STRATEGIES FOR MANAGING THE RISK OF HOLDING INVENTORY

Managing the risk associated with owning and maintaining inventory over time is an important concept to apply in supply chain management. Basically, this risk increasingly stems from (1) increasingly shorter product life cycles, (2) the proliferation of SKUs, implying the fragmentation of original aggregate demand due to the introduction of new products that are more difficult to forecast, and ( 3) the growing segmentation of markets, possibly implying the opening of new distribution centers or warehouses to guarantee the level of service. All these factors favorably contribute to the occurrence of misunderstandings in the allocation decisions (location) and replacement (how much, when and how) of inventories.

Normally, the degree of risk associated with keeping stocks is measured through the variability of demand for a given product and/or distribution channel, calculated as its standard deviation and coefficient of variation . There are currently several actions underway in supply chains to increase the degree of predictability of demand and, consequently, reduce its standard deviation, through access and sharing of sales information obtained in real time with consumers. Examples of this are rapid response programs between retailers and manufacturers, such as ECR (Efficient Consumer Response) and other Just-in-Time resupply initiatives.

The reflections of a greater variability in demand are immediately perceived in the levels of safety stock, a portion of stocks destined to guarantee the desired levels of product availability under conditions of uncertainty in demand and in the lead-time for resupply. These levels constitute a strategic product reserve, not providing direct support to the planning and execution of commercial, distribution or manufacturing operations at a given stage of the supply chain.

In addition to accessing information in real time as a way to reduce the calculated variability of demand at a given stage of the chain, there are three other ways to achieve this specific objective, having as a basic assumption the use of recent sales information as a triggering mechanism for sales activities. logistics and production that have a lead time relatively shorter than the response time required by the market:

• aggregating demand across different locations, implying a greater degree of centralization of inventories. By centralizing stocks, their movement towards the final destination is somewhat delayed, giving rise to postponement policies for physical distribution. Figure 1 illustrates these policies.

• aggregating demand across the product line, implying a greater proportion of semi-finished products in stock. Maintaining a larger share of semi-finished products implies delaying the final differentiation of products through simple value-adding operations such as assembly, packaging, painting, placement of accessories, etc. This decision gives rise to postponement policies in manufacturing. Figure 2 illustrates these operations.

• through a combination of the two previous measures.

Simply put, with regard to the postponement of distribution through physical centralization, it must be evaluated whether the reduction in safety stock levels will more than compensate for any increases in transportation expenses and in transit inventory levels, in addition to the real possibilities of eliminating expenses fixed with the operation and maintenance of facilities. On the other hand, with regard to the postponement of final production operations, it must be evaluated whether the reduction in finished product safety stock levels will compensate for any increases in the costs of executing final manufacturing operations and losses of economies of scale, since which can increase the share of variable expenses compared to fixed expenses. The following paragraphs comment on some critical points that must be observed when implementing postponement policies.

2. POSTAGE OF DISTRIBUTION

As mentioned, centralizing inventories can reduce safety stock levels and consequently the average inventory level across the entire system. Intuitively, it is expected that when demand in a given market region is above average, in another region demand may be below average, allowing items allocated to a given warehouse to be reallocated to others. The process of reallocating or transferring stocks between warehouses, however, may not be economically viable in a decentralized distribution system, considering the value of the product vis-à-vis unit transport costs. Unless the transfer costs are of a marginal nature, as there is already a flow of products between facilities, and taking into account that in a decentralized system products are normally of low added value, the economic viability of this operation is not justified.

Furthermore, the greater the coefficient of variation, the greater the benefits obtained with the physical centralization of inventories. This is because average inventory levels involve two components: one proportional to the expected average demand between two consecutive replenishments and the other proportional to the standard deviation or expected variability of demand over this time interval.

Finally, the benefits of physically centralizing inventories strongly depend on the behavior of demand in a given market region in relation to others. For example, the demands in any two markets are positively correlated if it is quite plausible that the demand, when increasing in a certain market, also increases in the other market. Similarly, when demand falls below the average in a given market, the same situation also occurs in the other. Intuitively, the benefit of centralizing inventories is proportionally smaller as the correlation between the demands of the two markets becomes increasingly positive.

Figure 3 graphically illustrates, for two distribution centers, situations in which demands are positively correlated, and there is no potential for significant reductions in safety stock levels with centralization, and situations in which demands are negatively correlated, with significant reductions in safety stocks based on offsetting up and down variations in demand at the two distribution centers.

Consider the logistics network shown in figure 1. Table 1 presents the demand verified over ten weeks for each of the two distribution centers, in addition to its mean, standard deviation and coefficient of variation. To find the total demand of the network, the demand of each distribution center must be added, noting that the average demand of the logistics network is the sum of the average demands of each distribution center, but that the standard deviation and the coefficient of variation of the network are not respectively the sum of the deviations and the coefficients of each distribution center.

