Structural time series models in inventory control

Abstract Exponential smoothing methods are often used to forecast demand in computerized inventory control systems. These methods, by themselves, are rather ad hoc, but they can be given a proper statistical foundation by setting up a class of structural time series models. The purpose of the paper is to highlight the potential role of these models in inventory control. In particular they are used as the basis for deriving formulae for estimating the mean and variance of the lead time demand distribution under both constant and stochastic lead time assumptions.