From Here to There

From Here to There

Logistics Researchers Keep Shelves Stocked at Stores Everywhere

A vein of Matt Waller’s research would not exist if it weren’t for a decision Sam Walton made many years ago. By the late 1980s, the visionary and now iconic founder of Wal-Mart Stores Inc., which last year posted sales of $419 billion, had already broken a few cardinal rules of retail.

For one, he decided to share sales and other information with department managers with the hope that these individuals would use the information responsibly to improve overall performance. It worked. The information enabled managers to see beyond their department and empowered them to make decisions that served the customer and benefitted the company.

Walton then took one step further and expanded this concept outside the company. To better manage inventory, he decided he needed more information from suppliers, and, likewise, they needed more information from his stores. So at Walton’s behest, Wal-Mart bucked convention and started working closely with major suppliers, particularly Procter & Gamble. Then, against the advice of venerable business leaders outside the company and directors within, Walton took a big risk in the early ’90s and began providing point-of-sale data with all suppliers.

Critics feared suppliers would share the sales information with competitors, which would cause Wal-Mart to lose its competitive edge. Walton listened but held firm and followed his instincts. In fact, he did not mind if suppliers shared the information with other retailers. This plan also worked, perhaps more so than he imagined.

In addition to the desired effect of increased profits and satisfied customers, Walton’s radical idea created a revolution in supply-chain management practices. No longer did suppliers have to wait for manual orders from store managers. Relying on raw sales data, suppliers could forecast consumer demand independently. Walton gave them the freedom to ship more goods that were selling and less that weren’t. And because all parties had access to the same information, Wal-Mart, as long as their analysts were sharp, did not have to tolerate greedy or unscrupulous suppliers. While Walton’s idea may not have eliminated human error from supply-chain management processes, it did render them more scientific.

Waller was working on a doctorate in business logistics at Penn State when all this Wal-Mart business was going on. He remembers reading a story about it in a business magazine. Like others, he was surprised at Wal-Mart’s direction, but he also sensed that something big would come out of it. He did not know that it would lead to the company becoming the “most sophisticated supply chain management organization in the history of the world.”

A year later, 1994, Waller was teaching at the University of Arkansas, and since then he has monitored and occasionally analyzed Wal-Mart’s supply-chain management practices. He and other researchers have studied the practice of retailers sharing point-of-sale data, which industry practitioners have dubbed a “bottom-up” approach to forecasting consumer demand. As Wal-Mart’s stock continued to rise – in large part because of the company’s command of logistics and supply-chain management – and as more and more studies demonstrated the general superiority of sharing point-of-sale data, Waller expected every retailer to follow Wal-Mart’s example and adopt the bottom-up approach. But this has not happened.

“It still hasn’t caught on,” he says. “Only a handful of major retailers use it, which is kind of baffling when you consider Wal-Mart’s success. I think these other stores still worry they’ll lose a competitive edge if they share private information. But the main thing that holds them back, I think, is a basic lack of knowledge about supply-chain management practices.”

This condition – the general lack of knowledge, despite many studies showing that the bottom-up approach outperforms basic retailer order history – has driven much of Waller’s research focus over the past several years. The fundamental question is: How can suppliers improve accuracy of order forecasting, regardless of the amount of point-of-sale data? He says the inability to do this is the greatest obstacle to establishing and maintaining an appropriate amount of goods on retail shelves.

In late 2009, Waller and Brent Williams, professor at Auburn University at the time but now a member of the Sam M. Walton College of Business new department of supply chain management (see related story), released findings from a major study of the retail and consumer-packaged goods supply chain. (A previous Grocery Manufacturers of America study reported that half all retail out-of-stocks in the consumer goods industry were linked to poor ordering, replenishing and forecasting processes.)

