracy decreases moving backward into the supply chain. The reason forecast accuracy decreases moving backward in the supply chain can be illustrated by plotting sales over time for supply chain participants as depicted in Figure 2. Notice that actual sales demand recorded at the retail point-of-sale (POS) has moderate variation. This is because the retailer builds their forecast model using actual demand as tracked through their POS systems. But notice how much more chaotic and unpredictable the demand curve becomes as you move away from the actual point-of-sale. Demand as viewed by the supplier, wholesaler, and manufacturer is based on estimated sales which combined with latency and manually adjusted "safety stock" causes increasingly inaccurate and chaotic forecasts. Research indicates a fluctuation in actual customer demand of +/-5% will be interpreted by supply chain participants as a change in demand of up to +/-40%. As Figure 2 shows, although actual demand has only changed +/-5%, the reaction of supply chain participants is dramatically exaggerated and is known as the bullwhip effect. Much like cracking a whip, the user only needs a small motion in their wrist (point of sale) to cause a huge motion in the end of the whip (manufacturer). This is nothing really new. What Lee et al (1997) described technically as the "amplification of demand variability from a downstream site to an upstream site" was first observed by Forrester (1958) who used the term industrial dynamics to show how product demand moves from customers to manufacturers and to suppliers. Zymelman (1965) proposed how bullwhip could be reduced in the cotton industry through a control law that he established through analogue simulation. Burbidge (1981), based on a study of U.K. companies, was the first to propose a solution as to how production problems could be avoided if long set-up times can be reduced by aiming at small batch production, instead of so-called Economic Batch Quantity (EBQ) production where minimisation of cost through economies of scale was the norm. Sterman (1989) attributed the phenomenon to systematic errors of decision makers and managers as they looked at the dynamic environments of the business that is characterised by multiple feedback loops, time delays, and nonlinearity. In other words, it was seen as a natural phenomenon that was beyond their control, and the best that could be done was to design management systems to cope with it. The breakthrough research was carried out by Lee and his colleagues at Stanford University who, in the early 1990s, observed that it existed not only at P&G but in almost every company from different industries like Tesco, British Aerospace, and Rover. Data showed that the phenomenon, which others called the "whiplash", "whipsaw", or "acceleration" effect, manifested itself throughout the supply chain, which is the continuous link of suppliers from raw materials producers all the way to the customers. It was in 1997 that Lee and his colleagues used the term "bullwhip effect" to describe the phenomenon, and it was the name that has stuck ever since. Several industry studies over the years have been recently summarised by Geary et al. (2006). Aside from revealing that the bullwhip effect has been observed by practitioners early in the 20th century, they also identified ten published causes of the phenomenon and suggested solutions, both tried and tested and a number that have not been tried, to cope with the problem. Before proceeding with the causes and solutions of the bullwhip, it would be interesting to consider briefly the effects in terms of business pain and costs that show that the problem exists and that it needs to be solved. The Pain of Getting Bullwhipped It can be assumed that getting hit by a bullwhip is painful, so it is best avoided unless by a bull that needs to be brought back to its corral. For years, supply chains have been hit by their own bullwhip, and the results are rather painful. How painful is it A recent (Manhattan, 2006) estimate by the U.S. Department of Commerce put the pain of the bullwhip effect at anywhere from $1.1 of inventory tied up along the supply chain at any one time. This total, which could rise to $3 trillion if inventories in Europe and Asia are included, can support some $3.2 trillion in annual retail sales in the U.S., where the inventory is spread out across the value chain: $400 billion at retail outlets, $290 billion at wholesalers or distributors, and the remaining $450 billion with manufacturers. Given these figures, it would be easy to assume that stock-outs at retail outlets should be low, but this is not the case. Manhattan (2006, p. 2) reports that 8.2% of shoppers on average fail to find their product in stock, that these stock-out events represent 6.5% of all retail sales, which translates to net lost sales of 3.1%. In the retail industry where margins are being driven ever lower by global competition, this is increasingly getting to be unacceptable, not to mention the effects of a loss in customer goodwill. An Economist survey (Markillie, 2006) pointed out how supply chains are becoming a source of strategic advantage for large companies that learn how to use it to increase revenue and profits without necessarily lowering costs. By re-engineering their supply chains, companies like Sainsbury's, Boots plc, and Marks & Spencer have become more responsive to customer demand by smoothing out bumps in the supply chain and reducing bullwhip inefficiencies. Why the bullwhip can cost so much is easy to understand. Whilst customer demand may be stable, there are several factors that work along the supply chain that lead to such costly consequences as over-capacity and stock-outs on the upswing, and stockholding and obsolescence costs on the downswing. These factors are causes of the bullwhip effect, so any reasonable attempt at finding solutions should begin with a better understanding of the causes to the problem. Ten Causes of the Bullwhip Effect Geary et al. (2006, p. 9-10) enumerate ten known causes of the bullwhip effect by identifying ten principles for the design of supply chain systems. When these principles are not followed, the result is a bullwhip in the supply chain. 