An Optimal Two-stage Forecasting and Ordering Contract Model using Neural Networks and Options Pricing

• Main Authors: Teddy Chih-Shuo Chuang, 莊智碩
• Other Authors: Chih-Peng Chu
• Format: Others
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/36785727828155934346

碩士 === 國立東華大學 === 運籌管理研究所 === 101 === In the supply chain, demand uncertainty is a constant key challenge for retailers when placing product orders to suppliers because uncertain demands may generate serious mismatch between supply and demand, and, as a result, jeopardize the business and profit. If a retailer cannot attain right demand forecasting and hence order a quantity smaller than the actual market demand, he will endure the risk of opportunity and profit loss due to product shortage. On the contrary, if the retailer orders an excessive quantity, he will face resource wastes in both capital and inventory. A company able to minimize the threat of unpredictable demand change by performing good forecasting to catch the right demand trend and order the right quantity will take a crucial step towards success. In pursuit of desirable supply chain management, catching the right demand trend in the rapidly changing world has become an increasingly important issue for business runners. Realizing the fact that demand forecasting will turn out more accurate results when it is performed near the selling season, we propose, in this research, a two-stage forecasting and ordering contract model based on artificial neural networks and financial options to improve the demand forecasting accuracy and order flexibility. Under the new model, a retailer can place an initial order to the supply based on the result of the first demand forecasting by using neural networks. When placing the initial order, the retailer will meanwhile purchase a fixed amount of call and put options according to the calculated prediction interval, to ensure order flexibility. The retailer then carries out a second demand forecasting in the lead time (between the initial order and product acquirement) and, based on the newly updated forecasting result (which will get closer to the actual demand), decides whether to adjust the order quantity -- by exercising the purchased options -- to suit the practical need. We have tested and verified the performance of the proposed model by extensive actual data and have obtained quite favorable evaluation results. To begin with, our new model can substantially increase the forecasting accuracy to reduce the prediction mismatch between supply and demand, and to decide on an appropriate amount of options. It also provide the retailer with a good chance to change, increase or decrease, the order quantity according to updated forecasting results and purchased options. The practice of the new model will benefit the supplier as well because it enables the supplier to manufacture products and manage inventory more flexibly and efficiently. To sum up, our two-stage forecasting and ordering contract model can benefit both retailers and suppliers in the two-echelon supply chain.