Optimal policy structure characterization for a two-period dual-sourcing inventory control model with forecast updating Ali Cheaitou a,n , Christian van Delft b,1 , Zied Jemai c,2 , Yves Dallery c a Industrial Engineering and Engineering Management Department, College of Engineering, University of Sharjah, P. O. Box 27272, Sharjah, United Arab Emirates b HEC School of Management, Paris (GREGHEC), 1 Rue de la Libération, 78351 Jouys-en-Josas Cedex, France c Ecole Centrale Paris, Laboratoire Genie Industriel, Grande Voie des Vignes, 92295 Châtenay-Malabry Cedex, France article info Article history: Received 1 October 2012 Accepted 25 July 2014 Available online 22 August 2014 Keywords: Inventory control Forecast updating Dual supply Short life-cycle products Optimal solution abstract A proposed single-product, stochastic, two-period inventory control model combines demand forecast updating with the flexibility of two supply sources. Demand is modeled by two independent, random variables over a two-period selling season. At the beginning of the first period, two quantities are ordered using two different supply options: a local supplier who delivers the ordered quantity immediately and a second supplier who delivers the ordered quantity at the beginning of the second period. The local supplier charges a higher cost per ordered unit. The model considers an initial inventory, so the decision maker has an opportunity at the beginning of the first period to return part of the available inventory to the supplier (or sell it in a parallel market). At the end of the first period, any unsatisfied demand is backlogged to be satisfied in the next period. At the beginning of the second period, exogenous market information updates the second-period demand forecast. According to this updated forecast and the actual inventory level, an additional quantity is ordered using the local procurement source or another quantity is returned to the supplier (or sold in a parallel market). With a dynamic programming approach, this research exhibits the structure of the optimal policy, characterized by order-up-to and salvage-up-to levels. The findings provide the structure of the second-period conditional optimal policy and analytical insights that characterize the first-period optimal policy. Furthermore, a numerical study reveals the impact of information quality on the optimal policy and the trade-off between the two procurement options. & 2014 Elsevier B.V. All rights reserved. 1. Introduction We study short life cycle products for which demand occurs during a short selling season. The basic inventory management model that fits this type of situation is the well-known News- vendor model in which a decision maker orders a single quantity to fulfill uncertain single-period demand. At the end of the period, unsatisfied demands are lost (maybe inducing a penalty shortage cost), and remaining inventory is salvaged at a salvage value. Vast research considers the Newsvendor models (for a review, see Khouja, 1999) and proposes extensions to improve this model. A key extension is the two-period setting, which allows decision makers to react to the level of demand. Understocking and/or overstocking is costly, so in a stochastic setting, decision makers must use forecasts of future demand to minimize such costs. In some cases, it is possible to improve the quality of these forecasts during the decision process, such as by using information technology- enabled sales or market data. Between successive decisions, new information provides updates to the demand forecast for the remaining of the planning horizon. This new information may take two forms: endogenous, such as actual demand for the same product in previous periods, or exogenous, such as sales of a preseason product. We focus on the effective use of exogenous information. Tan et al. (2007) suggest ways to improve demand forecasts using exogenous information. For example, if a company uses sales representatives to market its products, the information collected from the distribution of sales vouchers or quotations, which suggest that customers are interested in buying, can update demand fore- casts. In Internet retailing, the number of visits to a commercial website indicates the level of interest in some of the offered products; a precise visit to a certain subpage could indicate the interest of the buyer by a specific product. Furthermore, the company can ask online visitors to complete ”wish lists” that indicate products Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/ijpe Int. J. Production Economics http://dx.doi.org/10.1016/j.ijpe.2014.07.028 0925-5273/& 2014 Elsevier B.V. All rights reserved. n Corresponding author. Tel.: þ971 6 505 3921; fax: þ971 6 505 3963. E-mail addresses: acheaitou@sharjah.ac.ae (A. Cheaitou), vandelft@hec.fr (C. van Delft), zied.jemai@ecp.fr (Z. Jemai), yves.dallery@ecp.fr (Y. Dallery). 1 Tel.: þ33 139 677 305; fax: þ33 139 679 415. 2 Tel.: þ33 141 131 285; fax: þ33 141 131 272. Int. J. Production Economics 157 (2014) 238–249