1. Poor control systems result in guesswork when making sales forecasts. Thus, agents across the chain compensate by over-estimating what is really needed. 2. Time wastage in achieving such tasks as design, production, sourcing, and delivery leads to the accumulation of non-value added time and toleration of delayed deliveries. This also shows a lack of process capability or incapacity to deliver and a propensity to compensate through re-works. 3. Lack of information transparency characterised by supply chain players not being able to access up-to-the-minute data free of "noise" and "bias" that results in information delays and double-guessing of other players in the system results in over-ordering to compensate for potential non-delivery of certain components. 4. Non-elimination of echelons appropriate to the goals of the supply chain, which goes against the target of having only the optimum level of inventories (zero in some cases) and having these minimum stocks in the right place at the right time. 5. Poor synchronisation of orders and deliveries that should be visible at discrete points in time across the supply chain. Multiple customers working on EBQ (Economic Batch Quantity) re-order principles produce an emphatic bullwhip that could be eliminated by continuous ordering synchronised throughout the chain. 6. Multiplier or knock-on effect, usually between product manufacturers and their capital equipment suppliers, where orders are multiplied over time. Anderson et al. (2000) demonstrated this in the machine-tool industry. 7. Safety factors and trend detection capabilities tend to inflate demand forecasts, which are already a problem because they are rarely right. A complication arises when product substitution takes place where what is actually available is sold in place of stock-out items, thus creating a demand where there was actually none. 8. Time-phased aggregation of orders, a consequence of the EBQ practice, result in "lumpy" deliveries that come back around the ordering loop as "lumpy" orders, which causes bullwhip. 9. Price fluctuations from sales and discounts caused by marketing programmes that are deliberately designed to empty over-full pipelines. Fisher (1997) demonstrated that this causes a backlash of over-ordering to take advantage of discounts, and when the retailer accumulates enough stock, orders drop to zero, creating a boom-and-bust scenario. 10. Gaming or hedging against unpredictable supply in an actual or perceived shortage situation, also known as hoarding, resulting in double-guessing amongst suppliers and customers trying to protect their businesses against what other players are doing. How Do You Solve a Problem like the Bullwhip Is better consumer demand forecasting a solution Yes, but it is not the only solution, because there are many other root causes of the bullwhip that have their own sets of solutions that are not limited to improving demand forecasts. The first real step in finding a solution is admitting that the problem, and a solution, exists. Otherwise, nothing happens in the supply chain as purchasing or inventory managers and their cohorts in production continue to bleed the company dry by not managing their inventories well. The financial impact of the bullwhip has been discussed above, so any actions that would mitigate the bullwhip's impact would find its way to the bottom line. Surprisingly, acknowledgement of the problem is often not the case, as many managers are unwilling to admit that there is a bullwhip in their supply chain. McCullen and Towill (2002) in their study were able to identify ten common reactions: ignorance, arrogance, negligence, indifference, despondence, transference, intolerance, avoidance, decadence, and acceptance. Any of these is enough to paralyse anti-bullwhip efforts. In effect, whilst some managers may see the problem as without a solution, others assume that the solution would not work or does not apply, or they submit themselves blindly to bullwhip as a fact of business life. Loughman et al. (2000) argued strongly that people factors need to be addressed because otherwise, any planned improvements would be useless. No matter what sophisticated Customer Relationship Management (CRM) tools are used, implementing the plans would be frustrating if the key players in the supply chain such as the people in sales, purchasing, even accounting and finance do not foster the desire to change. People, they pointed out, do not always use "rational" methods in decision-making. They have personal agendas; do not always do what they are told even if they know it is the right thing; sabotage processes they fear or dislike; or they just dislike complicating the work they are used to. Once the problem and the possibility of a solution exist, the next logical step is to measure the amplitude of the bullwhip and determine what savings could be made. The