Abstract

This paper examines the assertion that partial pooling of customers is sometimes favored over complete pooling on the sole basis of demand correlation. First, the management of inventory within a supply chain is discussed, with specific attention paid to risk-pooling. Then, the claim that partial pooling can dominate is theoretically discussed. The conditions under which previous research found that partial aggregation could, at times, be preferable are investigated next. From this, methods are proposed for checking correlation matrices to ensure their validity. It is concluded that partial pooling can never do better and that examples supporting partial aggregation are based on inconsistent correlation matrices.

Scope and purpose

Arising in part from uncertainty, inventory is a key component of the supply chain. This paper reviews alternative methods for managing demand uncertainty across various levels of the supply chain as well as within a single echelon of the system. Particular attention is paid to the fulfillment of customer demands from multiple stock-keeping facilities versus from a single location, with the superiority of the latter presented in terms of requisite safety stock levels. In doing so, methods for ensuring the feasibility of correlation matrices for simulation and numerical analyses are also highlighted.

Author Keywords: Single echelon inventories; Decentralization of facilities; Customer grouping; Correlation matrices

Article Outline

1. Introduction

2. Statistical economies of scale

3. Partial pooling is never better

4. Results based on inconsistent correlation matrices

4.1. The correlation structure for three random variables

4.2. The presence of more than three random variables

5. Conclusion

Acknowledgements

Appendix A. The structure of a valid correlation matrix

References

Vitae