A Two-Echelon Inventory Model with Lost Sales

Almost all multi-echelon inventory models assume that demand not satis ed immediately can be backordered. In some situations this assumption is not realistic. For example, it may be more representative to model stockouts as lost sales when the retailers are in a competitive market and customers can easily turn to another rm when purchasing the good. Assuming lost sales at the retailers, we consider a one warehouse several retailers inventory system. Using the well-known METRIC-approximation as a framework, we present a heuristic for nding cost e ective base-stock policies. In a numerical study we nd that the cost of the policies suggested by the heuristic is on average 0.40% above the cost of the (S 1; S)-optimal policy.