simplest solution was that recommended by Watson (1994) and known in the industry as UDSO for Understand-Document-Simplify-Optimise. This means understanding who the players in the supply chain are and mapping their performance characteristics, documenting data along each step of the chain, looking out for inefficiencies marked by stock-outs or over-supply of products, and then thinking how these inefficiencies could be avoided. On paper, this is a simple process but in reality, this can be quite complex. The complexity is in the inputs required, which would need a high degree of precision, such as customer orders and schedules, raw materials supply scheduling, production processes and times, etc. Once the data are in, analysis takes place, and this is where needed insights may be more complicated, such as looking for more efficient ways to ship products, whether packaging can be improved, and so on. As can be immediately seen, it is not enough to know what the customers exactly want and when they want it. If the other parts of the supply chain fail to deliver on schedule, a bullwhip effect takes place. A more recent study (Bjork, Hejazi, and Carlsson, 2007) showed that despite the many publications that explain the causes of the bullwhip effect, the countermeasures proposed are often based on mathematical models in which different assumptions are made to simplify the complexity of the real world. Having transparent demand data information at a downstream site available to the upstream site is another solution, as this would counteract the bullwhip due to several of the ten known causes that lead to poor demand forecasts. One example is the so-called Vendor Managed Inventory, where a supplier takes charge of determining delivery schedules and quantities based on Point-of-Sale information transferred across the supply chain from retailers to distributors and thence to manufacturers and raw materials suppliers through an Electronic Data Interchange (EDI). This model is used by several companies to reduce inventories by as much as 25% (Lee et al., 1997, p. 99). If sharing information and forecasting collaboration is as valuable as suggested, why is it not more common A study by Smros (2004) showed that despite successful pilot implementation of the Collaborative Planning, Forecasting and Replenishment (CPFR) initiative between grocery retailers and suppliers, the adoption rate of collaborative practices has been slower than expected. She discovered empirical evidence showing that most collaboration models and the CPFR initiative seem to build on invalid assumptions concerning retailers' forecasting needs, resources, and processes, as well as suppliers' capabilities to benefit from the more accurate demand and forecast information made available through collaboration. Four critical factors were identified as reasons for the slow uptake: retailers' limited forecasting resources and lack of processes needed to support intensive CPFR-style collaboration, retailers' and suppliers' divergent forecasting and collaboration needs, supplier difficulties in benefiting from the collaboration, and the unwillingness to invest in information technology (IT). A more recent study (Overby et al., 2005) by Forrester Research showed considerable improvements in manufacturer-retailer collaboration, thanks to the emergence of more sophisticated technologies and standards. Collaboration and information sharing, however, continue to be difficult due to mutual mistrust amongst supply chain partners. Such mistrust that are cultural and financial in nature can be addressed by aligning incentives, such as greater reliability in demand planning and forecasting, more coordination on price and promotion efforts, and measuring and rewarding collaboration using financial metrics such as increased revenue and inventory turns. Figures 3 and 4 highlight their findings. Sainsbury's demonstrated that supply chain improvement and elimination of the bullwhip can be an important component of business transformation (Sun, 2003). According to its supply chain development director, costs were reduced after bullwhip elimination through delivery of fresher products after adopting a three-point strategy of network renewal to gain employee acceptance of the revamped supply chain, an integrated system of responses, and a focus on the company's people and culture. The company invested in new facilities and programmes to benefit from accuracy, simplified paperwork, product-friendly deliveries, and a closer relationship with depot managers. It also implemented faster shelf replenishment, faster vehicle turnaround, delivery from a maximum of three depots, timely delivery and notification, improvements in inventory accuracy, and an easy return of crates and roll cages. Integration covered planning and operations as well as suppliers, product development functionality, and in-bound shipments to warehouse operations. The major challenges for Sainsbury were capability, compliance, and measurement, which included delivery standards, the use of readable codes and pallet labels, delivery times, pallet standards, transmission of electronic delivery notes over the Web, and data integrity. Its most important challenge, however, was to teach old dogs and new dogs new tricks. This involved driving behavioural change through measurement, development, and partnership with the help of managerial leadership, which it considers critical, and simple, clear, and frequent communications. The result was Sainsbury's ability to migrate from a custom-built, paper-based, and