by David Jacoby
The voyage from the farm to the home can take a whopping 110 days,10 due in part to rapid consolidation in the processing industry, which has increased the physical distance between the growers and the households that consume the end product. In response, the Home Grown Cereals Authority (HGCA) worked with the Cereals Industry Forum to foster better communication between farmers, storage, marketing and trading companies, primary and secondary processors, and transporters with the objective of accelerating the flow of product through the pipeline.11
Automotive supply chain
Automobile manufacture used to be carried out by a small number of national car companies in their domestic plants, but globalisation has transformed that model dramatically. Today steel coils are produced in Japan and exported to other countries worldwide. For example, US automakers import steel coils to New Orleans, and then ship them by rail and truck to Chicago, Indianapolis and Detroit. Steel slitters and metal stampers make parts out of them and deliver them to auto assembly plants in places such as Lordstown, Ohio, which also receive components like engines and transmissions from other suppliers. Once the finished vehicles are complete, they are sent by rail to mixing centres, where they are regrouped by destination and trucked to dealers nationwide. As the vehicle requires service parts, automakers operate networks of stocking centres that provide parts as rapidly as needed. In Europe, service parts are distributed by truck to distribution centres or in some cases airfreighted, for example from manufacturing plants in continental Europe to distribution centres or service centres in the UK or Ireland.
Chemical supply chain
The journey from crude oil to a white wall involves many steps, at least if you use oil-based paints. Crude oil is shipped from places such as the Arabian Gulf to ports where extensive petroleum processing facilities are located. In the United States, many of these facilities are located in Texas and Louisiana, and Europe has similar capabilities at ports such as Antwerp (Belgium), Rotterdam (Netherlands) and now Teesside (UK). These processing plants convert the petroleum, sometimes in two or more stages of refining, into base chemical products such as polymers, which are used in adhesives, building materials, paper, cloth, fibres, plastics, ceramics, concrete and house paint. The polymers are transported by rail to more specialised chemical plants that make them into emulsions. The emulsions are then sent by rail to the paint manufacturers. The paint is ultimately shipped by rail or truck to distributors, who package it in smaller units such as cans or drums. Large distributors have many distribution centres, which collect different types of products and sort them into outbound loads mixed with many different shipments for customers in a wide geographic range. The long and multi-tiered supply chain involved in paint manufacture means that there is a lot of inventory and about a third of it is in between processing locations at any point in time. Therefore, management of these inventories is like a moving target, and takes skill and collaboration to control so that the right product gets to the right place at the right time.
2 The bullwhip problem
Long supply chains dramatically increased the complexity of almost any firm involved in selling any product since 1980. International trade expanded, and starting with Wal-Mart on the consumer side and maintenance, repair and operating (MRO) supplies on the industrial side, most western companies began importing some raw materials or components from Asia during this period.
Without systems, processes and information to manage the long pipeline of product, retailers were doomed. With 40 days on the water and in customs, and 40 days in domestic distribution centres and trucks, consumer products companies and retailers needed to have a clear picture of which T-shirts and toys were going to be in demand about three months before they reached the shelves. They also had to know how much of what they already had in the stores would be sold at that point – a nearly impossible task.
Most retailers struggled to match supply with demand, and failed. The number of green size 8 T-shirts on the shelf never seemed to be right. Either there would be too many, and they would be heavily discounted or thrown away as obsolete; or there would be too few. The latter case would be the hardest to diagnose and to remedy, since it was usually impossible to tell how many would have been sold if they had been available on the shelf at the right time.
The period was expensive and painful. Retailers lost margin and threw away large amounts of unsold product. Consumers could not find what they wanted on the shelves, and frequently settled for something other than what they really wanted.
Retail stock-outs
The problem became most evident in the grocery industry and in fast-moving consumer goods (FMCG), which includes lower-value household items that are sold in consistently high volume such as toilet paper and toothpaste. One study1 forecast that grocery chains could save 10% of their costs by implementing a system that replenished these goods based on actual consumption rather than on projected consumption. Retailers and the companies that supplied them, called consumer packaged goods (CPG) companies, established a movement around the effort to do this. It was called efficient consumer response (ECR). Fourteen major US retailers joined the effort.
Implementing ECR was a big challenge. Most companies only knew how to operate on a push system, whereby they would forecast consumption and replenish by assuming that a certain amount of the forecast had been consumed, even if it was not. An elaborate system of reorder points triggered replenishment, most often before the product was actually sold, in order to have it available in the store when the product really did sell.
A French grocery chain, Monoprix, was involved in one of the early efforts to implement ECR. Several major chains had achieved some success with this in the United States, and Monoprix and its hypermarket competitors were pioneering the technique in France in the early 1990s. Implemented in real-time to various degrees, all the chains were trying to record actual sales and then pull the exact amount sold all the way through the supply chain by transmitting readings from the retail store systems back to the source supplier.
