The bullwhip lives
Supply chains still prone to inventory oscillations, but Sara Lee, Merloni and others dampen the effect
By Malcolm Wheatley, contributing editor -- Manufacturing Business Technology, 2/1/2004 7:00:00 AM
It's a phenomenon known to business school students everywhere. Early in their studies of supply chain management, they get to play a simulation game. With an innocuous name (The Beer Game is a favorite), and based around a simple production scenario, students get first-hand experience "running" a supply chain, making the kind of decisions that manufacturing executives make in the real world every day.
But whenever it's played, The Beer Game follows a predictable path. Despite steady consumer demand, production and inventories further down the supply chain oscillate wildly. Despite the fact that demand for beer remains pretty much the same from week to week, the supply chain plunges from one crisis to another, overstocked one week and working overtime the next.
What's going on? Simple: it's the "bullwhip effect" at work. So-called because a chart of production or inventories typically follows a bullwhiplike series of extreme curves, the game usually leaves students with a sober realization that running a manufacturing plant or supply chain isn't as easy as they might have supposed.
Trouble is, despite being identified several decades ago, neither college professors nor businesses are any nearer to proclaiming the bullwhip effect to be dead. Indeed, a few years back, the bullwhip effect notched up one of its biggest scalps to date, when Cisco Systems' CEO John Chambers was forced to take a massive inventory write-down.
So intractable are its effects that some leading experts—including Stanford University's Prof. Hau Lee, a leading academic researcher in the field—doubt that the bullwhip effect will ever fully be beaten. "The bullwhip will continue to be a problem for a long time to come," Lee predicts.
So why, exactly, does the bullwhip effect occur? Lee says part of the difficulty of dealing with the bullwhip effect is that it doesn't have a single identifiable cause. That said, the major culprits are easy to pin down. Probably the most common cause is the amplification of the inherent uncertainty that surrounds forecasts. Demand suddenly goes up 10 percent, so you worry about running out of inventory—and order more from the supplier, opting for 15 percent more than usual to be on the safe side. The supplier in turn orders more from his supplier—this time increasing the uptick by 20 percent, as there's clearly something significant affecting demand. And so on. The same thing happens in reverse when there's a demand downturn.
But other factors also are at work. Supply chains that feature pricing promotions, or volume discounts, for example, also are prone to the problem. "Order batching"—where stable end-consumer demand is lumped into batches to (supposedly) facilitate order administration, and reduce processing costs—is another prime cause. So too is the practice—common in some industries, such as semiconductor manufacture—of rationing supplies in time of high demand. In fact, almost any business practice that makes perceived demand lumpier than it really is can cause the bullwhip effect.
Which isn't to say that business is powerless. "Many of today's supply chain solutions address the bullwhip effect in one form or another," says Noha Tohamy, senior supply chain analyst at Forrester Research, a Cambridge, Mass.-based analyst firm. "Every year, we are making progress toward understanding what 'real' demand is, and consequently communicating it better." To Tohamy, it's clear that solutions for collaborative planning and forecasting, supply chain visibility, and supply chain execution all play their part in either determining that real level of demand, or communicating demand and supply better.
Dampen the effect
That manufacturers' employ electronic collaboration techniques to dampen the bullwhip is good news to Jim DeMin, technical manager at Infonet Services Corp., El Segundo, Calif. While beating the bullwhip is tough, the benefits of even ameliorating its effects are usually well worth the investment, he says. For proof, one need look no further than those major manufacturers supplying the retailing and automotive industries, where such tools are a condition of doing business.
"These manufacturers are finding that although they are selling their products to such customers at lower prices than to other customers with whom they don't electronically collaborate, the profit margins that they achieve on the business are actually higher," he says. That's why, says DeMin, even though the required electronic networks represent what he characterizes as "very daunting network challenges," the investment usually is worthwhile.
Thanks to the dampening effect that collaborative networks have on the bullwhip, "even seemingly minor improvements in network performance, availability, and reliability can yield an orders-of-magnitude contribution to business performance," DeMin adds.
Even so, warns Katherine Jones, a managing director for enterprise business applications with analyst firm Aberdeen Group, Palo Alto, Calif., automated tools and techniques can only go so far in minimizing the bullwhip. "The real problem," she says, "is how people think about the supply chain. People will always be tempted to 'feed the forecast'. They think of the just-in-case scenario, not the just-in-time scenario." In Jones' view, computer-based solutions can only go so far when dealing with the vagaries of human-based problems.
