Delegation of Stocking Decisions under Asymmetric Demand Information
MetadataShow full item record
AbstractProblem Definition: Headquarters of a retailer delegates stocking decisions to store managers in its various stores. Store managers have complete information of the local demand process, whereas headquarters has partial information. The problem is how to incentivize the managers to make stocking decisions that minimizes headquarters’ expected overage and underage costs. Academic/Practical Relevance: We investigate performance measurement schemes in which headquarters incite store managers make stocking decisions in headquarters’ best interests using their private information. Adopting such schemes helps retailers reduce stock-outs and attain desired service levels across their stores. Methodology: Headquarters knows that the underlying demand process at a store for a product over a replenishment cycle is one of N possible demand processes, whereas store manager knows the specific process with certainty. Store manager creates a single order before the cycle. Headquarters use an incentive scheme which is based on the leftover inventory at the end of the cycle and whether there is a stock-out at a pre-specified inspection time before the end of the cycle. Headquarters’ problem to determine the inspection time and importance given to a stock-out occasion relative to leftover stock is formulated as a constrained convex optimization problem. Results: The proposed “early inspection” scheme leads to perfect alignment under certain conditions. Under more general conditions, it provides a near-perfect alignment and performs strictly better than stock-out inspection at the end. Using historical sales data of a retailer, we show that this scheme can lead to considerable cost reduction. Managerial Implications: Even though attaining high on-shelf availability is crucial in retail, stock-out related measures are not reflected in store managers’ performance scorecards. We prescribe a novel, easy and practical method for headquarters with which they can increase on-shelf-availability by relying on their store managers’ private demand information. The proposed method decidedly outperforms the computer aided ordering systems that are commonly used in practice.
Working paper. Please do not cite without the consent of authors.