Tesco also pioneered supply chain concepts during the 1980s and early 1990s. The company implemented lean processes and systems that were designed to help it synchronise its supply chain. For example, it implemented point-of-sale scanning and electronic data interchange (EDI). It also used cross-docks to consolidate multiple shipments on to a single vehicle so that its store deliveries could be more efficient. These advantages gave it an advantage when expanding internationally and competing against entrenched local stores.2
Even though the bulk of early effort was put into consumer goods and retailing, the principles applied to other industries as well. Much later, researchers quantified the opportunity. One study concluded that forecast errors at bulk chemical producers ranged from 10% to 24%, and had an average error rate of 26% and a median error rate of 11%.3
Academics’ discovery of the bullwhip effect
Academics were the first to identify the root cause of retail stock-outs, which came to be called the bullwhip effect. Yanfeng Ouyang describes the concept in practical terms as “where a small perturbation at the handle (customers) causes huge movements at the tip (upstream suppliers)”.4 It is called the bullwhip effect because the phenomenon resembles the way a whip oscillates when flexed. A Procter & Gamble (P&G) executive used the term “bullwhip” in the early 1990s to describe the fact that although diaper (nappy) usage was relatively constant, P&G’S order flow from retailers was cyclical and highly volatile.5
John Sterman, in Octavio Carranza Torres’ book The Bullwhip Effect in Supply Chains explored this issue in detail as it applies to economies and industries.6 He points out that the oil and gas industry provides a stark example of the bullwhip effect. While the volume of oil and gas consumed is remarkably constant, drilling owners and operators and oilfield service activity is highly cyclical. Perceived disturbances in demand and supply, transmitted through the price of crude, cause predictable upward and downward swings of up to nearly 50% in well-drilling a
ctivity, based on the 33-year trend shown in Figure 2.1.
Figure 2.1 The bullwhip effect in the oil industry: oscillation in the oil and gas supply chain
Source: Author’s analysis of US Federal Reserve data
In other industries, the impact is more subtle. The semiconductor industry showed signs of amplitude magnification – minor swings in industrial production led to wider and successively increasing fluctuations in semiconductor production – between its birth in the 1950s and the early 1990s (see Figure 2.2). In this case, the semiconductor production responds belatedly to industrial production and overcorrects for fluctuations therein. However, while the swings in industrial production are + or –10%, the resulting swings in semiconductor production are up to five times that amount. The bubble in the 1995–2001 period was a combination of bullwhip and the internet boom and bust.
Figure 2.2 The bullwhip effect in the semiconductor industry
Source: Author’s analysis of US Federal Reserve data
A similar phenomenon can be observed in the machine tool industry, which responded belatedly and excessively to the peaks and valleys of the automotive production cycle. While US automotive demand fluctuated by no more than 20% between 1970 and 2000, machine tool orders lagged behind auto production and then overcompensated, resulting in boom-and-bust cycles that damaged sales and share prices.
Longer supply chains magnify the impact of the bullwhip effect and increase the amount of inventory held across the system. Conversely, the fewer the layers in a supply chain, the less the resulting bullwhip effect. Hewlett-Packard measured the bullwhip effect by comparing the standard deviation of orders at the stores with the standard deviation of production at the upstream suppliers.7
The rules of ordering, such as the timing of order placement, the acceptance or refusal of back orders, order quantities and lot sizes, and cancellation rights and penalties, can have an enormous impact on the total system inventory and the bullwhip effect when, for example, there is a holiday demand surge.
The effect of ordering rules and exceptions can be chaotic in a large and interdependent system. Chaos theory, first outlined by a French scientist, Henri Poincaré, in 1890, tried to decode the spread of chaotic patterns. Edward Lorenz introduced the “butterfly effect” to point out the sensitivity of the ensuing pattern to tiny changes in the initial condition. Lorenz initially published his theory in 1963 at the New York Academy of Science, and in 1972 he gave a now famous presentation entitled “Predictability: Does the Flap of a Butterfly’s Wings in Brazil Set Off a Tornado in Texas?” to the American Association for the Advancement of Science in Washington, DC.
The Lorenz attractor was one of the first mathematical systems that described chaos. Using three equations, Lorenz determined the evolution of a system (see Figure 2.3). Lewis Dartnell of University College London explains: “The Lorenz attractor never settles on a single point, nor ever repeats itself – it forever whirls round and round the double-lobes seen in the diagram.” The system is deterministic, prescribed by three equations, and there is no way of predicting or calculating what value the system will take in the future without actually simulating it all the way and finding out. “It is like a fly that buzzes around inside a room, its direction of flight at any time is determined by equations depending on where it currently is relative to the bottom corner (i.e. its x,y,z co-ordinates inside the room).”