That's not to say businesses in the grip of the bullwhip effect are powerless. As manufacturers begin to recognize the underlying causes, they are getting smarter at applying fixes. In other words, if the root cause of the bullwhip effect is a supplier-based problem, then improved forecasting isn't going to have a major impact. Improved supply chain execution capabilities, on the other hand, just might. What's more, better algorithms are only part of the picture: to beat the bullwhip, say the experts, managers need to tackle the organizational and human-based causal factors of the bullwhip, too.
For Italian manufacturer Merloni Elettrodomestici—Europe's third-largest producer of domestic appliances such as refrigerators, washing machines, and dishwashers—beating the bullwhip has involved both improved forecasting and demand planning capabilities sourced from Manugistics, as well as rigorous centralized control of the companies' 14 plants located in France, Italy, Poland, Portugal, Russia, and Turkey.
"Our organization is strongly centralized, and the effect of any forecast inaccuracy—as well as the potential bullwhip effect—is kept under control through our central planning staff, with its control and supervision mitigating a lot of both effects," explains Marco Rossi, a supply chain executive with Merloni.
Spotting root causes
At medical products distributor PSS/World Medical, Jacksonville, Fla., the rollout of a demand planning system from i2 Technologies helped the business see just how great were the human-induced bullwhip effects that it suffered, explains Tom Strubel, director of supply chain systems and business intelligence.
Previously, explains Strubel, demand variability had been buffered by high inventory levels. Implementing the i2 system alongside an ERP solution from PeopleSoft quickly highlighted that a major source of perceived demand variability was no such thing—but was instead a reflection of the number of working days in each calendar month.
"A typical month has 20 days in it—but some have 19, and some have 22," says Strubel. "Especially in the months of August, September, and October, demand was varying by between 10 percent and 25 percent, simply due to calendar effects. When we did our i2 rollout, we didn't at first recognize this—but when we looked into it, it became so obvious." The apparent variability had always been there, he notes, but hidden by inventory.
Other demand-planning changes at PSS/World Medical also minimized the bullwhip effect, including a management review process for sales forecasts—to determine their likelihood—and a more frequent review of actual sales or certain bullwhip-prone items. "We now have our lowest inventory levels ever, and our highest-ever order fill rates," says Strubel. "Previously, these were mutually incompatible."
How much, however, is it possible to dampen—or kill—the bullwhip effect by tackling one of its prime causes in isolation: forecast inaccuracy? And just forecast inaccuracy—without, in other words, making other "soft" changes such as significant organizational improvements?
Some insight comes from Sara Lee Household & Body Care, an Exton, Pa.-based producer of consumer brands including Sanex, Ambi-Pur, and Radox. Since integrating its sales & operations planning process with Prescient's demand planning solution, the bullwhip effect has been greatly alleviated, reports Gary Kahler, director of Sales & Operations Planning. Forecast accuracy has dramatically improved, climbing to 73 percent, and perfect-order fill rate is now 93 percent—both of which have significantly improved overall supply chain performance.
"The biggest change is to shift our focus from forecasting total SKU volume to forecasting volume at the customer level. While we can still generate a bullwhip effect on a small number of SKUs due to forecast error, the incidence of bullwhip problems has been greatly reduced as our forecast accuracy results have continued to improve," concludes Kahler.
Understanding the "bullwhip effect"
The bullwhip effect—the tendency of inventory and production levels to oscillate as misperceived or incomplete information is passed along a supply chain—has more than one cause. Here are a few of them:
| Cause: | Possible cure: |
| Poor demand forecasting | Improved forecasting systems that help companies settle on "one number" forecasts internally, and collaborate externally. |
| Inaccurate demand picture from poorly executed promotions and discount programs | Ensure suppliers understand promotion-based demand spikes are temporary. |
| Poorly executed order batching signals | Tie in order management and other demand more directly with master planning and scheduling. |
| Supply-side uncertainties | Sharing of accurate forecasts and schedules with key suppliers; supply chain visibility and control solutions to obtain an accurate picture of finished goods and other inventory already in the supply chain. |
| Source: MSI | |