Photo by Russell Cothren

To try to reconcile forecasting discrepancies between the point-of-sale approach and forecasting based on store orders – also called “top-down” forecasting – Waller and Williams applied something called vector error correction model, or VECM, to logistics. Frequently used in macroeconomics and other disciplines, VECM is a dynamic system that adds error-correction features to performance models that have multiple variables. A critical function of VECM is that it moderates deviations of the current state with long-term results and then applies this equilibrium to short-term dynamics.

Would VECM improve the accuracy of short-term forecasting orders? Armed with 104 weeks of data from a global consumer packaged-goods company, Waller and Williams tested the performance of the error-correction model in the ready-to-eat cereal, canned soup and yogurt categories. They first found a relationship – a “long-run equilibrium” – between forecasting based on point-of-sale data versus store orders to distribution centers. Establishing theoretical evidence for this implied that variables within the relationship followed an error-correction process and allowed the researchers to empirically examine whether conventional statistical conditions for using the error-correction model were useful.

“We found several combinations where point-of-sale and distribution-center orders were co-integrated,” says Waller.

However, in a majority of the combinations in which point-of-sale information was non-stationary – meaning statistical properties of sales data changed over time – orders remained stationary. The researchers applied the model under these conditions and found that it improved accuracy for short-term forecasting of orders. Their analysis demonstrated that the model generally improved forecast accuracy even when some of the statistical conditions for applying it did not hold.

The findings, Waller says, will improve important supply-chain measurement standards, such as inventory turnover, gross margin return on inventory investment and in-stock levels. Improvements in these areas will lead to greater service and convenience for consumers and increased profits for retailers.

“Clearly there are benefits to be gained from use of the model,” Waller says. “From a statistical perspective, it isn’t complicated, and it can be integrated into most commercial logistics software packages. The only problem is that it requires both order history and point-of-sale data, which, as we know, must be obtained from the retailer. So in this sense, this study just adds to the body of research suggesting that there is substantial benefit to incorporating point-of-sale information into an order forecast.”

In a subsequent study, “Top-Down vs. Bottom-Up Demand Forecasts: The Value of Shared Point-of-Sale Data in the Retail Supply Chain,” Waller and Williams dug deeper into top-down versus bottom-up debate by examining specific contexts in which either approach might be appropriate. Although findings from this study did not refute previous statements about superiority of sharing point-of-sale data, they did illuminate the limitations of working solely off point-of-sale data and highlighted several situations in which a top-down approach led to greater forecasting accuracy and other benefits.

Analyzing point-of-sale and order data for 10 ready-to-eat cereals from 18 regional U.S. grocery distribution centers, Waller and Williams confirmed that relying on retail point-of-sale data can increase the accuracy of predictions and reduce forecasting error, especially when the supplier is trying to predict consumer demand. However, the researchers quickly pointed out that a top-down approach works better when point-of-sale data is not readily available and when suppliers are trying to forecast demand at the account level, meaning forecasting for specific clients and their distribution centers.

Before a further explanation of how forecast demand works at the account level, it is important to mention what demand forecasts mean to suppliers. Most importantly, forecasts enable suppliers to establish and adjust production schedules, plan and procure transportation, and position inventory across the distribution network. In the long-term, demand forecasts help suppliers determine projected inventory requirements, create production schedules, determine capacity requirements and procure raw materials. In short, forecasting demand means everything to suppliers.

For something as critical as transportation planning, suppliers forecast demand for “ship-to” locations, which generally are retailer-owned distribution centers. This bottom-up approach, referred to as “ship-to demand forecasts,” dictates that the supplier creates demand forecasts for each ship-to location. However, because a few large retailers may account for a large portion of the total retail market share and require different management techniques, many suppliers create “account-level demand forecasts.” These represent the total expected demand across all of the retail customer’s ship-to locations, rather than individual totals for each location. At the account level, the supplier creates a single forecast for the customer’s total demand, which includes all distribution centers, and then disaggregates the total forecast into individual forecasts for each ship-to location.

