Demand patterns for products with seasonality and or short life-cycles do not follow a clear discernible pattern (to allow predictive time-series modeling of demand) for individual sales events or seasons due to such factors as considerable demand volatility, product promotions, and unforeseen marketplace events. Suppliers supporting such atypical demand patterns typically incur higher holding costs, lower capacity utilization, and lower order fill-rates, particularly under long lead-times and uncertainty in effective capacity. Retailers on the other hand struggle with product overages and supply shortages. On the other hand, atypical demand settings bring huge financial opportunity to supply chain players, and are pervasive. It is suggested in the literature that an effective means to reap these benefits is through increased information sharing between retailers and suppliers, superior forecasting with forecast update techniques, proper replenishment, and custom designed inventory/manufacturing policies. We also believe that sharing of order forecasts, also known as soft-orders, in advance by the buyer could be beneficial to both parties involved. This dissertation in particular studies a two-player supply chain, facing atypical demand. Among the two-players is a buyer (retailer/distributor/vendor) that makes ordering decision(s) in the presence of upstream supply uncertainty and demand forecast revision(s). We propose a stochastic dynamic programming model to optimally deicide on soft-order(s) and a final firm-order under a deposit scheme for initial soft-order(s). While sharing of upstream soft-order inventory position information by the supplier before receiving a final order is not a common industrial practice, nor is it discussed in the literature, our analysis shows that such information sharing is beneficial under certain conditions. Second player of the supply chain is a supplier (manufacturer) that makes production release decision(s) in the presence of limited and random effective capacity, and final order uncertainty. Our stochastic dynamic programming model for optimal production release decision making reveals that substantial savings in order fulfillment cost (that includes holding, overage, and underage costs) can be realized in the presence of advance soft-order(s). Soft-orders can also be shown to improve order fill-rate for the buyer. This research explores complex interactions of factors that affect the operational decision making process, such as costs, demand uncertainty, supply uncertainty, effective capacity severity, information accuracy, information volatility, intentional manipulation of information etc. Through extensive analysis of the operational policies, we provide managerial insights, many of which are intuitively appealing, such as, additional information never increases cost of an optimal decision; many are also counterintuitive, for example, dynamic programming models cannot fully compensate for intentional soft-order inflation by the buyer, even under conditions of a stable and linear order inflation pattern, in the absence of deposits.
