BlueLinx می تواند از روش های مدیریت موجودی نوآورانه برای خرید پیشروی کالا بهره مند گردد

Commodity prices often fluctuate significantly from one purchasing opportunity to the next. These fluctuations allow firms to benefit from forward buying (buying for future demand in addition to current demand) when prices are low. We propose a combined heuristic to determine the optimal number of future periods a firm should purchase at each ordering opportunity in order to maximize total expected profit when there is uncertainty in future demand and future buying price. We compare our heuristic with existing methods via simulation using real demand data from BlueLinx, a two-stage distributor of building products. The results show that our combined heuristic performs better than any existing methods considering forward buying or safety stock separately. We also compare our heuristic to the optimal inventory management policy by full enumeration for a smaller data set. The proposed heuristic is shown to be close to optimal. This study is the first to decide both the optimal number of future periods to buy for uncertain purchase price and the appropriate purchasing quantity with safety stock for uncertain demand simultaneously. The experience suggests that the proposed combined heuristic is simple and can be very beneficial for any company where forward buying is possible.

This paper describes a heuristic developed to improve the purchasing decisions of BlueLinx Corporation, a two-stage distributor of building product materials with annual revenues around eight billion US dollars. Purchasing and selling commodities at BlueLinx is a complex process due to both fluctuating purchase prices and highly seasonal and uncertain customer demand. The company purchases bulk commodities from suppliers such as lumber mills and sells smaller truckloads to customers as requested. The customers do not procure commodities directly from the mill because (1) they do not purchase enough volume at one time to satisfy the minimum mill quantity requirement, or (2) they do not want to give up the flexibility of shipment size and destination that is absorbed by the two-stage distributor. BlueLinx can charge a positive margin by absorbing lead-times, breaking bulk, and providing fast deliveries. However, a highly variable portion of their profit or loss is derived solely from the difference between the price they purchase the commodities at versus the price they sell them at. Due to the competitive nature of their business, the price BlueLinx can charge for its product is determined by market forces and may be considered exogenous to BlueLinx's decision making. Thus, strategic purchasing that minimizes the cost of acquiring the product provides BlueLinx with the largest opportunity for improving profits. In this paper, we provide insights into BlueLinx's problem by modeling a two-stage distributor that has a purchasing opportunity at a known, current cost with forecasts for future demands and a known distribution for future costs. The distributor's decision is whether to buy enough products to satisfy demand only in the period 0 (first or current period where inventory on hand can be increased by a purchase) or to also buy to meet demand in future periods (forward buy periods beyond the vendor delivery lead-time). We propose a heuristic for this problem that is a combination of two existing methods for determining the optimal number of future periods to buy and the order-up-to levels under an uncertain cost and demand environment. The goal is to maximize the total expected profit. We use actual sales data (simulated through a bootstrapping technique) from BlueLinx for the years 2001–2005 to demonstrate the effectiveness of the proposed heuristic. The results show that our combined heuristic performs better than any existing methods considering forward buying or safety stock separately. We also compare our heuristic to the optimal inventory management policy by full enumeration for a smaller data set. The proposed heuristic results also show our method to be close to optimal. This study is the first to decide both the optimal number of future periods to buy for uncertain purchase price and the appropriate purchasing quantity with safety stock for uncertain demand simultaneously. The study suggests that the proposed combined heuristic is simple and can be very beneficial for any company where forward buying is possible. We begin by describing BlueLinx's purchasing environment.

Given the fluctuating prices of commodities, forward buys make sense. Even though wood and other commodities are non-perishable, the newsvendor formula can be applied to the demand distribution to balance profit and holding costs. Golabi's method should be used to determine the number of periods to forward buy given supplier lead-times, current purchase price, and expected future purchase prices. However, the quantity to procure should not be the point forecast but rather it should be based upon the distribution of demand. The distribution of demand should reflect more uncertainty through increasing prediction intervals. By utilizing Gavirneni's application of the newsvendor equation to the increasingly spread out demand distribution as part of our GOGA heuristic, we achieve better results overall than either Golabi or Gavirneni's methods achieve in isolation. Forward buys as outlined in Golabi's method clearly increases profits over not doing so as shown in Tables 4 and 5. The modified newsvendor equation of Gavirneni provides advantages over buying no safety stock as shown in these same two tables. Our GOGA method combines forward buys and safety stock based upon a modified newsvendor equation for maximum profit improvement. For plywood purchasing the GOGA method has the best expected profit overall. In Table 6, we showed that the GOGA heuristic achieves near optimal or optimal for smaller data sets. The real world demand and price data are much larger and less predictable, so the GOGA results when applied to the BlueLinx data cannot be said to be either close or far from optimal with any certainty. The Director of Global Sourcing at BlueLinx has written that he is convinced by the simulation results that his company will be able to significantly improve the bottom line with this new method. He has formally requested BlueLinx's IT resources to review options for developing programs utilizing these formulas into their enterprise resource planning (ERP) system. The company believes that this more formal approach will improve their profitability on commodity buys but has requested that only the method be shared publicly, not its expected dollar profit improvement or volume of commodity sales as those are competitive secrets. An article in the Atlanta Journal-Constitution [16] on November 2, 2006, stated, “The slump in the housing industry is causing real pain, and not just among home builders …… In a conference call with analysts, CEO Stephen E. Macadam said BlueLinx eliminated about 8 percent of its work force, including 175 salaried employees and 100 hourly workers.” The timeline for the implementation of the GOGA method is now uncertain. This new method will be programmed into the existing routines in the homegrown ERP system. A table needs to be added to store the historical commodity price data to allow future price forecasting via an autoregressive function. The role of adding and ownership of the data in this new table will be assigned as additional duties to an employee in the purchasing department. Demand forecasting and order up to formulas currently exist in the ERP routines, but do require modifications to perform the new GOGA method. The change to the calculations is behind the scenes and will result in new quantities in the recommended purchase quantity field. Because product specialists already order the system recommended amount on a PO, no new training is required for existing procurement staff. It is important to note that although this method was demonstrated on plywood for one particular company, it is applicable to any industry where purchases prices fluctuate from purchase period to purchase period. To model the long lead-time vendor, we shift the Golabi period 0 out three months and use wider prediction intervals to reflect the increased demand uncertainty. Some real examples from companies in different industries are; Boeing's Commercial Aircraft Group requires titanium to build the frame for the cockpit. Potash Corporation buys natural gas to process the potash into dry, granular fertilizer. A plastic film converter faces highly variable prices of mill rolls of polystyrene from Dow because this plastic is petroleum based. RR Donnelly is faced with volatile paper prices from mills to print and bind books. Pepsi-Cola General bottlers must decide how much sugar to buy and when. As these few examples illustrate, for many products, and in many industries, fluctuating purchases prices are a normal part of business decision making. The GOGA method can be applied to all these situations for profit improvement.