Bullwhip Effect Of Supply Chain Research

An important observation in supply chain management, known as the bullwhip effect, can lead to tremendous inefficiencies: excessive inventory investment, poor customer service, lost revenues; misguided capacity plans, ineffective transportation, and missed production schedules. So the bullwhip effect research is very important in supply chain management theory and operation.In this paper, the supply chain bullwhip effect is investigated and the associated extensive work is conducted. Based on the bullwhip effect theory model of simple two level supply chain with AR(1) stationary demand process presented by Prof. H.L.Lee, an extensive work is performed systematically with respects to the influences on different stationary demand process, and different information forecasting and information sharing on two or multi level supply chain bullwhip effect, which mainly includes: (1) to establish the two or multi level supply chain bullwhip effect theory model without information sharing on the basis of AR(1) and ARMA(1,1) stationary demand process, moving average forecasting method (MA) /exponential weighted moving average method (EWMA)/mean square error-optimal forecasting method (MSE-optimal), and order-up-to inventory policy; (2) to establish the two or multi level supply chain bullwhip effect theory model with information sharing on the basis of AR(1) and ARMA(1,1) stationary demand process, MA/EWMA/MSE-optimal forecasting scheme , and order-up-to inventory policy.Meanwhile, the corresponding models related to the mentioned bullwhip effect theory model of two or multi level supply chain are simulated by employing the EXTEND system simulation software and compared with the theoretical results. Consequently, the sensitivity of affect factors for supply chain bullwhip effect, such as lead time L , observation length p, correlation coefficient p, are discussed by orthogonal tests with variance analysis method, which provides some references and guides for the decrease of the supply chain bullwhip effect.Furthermore, the supply chain bullwhip effect based on AR(1) and ARMA(1,1) stationary process, MA and MSE-optimal forecasting scheme are compared, with or without information sharing, which can be regarded as the rules for employing MA or MSE-optimal forecasting scheme and sharing information.Finally, the demand wave phenomenon of container shipping market is investigated by adopting the supply chain bullwhip effect concept.The following conclusions of this paper can be drawn as:(1) For the supply chain with AR(1) stationary process demand, MA forecasting method and order-up-to inventory policy, the bullwhip effect always exists with the value increasing along with the increase of L and decrease of p. Regarding p, the bullwhip value decreases along with the increase ofρwhen 0≤ρ≤1; while -1≤ρ≤0, it has two cases: provided p is even, the bullwhip value increases along with the increase ofρ, and for the odd, it is just reversed. The above conclusion is also suitable for ARMA(1,1) stationary process demand.(2) For the supply chain with AR(1) stationary process demand, EWMA forecasting method and order-up-to inventory policy, the bullwhip effect always exists with the value increasing along with the increase of L, the decrease of p and the increase of smoothing constant a.(3) For the supply chain with AR(1) stationary process demand, MSE-optimal forecasting method and order-up-to inventory policy, the bullwhip effect only exists when 0≤ρ≤1 and the value increases along with the increasing L. As forρ, it has a critical valueρ~*. The bullwhip value increases along with the increasingρin case ofρ<ρ~* and decreases forρ>ρ~*. However, for the ARMA(1,1) stationary process, the bullwhip effect exists only ifρ>θ, with the influence ofρsimilar to that of the AR(1) stationary process demand.(4) For the supply chain with AR(1) or ARMA(1,1) stationary process demand, MA forecasting method and order-up-to inventory policy, the ordering process is a ARMA(1 ,p) stationary process; for that with AR(1) stationary process demand, EWMA forecasting method and order-up-to inventory policy, a ARMA(1,(?)) one; and for that with AR(1) or ARMA(1,1) stationary process demand, MSE-optimal forecasting method and order-up-to inventory policy, a ARMA(1,1) one.(5) When MA forecasting method is employed, sharing demand information can remarkably reduce the supply chain bullwhip effect, which is independent on AR(1) or ARMA(1,1) stationary process demand.(6) For AR(1) stationary process demand and MSE-optimal forecasting method, sharing demand information can reduce the supply chain bullwhip only when 0≤ρ≤1. While for the ARMA(1,1), there exists a critical valueρ~* that leads to the decrease of the supply chain bullwhip effect whenρ>ρ~*.(7) Whatever with or without information sharing, there is a critical value L~* for AR(1) stationary process demand. When L< L~*, MA forecasting method is inclined to decrease the supply chain bullwhip effect better than MSE-optimal method; while for L> L~*, MSE-optimal forecasting method is better than the MA one in the decrease of supply chain bullwhip effect. Similarly to ARMA(1,1) stationary process demand.