| One important supply chain research problem that has drawn much attention is the phenomenon known as bullwhip effect. In this thesis, at first, we study the bullwhip effect under order-up-to policy. We compare the bullwhip ratio under three different forecasting methods and demonstrate that the optimal forecast scheme has advantages over EWMA and moving average method. We also summarize the order process behavior when different forecasting methods are employed. Secondly, we study the bullwhip effect and inventory variation under direct-forecast ordering policy. We find that bullwhip effect does not exist as opposed to the other researcher's work when an optimal forecast scheme is used. We use two different forecast methods---one is MSE optimal, the other is exponential smoothing to study the inventory variation. We find inventory variations under both forecasting techniques are non-stationary, the variation under optimal forecasting is less than variation under non-optimal forecasting after certain time periods. The results here indicate if we adopt optimal forecasting, we can both decrease bullwhip effect and inventory variation in the long run.;One way to reduce the bullwhip effect is to adopt direct channel, so thirdly we propose a dynamic system to study the channel design in supply chains. For the constant market size, we provide the analytical solution and present some strategic suggestions for whether to open a direct channel or not. For inconstant market size, we use simulation to study the channel redistribution. Our findings indicate no matter the market size is deterministic, white noise, or non-stationary, the ratio of demand on direct channel to the demand on indirect channel is same when the redistribution is stable. So the results of demand redistribution when market size is constant are pretty robust, it really provides managerial implication for the channel design issue. |
