A Decision Support System for Vendor Managed Inventory.(Brief Article)

Recently, vendors and retailers have begun to forge cooperative agreements to manage inventory, which requires sharing demand information and setting mutually agreed upon performance targets for the supply chain. This paper describes the market forecasting and inventory management components of a Vendor Managed Inventory (VMI) decision support system and how this system was implemented by a major apparel manufacturer and over 30 of its retail partners. The DSS also helped the vendor and retailers arrive at jointly agreed upon customer service level and inventory turnover targets. As a result of implementing this VMI system, customer service levels improved dramatically, often coupled with a significant improvement in inventory turnover. The VMI performance results relative to the existing system and related insights for supply chain coordination are discussed.

Vendor Managed Inventory (VMI) systems have been initiated by certain manufacturers to improve both retail customer service levels and inventory turnover. VMI systems achieve these goals through more accurate sales forecasting methods and more effective distribution of inventory in the supply chain. The VMI system allows the retailer to expand the assortment of the vendor's products that can be offered within a given retail space. This improves the profitability of the vendor's brand for both the retailer and the vendor. Retailers working alone are generally not able to achieve the same productivity increases because the vendor is the one able to provide a more responsive replenishment system based on more precise demand information.

The VMI system we describe uses a decision support system (DSS) to develop a specific model for each retailer that provides weekly sales forecasts (even in difficult promotional retail environments). The DSS further assists the user with decision methods for inventory management and models for measuring and improving system performance. Under the VMI agreement, the retailer provides seasonal promotional plans and sales data to the vendor. The vendor then produces the sales forecasts and supplies the inventory to meet agreed upon customer service levels and inventory turnover targets. The DSS also helps the vendor and each retail partner to jointly set customer service levels and inventory turnover targets by evaluating alternative inventory level and replenishment decisions and identifying achievable system performance levels.

The DSS described here has been implemented by a major apparel manufacturer in conjunction with over 30 of its key retail accounts. The VMI system has been highly successful. Over a three-year period, service levels have improved dramatically, with minor increases in the retailers' inventory levels. In many cases, the participating retailers' buyers have found that the VMI system's forecasts are more accurate than their own and they now rely on these forecasts for merchandise planning.

One of the vendor's primary objectives in this VMI system was to increase customer service level through better forecasts and more effective use of inventory. This would lend to a higher availability of sizes and color choices, thus making this vendor's brand relatively more attractive and therefore achieve increased sales. In principle, one could argue that it might be difficult to persuade the retailers to increase their service levels, as opposed to simply using the VMI's improved inventory management to increase their inventory turnover. However, in this instance, the participating retailers had set high goals for customer service level, but were failing to meet them because of inaccurate sales forecasting and/or suboptimal inventory allocation. Thus, when the VMI system caused the customer service level to increase with a relatively small amount of additional inventory investment, both the vendor and the retailers experienced improved results.

Related Supply Chain Literature

VMI can be viewed as an example of channel coordination, which has been studied by both marketing and supply chain researchers. The supply chain literature, recently surveyed by Tsay et al. (1998), develops an economic rationale for why retailers and vendors may choose different levels of inventory investment. In particular, Narayan and Raman (1997) and Cachon (1997) study the coordination of inventory levels when the retail price is fixed and develop models for the effects of VMI on service level and channel profits. They argue that the vendor may have a higher perceived cost of stockouts than the retailer. This may occur because the retailer's profit margin per unit is lower than the vendor's, or because a customer may simply substitute a different brand at the same store when a stockout occurs. This clearly results in a shortage cost that is larger for the vendor than for the retailer. These cost differences can cause retailers to select a lower customer service level than the manufacturer would prefer.

The marketing models, beginning with the seminal paper by Jeuland and Shugan (1983), compare both the prices and the "level of channel effort" that retailers choose, versus those the vendor would choose, to optimize the complete channel. For our VMI application, retail prices were fixed, due to legal considerations. Thus, the VMI channel coordination problem in this paper more closely resembles the situation studied by Desiraju and Moorthy (1997). They consider the case of demand uncertainty and propose jointly set performance requirements as a means of coordinating channel effort among the retailers. In the VMI context, the retailer's inventory investment and the corresponding customer service level can be viewed as an example of "channel effort" that has a decreasing marginal impact on demand. The conclusion of the related marketing literature is again that without channel coordination retailers will tend to provide a lower level of "effort" than is needed to maximize channel profits.

Collaborative, forecasting, planning and replenishment, or CPFR [1], systems also improve sales forecasts through information sharing between retailers and vendors. CPFR agreements typically go further than this paper's VMI system in that they develop joint business plans with specified financial remedies when targets are not met. On the other hand, a CPFR system need not include a vendor managed DSS for sales forecasting, inventory management and performance analysis. A CPFR approach to sales forecasting may not lead to the same improvements in accuracy as VMI, since they are typically collaborative subjective forecasting efforts. Quick Response (Q/R) systems, which have also been implemented by a number of vendors, focus on reducing the lead time for replenishment, which in turn reduces the need for safety stock and thus improves inventory turnover. A VMI system can potentially achieve these same objectives.

This paper makes the following contributions to retailing research in supply chain management. It:

* Combines promotion response and parameter updating models developed in the marketing and forecasting literature with inventory management and system performance models developed in the management science literature. This is a novel integration of these individual models into a single decision support system.

* Describes a major multicompany implementation of a VMI decision support system and the resulting improvements in supply chain performance.

* Provides an example of how certain incentive incompatibilities in retail supply chain management can be resolved and discusses the relationship to channel coordination theory.

VMI BACKGROUND AND BENEFITS

A key business motivation for developing VMI replenishment systems is to develop a deeper partnership between the vendor and key retail accounts. Some of the specific goals cited for this system were to: (1) give the retailers' customers the best opportunity to purchase the vendor's products, (2) help the retailers manage their inventory more effectively, and (3) assist the vendor in production scheduling [2].

To understand the importance of enhancing the vendor/retailer relationship, it is useful to consider the typical retail decision-making environment for forecasting and inventory management. Buyers for department store chains are responsible for 20 to 40 product categories, which may include thousands of individual SKUs at the style-color-size level. For example, a product category might be Levi 501 stone washed denim jeans. Buyers generate an aggregate (chainwide or regional level) forecast and use historical ratios to forecast sales allocations among the individual SKUs and stores (e.g., 40% of past sales were medium size, 2.2% of sales in the Eastridge store).

Buyers are also responsible for planning and executing temporary price promotions that include …