Because suppliers must effectively distribute inventory throughout their network, as well as make customer-specific marketing and sales plans, they must forecast at both ship-to and account levels, Waller says. He and Williams found that if a supplier forecasts retailer orders or actual demand at the ship-to level, the use of point-of-sale data can reduce forecast error. However, at the account level, forecasts based on basic order data appeared can achieve more accurate forecasts for large retailers. The findings also questioned the overall value of point-of-sale data when demand forecasts pertained to broader decisions, such as production or capacity planning. Still, for short-term decisions such as inventory or transportation planning, point-of-sale data increased forecast accuracy and improved performance.

An important insight gleaned from the study was that large retailers share point-of-sale data with suppliers because they have the technology and resources to do so. But, says Waller, this type of sharing may be even more beneficial for small retailers.
“The bottom line is that the choice of a method – top-down or bottom-up forecasting – really should depend on the availability of shared, point-of-sale data,” Waller says. “Clearly, sharing the right data in appropriate contexts leads to greater accuracy when forecasting demand in the retail supply chain.”

Logistics Department A Logical Outcome

Matt Waller sits in his office and marvels at the concentration of transportation leaders in Arkansas: J.B. Hunt, the world’s largest truckload carrier, in Lowell; FedEx Freight, the world’s largest less-than-truckload carrier, in Harrison; ABF Freight System in Fort Smith; USA Truck in Van Buren; PAM Transport in Tontitown; Willis Shaw Express in Elm Springs; Maverick in Little Rock. And these are only the major players. Include the smaller companies, and the list goes on and on.

Their presence, combined with Wal-Mart and logistics-intensive suppliers such as Tyson Foods, make Northwest Arkansas a supply chain management mecca of sorts, not only a physical hub but an incomparable locus of opportunity and knowledge. Waller knows there is perhaps no better place to study logistics and supply chain management.

“We have something really special here,” he says. “These companies – many of which have our graduates as senior executives – and Wal-Mart, the most innovative supply-chain management organization in history of the world, make this area a truly unique business environment. And that’s forgetting many logistics-oriented suppliers and manufacturers. For me, it’s exciting to have a part in educating supply chain leaders for these companies.”

Considering the convergence of major industry players and intellectual capital, it makes sense that one of the nation’s preeminent supply-chain management programs exists at the University of Arkansas. And as Waller alluded to above, these are indeed heady times. In late March, the U.S. News & World Report’s “2012 America’s Best Graduate Schools” guide recognized the university’s supply-chain management and logistics program by ranking it tied for ninth place among all public graduate business schools and tied for 14th place among all universities.

The announcement above came only two weeks after university officials, responding to market demands and following through on an idea that had been discussed for many years, announced that the supply chain management program had been elevated to official department status within the Sam M. Walton College of Business. The change became effective July 1. Waller, holder of the Garrison Chair in Supply Chain Management, was appointed department chair.

“As the United States moves toward a more competitive global economy, there will be an increasing demand for more efficient logistics systems and highly qualified people to manage them,” Walton College Dean Dan Worrell said at the time.

Waller concurs. He says there is high demand for logistics expertise, for graduates who understand supply-chain management practices and how distribution, transportation and retail work together to move goods from the manufacturer to consumer.

“Compared to other disciplines, supply-chain management is young,” he says. “But it’s been around for some time now – on this campus and others – and many academics have offered findings that have increased efficiency and improved business processes. So it’s gratifying to be recognized and to be acknowledged as a formal academic discipline.”


About The Author

Matt McGowan writes about research in the College of Engineering, Sam M. Walton College of Business, School of Law and other areas. He is the editor of Short Talks From the Hill, a podcast of the University of Arkansas. Reach him at 479-575-4246 or

University Relations Science and Research Team

University Relations Science and Research Team

Matt McGowan
science and research writer

Robert Whitby
science and research writer

Looking for an expert?

The University of Arkansas Campus Experts website is a searchable database of experts who can talk to the media on current events.

Trending Topics:
State and local economy
Environmental economics
Immigration politics

More on University of Arkansas Research

Visit the office of Research & Innovation for a complete list research awards and more information on research policies, support and analytics.

Connect with Us