Applications of Mean-VaR Models to Optimizing Oil Cracking Portfolio

• Main Author: 鄭互維
• Other Authors: 許和鈞
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/56964530938647830440

碩士 === 國立交通大學 === 經營管理研究所 === 95 === The purpose of this research is to study how to optimize a product portfolio under a price volatility to maintain a stable operating revenue by applying Mean-VaR models. The practical analysis is to discuss how a refinery with product portfolios composed of gasoline, heating oil, and residual fuel, uses historical simulation method, t-distribution method, and EWMA method, to estimate VaR of product portfolios to bring out the Mean-VaR models and efficient frontiers, and then to find out the most efficient product portfolio by the efficiency index modified by Campbell、Huisman, and Koedijk(2001).In the comparison among models, this study proceeds the backtests for past 250 periods by the method developed by Christofferson(1998). The results of backtest appear that EWMA method is the best model among the three ones in this study. Besides, the results also show that there are significant differences about return, weight of product portfolio, efficiency, among the three models. Besides, this study also discuss effects upon optimal product portfolios under different designated risk conditions. The results reveal that higher VaR, lower confidence level, longer holding time interval will enhance the expected return because of the higher risk, and meanwhile, the weight of higher risk product will rise, and vice versa.