Measuring and implementing the bullwhip effect under a generalized demand process

Abstract

The measure of the bullwhip effect, a phenomenon in which demand variability increases as one moves up the supply chain, is a major issue in Supply Chain Management. Although it is simply defined (it is the ratio of the unconditional variance of the order process to that of the demand process), explicit formulas are difficult to obtain. In this paper we investigate the theoretical and practical issues of Zhang [Manufacturing and Services Operations Management 6-2 (2004b) 195] with the purpose of quantifying the bullwhip effect. Considering a two-stage supply chain, the bullwhip effect is measured for an ARMA(p,q) demand process admitting an infinite moving average representation. As particular cases of this time series model, the AR(p), MA(q), ARMA(1,1), AR(1) and AR(2) are discussed. For some of them, explicit formulas are obtained. We show that for certain types of demand processes, the use of the optimal forecasting procedure that minimizes the mean squared forecasting error leads to significant reduction in the safety stock level. This highlights the potential economic benefits resulting from the use of this time series analysis. Finally, an R function called SCperf is programmed to calculate the bullwhip effect and other supply chain performance variables. It leads to a simple but powerful tool which could benefit both managers and researchers.