centralized legacy environment to a new paperless system that operates in real time through a Web browser that has been integrated into key points in the supply chain. The company has seen an improvement in underlying profit and has achieved a best-in-class cost per case with an in-store availability of 97 percent. Conclusion: Is Better Demand Forecasting the Solution Research has shown that, unfortunately, no forecasting tool is capable of predicting all types of demand fluctuations, as Geary et al. (2006) pointed out in their summary of best practices shown in Figure 5. Despite use of progressively more sophisticated forecasting tools, unless the people problem is solved, and every single player in the supply chain gets learns to use and benefit from demand forecasting tools, the bullwhip effect can never be eradicated. It may even worsen if these tools are used improperly. Demand forecasting is useful if information flow and coordination of orders across the whole supply chain takes place: from top management to the lowest level across the whole supply chain, everyone committed to follow the strategic imperative of removing identified sources of inefficiency. People commitment and proper use of tools would result in strategic partnerships and improved data integrity, crucial factors to develop seamless supply chains. Otherwise, the bullwhip effect will remain to be a problem that would continue leading to lower customer service levels and high inventory costs. Bibliography Anderson Jr., E.G., Fine, C.H. and Parker, G.G. (2000). Upstream volatility in the supply chain: the machine tool industry as a case study. POMS Series in Technology and Operations Management, 9, p. 239-261. Bjork, K.M., Hejazi, A. and Carlsson, C. (2007). The bullwhip effect: Complexity in theory and in a real-life application. International Journal of Integrated Supply Management, 3 (1), p. 82-102. Burbidge, J.L. (1981). Britain's counter productive plants. Management Today, November, p. 86-90. Childerhouse, P. (2002). Enabling seamless market-orientated supply chains. Ph.D. thesis, LSDG, Cardiff University. Fisher, M. (1997). What is the right supply chain for your product Harvard Business Review March/April, p. 105-116. Forrester, J. (1958). Industrial dynamics: A major breakthrough for decision-makers. Harvard Business Review, 36 (4), p. 37-66. Geary, S., Disney, S.M. and Towill, D.R. (2006). On bullwhip in supply chains: Historical review, present practice and expected future impact. International Journal of Production Economics, 101, p. 2-18. Gereffi, G. (2002). Outsourcing and changing patterns of international competition in the apparel commodity chain. Paper presented at the conference on: Responding to Globalisation: Societies, Groups, and Individuals, April 4-7, 2002. Boulder, Colorado Lee, H.L., Padmanabhan, V. and Whang, S. (1997). The bullwhip effect in supply chains. Sloan Management Review, Spring, p. 93-102. Loughman, T.P., Fleck, R.A. and Snipes, R. (2000). A cross-disciplinary model for improved information systems analysis. Industrial Management and Data Systems, 100 (8), p. 359-369. Manhattan Associates (2006). Demand chain optimization: Pitfalls and key principles. Business Value Research Series. New Jersey: Manhattan Associates. Markillie, P. (2006). "The physical internet: A survey of logistics". London: The Economist, 17-23 June, p. 1-16. McCullen, P. and Towill, D.R. (2002). Diagnosis and reduction of bullwhip in supply chains. International Journal of Supply Chain Management, 7 (3-4), p. 164-179. Overby, C., McDonnell, F. and Tohamy, N. (2005). Shared strategy study: The state of collaboration 2005. Joint study by Forrester Research, Consumer Goods Technology, and RIS News (November 2005). New Jersey: Edgell. Simchi-Levi, D., Kaminsky, P. and Simchi-Levi, E. (2004). Managing the supply chain: The definitive guide for the business professional. New York: McGraw-Hill. Sun Microsystems (2003). In-house expertise: Sainsbury's seeks supply chain success. Boardroom Minutes: Retail. Retrieved 12 February 2007, from: http://www.sun.com/br/retail_708/article_sainsbury.html Smros, J. (2004). Forecasting collaboration in the European grocery sector: Observations and hypotheses. Department of Industrial Engineering and Management Working Paper, Helsinki University of Technology. Sterman, J.D. (1989). Modelling managerial behaviour: Misperceptions of feedback in a dynamic decision-making experiment. Management Science, 35 (3), p. 321-339. Watson, G.H. (1994). Business systems engineering: Managing breakthrough changes for productivity and profit. New York: Wiley. Zymelman, M. (1965). A stabilisation policy for cotton textile cycle. Management Science, 11 (5), p. 572-580. List of Figures Fig. 1: Sample supply/commodity chain for the apparel industry. Fig. 2: Diagram of bullwhip effect in action. Fig. 3: Degree of actual manufacturer-distributor collaboration Fig. 4: Varying stages of adoption in supply chain collaboration Fig. 5: Best practices in supply chain management to eliminate bullwhip Fig. 1: Sample supply/commodity chain for the apparel industry. (Source: Gereffi, 2002, Figure 1) Fig. 2: Diagram of bullwhip effect in action (Source: Lee et al., 1997, Chart 1) Fig. 3: Degree of actual manufacturer-distributor collaboration (Source: Overby et al., 2005, p. 10SS) Fig. 4: Varying stages of adoption in supply chain collaboration (Source: Overby et al., 2005, p. 10SS) Fig. 5: Best practices in supply chain management to eliminate bullwhip (Source: Geary et al., 2006, Table 5) Read More