Figure 2.3 The butterfly effect
Source: University College London (copyright Lewis Dartnell at lewisdartnell.com)
Complex transportation networks such as airlines and railways can sometimes be described as exhibiting chaotic operational patterns. After system delays and disruptions – for example, after severe storms – the schedules can be so out of alignment that the only way to get them back on schedule is to reset by starting afresh the next day or the next week.8
The bullwhip effect is evident in macroeconomics as well as supply chain management (SCM) because of the human propensity for delayed response and overreaction. Stockmarkets over extended periods of time show signs of bullwhip overcorrection, resulting in boom-and-bust cycles. Business cycles are generally viewed as the result of the bullwhip effect of inventory flowing through supply chains and causing alternating excess demand (growth spurts accompanied by inflation) and overstocks (recessions sometimes accompanied by falling prices).
Barilla Pasta, an Italian foodmaker, was quick to measure the bullwhip effect. Demand at its central distribution centre had a standard deviation nearly four times that of the demand at its regional distribution centre. The implication was that almost three-quarters of the fluctuation in demand from one level to the next was attributable to problems in the restocking process.9
Although some bullwhip effect is inevitable in any system with feedback loops and delays, Barilla’s and other corporate and academic research identified aggravating but controllable factors, the effect of which becomes more severe as the number of layers in the supply chain increases. The factors are as follows:
Overcorrection. Buyers order before they are ready to consume, in order to keep a stock buffer “just in case”, or by opportunistically profiting from market price fluctuations, or by hoarding in order to prevent or mitigate a shortage or possible shortage. Overcorrection is the worst aggravator of bullwhip. This tendency is aggravated by long order cycle times, which reduce buyers’ sense of confidence and make them overorder. But when five players in succession each order 10% extra, the company farthest from the customer (usually the manufacturer) will end up ordering 61% extra (1.105).
Promotions. Sellers create incentives for buyers to buy when they would not have otherwise bought, thereby creating an initial disturbance that ripples through the supply chain. This is the case with quarter-end low pricing or clearances and many other promotions.
Batching. Buyers batch orders to get volume discounts and to lower their production set-up costs as determined by their reorder point formulas, their inventory costs as determined by economic order quantity (EOQ) formulas and their transportation costs as determined by their distribution resource planning (DRP) software.
Tweaking. Buyers update demand forecasts, creating changes to which their suppliers and their suppliers’ suppliers react.
At first glance, it may seem that “buy now”, “bulk up” and “tweaking” (factors 2–4) could be addressed by software and perfect information. However, as the beer game (see page 9) can illustrate if played with a scoreboard visible to all, even experts make bad ordering decisions in the face of good information, because it is still necessary to guess (“forecast”) retail demand (or trust the provider of that information) and to compute the requirements for the product through the whole supply chain, and to face impending shortages without ordering just a little extra. Moreover, supply chain software has to be extremely sophisticated to operate successfully across multiple tiers of a supply chain, and getting better information requires trading partner collaboration, which relies on trust between corporate entities that in many cases have had transactional, guarded relationships for years.
Why supply chain management became a buzz word
These four factors are clearly problematic, but if they were the only problems, SCM would remain the domain of a narrow band of logistics and supply professionals. SCM has gained currency and become mainstream, discussed and analysed in the press because of factors such as cost, security, compliance, safety and the environment.
Cost competition
Global cost competition
Competition and consolidation in the retail sector heightened awareness of the need for and benefits of SCM. Retail chains such as Macy’s were purchased by holding companies such as Federated Department Stores. Grocery chains such as the Great Atlantic & Pacific Tea Company (A&P) lost money and their ownership changed hands. In Europe and South America, hypermarkets such as Carrefour eroded the margins of traditional centre-city grocers and merchants. In Japan, western firms such as 7-Eleven and McDonald’s a
te into the market share of small stores everywhere. All players – both the winning and the losing parties – looked to SCM to gain a competitive edge.
The problem was not confined to retail. Industrial companies also faced cost pressure and looked to SCM for help. Starting with automotive companies, which faced direct cost pressure from Japanese competitors, European companies turned to better management of their service (after-market) parts supply chains for help in reducing costs. American automakers did likewise. Companies in other industries, such as chemicals and mining, faced price erosion (see Figure 2.4), which stimulated them to search for cost reduction solutions, and SCM has been a large focus of that search.
Figure 2.4 US GDP and manufacturing prices
Sources: Boston Strategies International analysis of data from the US Bureau of Economic Analysis; US National Association of Manufacturers
More recently, service-sector companies have faced a similar need to reduce their costs through operations improvement, and consultants have adapted SCM principles to help them become more competitive, under the name of service chain management.
Fuel and labour costs
Spikes in oil prices and consequently fuel costs heightened interest in SCM in 2007 and 2008. Oil prices increased by 380% between 1993 and 2008, and price volatility became the norm.
While some carriers, such as those in the parcel express business, implemented surcharges, others such as railways rebased their rates to embed fuel surcharges, effectively using increased fuel costs as a profit centre.