A model to study the relationship between forecast error, inventory investment, and customer service level in a multiperiod, probabilistic environment

With the advent of Just-In-Time manufacturing strategies, reduction of inventory costs have once again become the focus of all attention. However, efforts to reduce inventory while continuing to live with poor forecasts and unduly high service level requirements, is likely to be futile. This work uses a dynamic programming approach to establish trade-off curves, tying in forecast error, customer service level, and inventory investment. This work applies to realistic, dynamic settings wherein there is uncertainty associated with demand, and lead time could be fixed or probabilistic. This work further assumes form-free probability distributions and thus avoids errors introduced into the analysis from estimating parameters of the distribution from limited data.