An analysis of dynamic postponement in a multimodel supply chain

In this research, we consider a two-echelon supply chain anchored by a single supplier of a standardized, short-life-cycle product. Given our assumptions regarding demand independence across the markets served by this supplier, no lateral transshipments across those markets, and no capacity restrictions on the amount shipped in any period, the problem becomes one of fulfilling the demand of each customer independently. We therefore address the problem of shipping goods from a single supplier to a single customer. In general, there are n transit modes having different lead times and costs available to supply the customer; a transit mode with a shorter lead time is assumed to command a higher cost. Demand is assumed to be uncertain and non-stationary over a finite product life cycle. Our premise is that the strategic, pre-planned use of faster modes of transit can allow a firm to delay the shipment quantity decision in order to collect information to better estimate demand. In this research, we consider the case in which n is fixed at 2. We determine the conditions under which this dual-mode strategy has value and develop a dynamic programming methodology for determining the optimal mix of transit modes under three different scenarios describing different demand patterns and types of demand-related information. In the three modeling scenarios, we find that certain types of demand information lead to greater levels of logistics postponement and significant performance improvement over a single-mode system. Further, our results indicate that the type of demand information has a significant impact on whether the performance improvements of a dual-mode system persist as the length of the planning horizon increases. Finally, we find that the availability of a faster mode of transit---whether or not the faster mode is even used---always results in better supply chain performance.