Article ID: | iaor20116469 |
Volume: | 37 |
Issue: | 2 |
Start Page Number: | 300 |
End Page Number: | 311 |
Publication Date: | Apr 2009 |
Journal: | Omega |
Authors: | Meredith Jack, Akinc Umit |
Keywords: | programming: dynamic, programming: probabilistic, inventory |
The make‐to‐forecast manufacturing environment is characterized by the manufacture of individual, expensive‐to‐store units which come in alternative configurations or models. Since the production lead times are longer than customers are willing to wait, production is started in anticipation of orders for particular models and then modified while in production, if economically desirable, to match the actual customer orders. Ideally, no finished goods inventory occurs in this process. Occasionally, however, such unsold finished items (called ‘orphans’ here) are generated. Since their holding cost is substantial, especially if they are physically large and hard to store, the orphans become a serious managerial problem. The dilemma management faces with an orphan is: should the orphan be modified (to the specifications of a new order) to dispose of it now, perhaps even at a loss, or should it be held another period (thereby incurring the substantial holding cost) for a possibly better matching order with a more tolerable modification cost? To meet this management challenge we formulate the dilemma as a stochastic dynamic programming problem, similar to the well‐known optimal stopping problem. We obtain optimal policies and analyze their properties for the typical case where there is only one orphan. Last, we extend the approach to the case of multiple orphans for those rare instances when there may be more than one orphan.