An Econometric Investigation of the Bullwhip Effect---The Influence of Demand and Supply in the Automobile Industry

The issues relating to the bullwhip effect have been studied rigorously in past, even before the advent of supply chain management concepts. The bullwhip phenomenon refers to the increasing amplification of demand variation as one moves upstream in the supply chain. A variety of causes and resulting unfavorable performance effects in supply chains have been investigated, using wide-ranging methodologies. Many beneficial solutions have also been provided to counter bullwhip effects and improve overall supply chain performance. However, despite considerable research efforts in the past, the number of empirical studies conducted to date is rather limited. This study, consisting of three essays, is aimed at contributing to this body of literature. It is based on secondary data obtained from the U.S. automotive industry to test and investigate bullwhip effect-related issues.;The automobile industry is considered due to availability of data, its important role in the economy, and dynamic business environment. The first essay is aimed at investigating individually: (1) the existence of bullwhip effect, (2) the impact of inaccurate demand process specification on bullwhip effect, (3) the effects of the demand forecasting policies on bullwhip effect, and, (4) demand-supply equilibrium at the product level, in selected U.S. automobile industry supply chain. It is concluded that, first, the bullwhip effects were present in the automobile supply chain. Second, it is found that mis-specified demand series may cause poorer forecast accuracies and stronger bullwhip effects. Third, the choice of forecasting policy is found to have an impact on forecast accuracy, and the use of relatively more complex forecasting techniques such as Holt-Winters and time series method (compared to moving average and simple exponential smoothing methods) leads to less pronounced bullwhip effects. Finally, pair-wise correlations among demand, production and inventory were significant, but the equilibrium status was not found to be present. This essay contributes to the field in the following ways. First, the results, confirming the existence of bullwhip effect in specific manufacturing supply chain, differs from the conclusions of studies such as Cachon et al. (2007) which was an aggregate industry-level study. Second, in contrast of the conclusions of studies such as Hosoda and Disney (2009), it was found that demand process misspecification has impacts on the bullwhip effect and the accuracy of demand forecasting.;The second essay is aimed at investigating the relative forecasting accuracies of univariate and multivariate time series approaches to further examine whether multivariate time series methods are more effective than the simpler, univariate methods. In addition, this essay is aimed at evaluating causal relationships among demand, supply and inventory variables. This essay contributes to the supply chain literature in the following aspects. First, multivariate time series model is introduced into the supply chain management field for the first time to investigate the forecast accuracy and test the relationship among the considered variables. The choice of univariate and multivariate forecasting methods was a major research target in the forecasting field. Second, although a few prior studies used the multivariate approach to describe the demand process, this study is the first to present the numerical results of the model specification and conduct forecasting based on the model. Third, this study is the first to investigate the causal relationships among demand, supply, and inventory using multivariate time series method. Finally, this study provides another effective and theoretically-robust tool to test the existence of equilibrium conditions in the supply chain.;The major objective of the third essay is to describe the transition of different types of volatility in the supply chain. This essay contributes to the literature in the several aspects. First, a multivariate generalized autoregressive heteroscedasticity model is used for the first time to empirically establish the existence of the volatility transition in the supply chain. Second, this study is also the first to consider the influence of a balanced demand-supply system where the demand, supply, and inventory are all stationary processes. The major contribution for this finding is that it is necessary to explicitly consider the effect of volatility when developing policies for supply chain inventory management in the presence of bullwhip effects. (Abstract shortened by UMI.)...