Impact of Exponential Smoothing on Inventory Costs in Supply Chains

Abstract

It is common for firms to forecast stationary demand using simple exponential smoothing due to the ease of computation and understanding of the methodology. In this paper we show that the use of this methodology can be extremely costly in the context of inventory in a two-stage supply chain when the retailer faces AR(1) demand. We show that under the myopic order-up-to level policy, a retailer using exponential smoothing may have expected inventory-related costs more than ten times higher than when compared to using the optimal forecast. We demonstrate that when the AR(1) coefficient is less than the exponential smoothing parameter, the supplier’s expected inventory-related cost is less when the retailer uses optimal forecasting as opposed to exponential smoothing. We show there exists an additional set of cases where the sum of the expected inventory-related costs of the retailer and the supplier is less when the retailer uses optimal forecasting as opposed to exponential smoothing even though the supplier’s expected cost is higher. In this paper, we study the impact on the naive retailer, the sophisticated supplier, and the two-stage chain as a whole of the supplier sharing its forecasting expertise with the retailer. We provide explicit formulas for the supplier’s demand and the mean squared forecast errors for both players under various scenarios.