The fact that the variability of demand when considering the logistics network as a whole (standard deviation of 9,5 units) is substantially smaller than the sum of the variability of demands allocated to each distribution center (9,5 + 9,1 = 18,6 ,48,9), indicates a strong potential for reducing safety stocks provided by centralization. Specifically in this example, there is a potential for a 1% reduction in safety stock levels (9,5 – 18,6 / XNUMX) . This potential can translate into substantial reductions in total logistics costs in circumstances where products are normally centralized: high added value, low turnover, high perishability/obsolescence, low weight or volume, etc.

On the other hand, the demand contained in table 2 for each of the distribution centers and for the logistics network, indicates that its variability (standard deviation of 12,5 units) is slightly lower than the sum of the variability of the two distribution centers ( 9,5 + 4,1 = 13,6), with potential for a reduction of just over 8% in safety stock levels. A reverse reading that can be made based on this result is that if the physical centralization of stocks would reduce safety stock levels by only 8%, its physical decentralization could also have reduced impacts in terms of increases in safety stock levels, depending on which market region would be used as a starting point for disaggregating demand.

This result may be interesting from the point of view of the total logistics cost in circumstances that favor the decentralization of inventories: low added value, potential to explore economies of scale in transportation, high turnover, small risk of perishability or obsolescence, considerable weight or volume.

DELAY OF MANUFACTURING

One of the most widespread concepts in inventory management is the fact that aggregate demand information is always more accurate than the disaggregated one. In this way, demand can be forecast more accurately for a country than for a city, for a complete line of products than for a single model, etc. Unfortunately, in the manufacturing planning and decision-making environment, aggregated forecasts are not easily or often employed, as the manufacturing manager must know exactly how much of each item is required before starting production. By adopting manufacturing postponement policies, it is possible to use more aggregated demand information in operations planning. For this, it becomes necessary to redesign the manufacturing process (and eventually the product design) in such a way that decisions about which specific products should be produced are delayed at least until the semi-finished products have been completed (see figure 2 ).

It is important to note that if some stages of manufacturing are postponed, the semi-finished products will have a lower added value than the finished ones, implying, therefore, lower opportunity costs of maintaining inventories. In some cases, taxes and duties to be collected are lower for semi-finished or unconfigured products than for finished products. By completing the manufacturing process at a distribution center close to where the products are consumed, it may be possible to reduce the applicable tax burden. However, it is not always possible to implement manufacturing postponement. Production lead times may be too long, or it may be necessary to take advantage of economies of scale in transportation and production, especially when products have low added value.

Figure 4 graphically illustrates a typical situation for three SKUs where their demands over time are negatively correlated, that is, when there is an increase in sales of one SKU, there is necessarily a decrease in at least one other. This situation is quite common in manufacturers of non-durable consumer goods, where, due to the growing number of SKUs launched, there is an effect known as “cannibalism” between products. In other words, the increase in sales of a given SKU occurs at the expense of the reduction in sales of another SKU produced by the same company, and not as a result of growth in market share. In these situations, the adoption of postponement policies for final manufacturing operations can significantly contribute to the reduction in finished product safety stock levels.

Analogously to the other two previous tables, Table 3 indicates that the demand variability for the basic product (standard deviation of 9,5 units) is smaller than the sum of the individual variability of the three types of SKUs (9,5 + 7,0 + 7,0 = 23,5). This potential for reductions in safety stock levels can translate into substantial reductions in total logistics costs in circumstances that favor the postponement of manufacturing: lack of economies of scale in final operations, short response time, higher proportion of relatively variable costs fixed costs, the operation adds substantial value to the product from the perspective of the end customer, etc.

CONCLUSION

This article presents two strategies emerging in several supply chains to manage the risk of holding inventories: delaying distribution and delaying manufacturing. The growing proliferation of products and markets served is leading several companies to rethink issues such as the location of inventories (a greater or lesser degree of centralization) and the chaining of production operations over time (production to stock or production against order). It is noticed that analyzes of the demand variability profile by product and by distribution center, as exemplified in this text, are extremely useful for making these decisions, since they indicate the potential for reducing safety stocks. Its adoption must occur concomitantly with the elaboration of a conceptual framework related to the characteristics of the operation, product and market in question.

BIBLIOGRAPHY

EVERS, PT, BEIER, FJ, 1998, “Operational Aspects of Inventory Consolidation Decision Making”, Journal of Business Logistics, Vol.19, N0.1, pp.173-189.

SIMCHI-LEVI, D., KAMINSKY, P., 2000, Designing and Managing the Supply Chain – Concepts, Strategies and Case Studies, 1st ed, New York, McGraw-Hill.

WANKE, P, 2000, “Logistic Positioning and the Definition of Customer Service Policy”, Tecnologística Magazine, June, Year 6, n. 5, pp.32-39.

ZINN, W., BOWERSOX, D., 1988, “Planning Physical Distribution with the Principle of Postponement”, Journal of Business Logistics, Vol.9, No.2, pp.117-136.

ZINN, W., LEVY, M., BOWERSOX, D., 1989, “Measuring the Effect of Inventory Centralization/Decentralization on Aggregate Safety Stock: 'The Square Root Law' Revisited”, Journal of Business Logistics, Vol.10, No. .1, pp.1-13.