Supply chain games: modeling in an intertemporal framework with periodic review

In this chapter we consider firms which employ a periodic review policy and handle products characterized by a relatively short life-cycle, so that a single inventory update at a predetermined point on the production horizon (i.e., two-period approach) may suffice to identify demand over the remaining part of the production horizon. Fisher et al. (2000) reported an example of this approach in the apparel industry where highly accurate demand forecasts were made after observing only 11% of demand. Fisher and Raman (1996) reported further examples in which very accurate forecasts were obtained after observing 20% of demand.

The point of update is assumed to be chosen from previous experience and may generally involve different considerations, such as the change from high- to off-season, expected customer fatigue towards the product, or the impact of competitors catching up with production of a similar product. As a result, demand realization determined for the second part of the production horizon may be different from that observed for the first part. In contrast to the previous chapter, where customer demand along with inventories was observed either continuously or only by the end of production horizon, we assume here that an update is possible before the end of the horizon and that the probability distribution of demand for the second part of the production horizon depends on demand realization over the first part of the production horizon (see also Kogan et al. 2004, 2007; Kogan and Herbon 2007).