Value-at-Risk Analysis of Shipping Freight Indices with the Long Memory Volatility Process

• Main Authors: Chang, Chao-Chi, 張超琦
• Other Authors: Chou, Heng-Chih
• Format: Others
• Language: en_US
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/km4277

博士 === 國立臺灣海洋大學 === 航運管理學系 === 103 === This study aims to apply Value-at-Risk (VaR) models to evaluate the risk of dry bulk shipping freight indices, tanker shipping freight indices, and container shipping freight indices, when there is a long memory volatility process. Assuming that investors in the freight market can venture by holding not only left tail but also right tail risk under the distribution of profit/loss, this study compares the performance of the VaR models with the normal, Student-t and skewed Student-t distributions for both left tail and right tail based on the chosen models. The left tail of the distribution of profit/loss represents the maximum potential loss when the freight indices slump and the shipowners will face the great risk. On the other side, the right tail of the distribution of profit/loss represents the maximum potential loss when the freight indices skyrocket and the shippers will face the great risk. The underestimation or overestimation of VaR will make the shippers or shipowners get huge loss. For the dry bulk shipping industry, the densities of the BDI, BPI, BCI, BHSI, and BSI exhibit the asymmetric long memory property of volatility and the fat-tail phenomenon. Besides, the asymmetric FIAPARCH model performs better for the BDI, BCI, BHSI and BSI, whereas the HYGARCH model performs better for the BPI. For the backtesting results, the VaR models with the skewed Student-t distributed innovation are preferred for both left tail and right tail. For the tanker shipping industry, the densities of the BDTI and BCTI also exhibit the asymmetric long memory property of volatility and the fat-tail phenomenon. In addition, the FIEGARCH model performs better for the BCTI, whereas the HYGARCH model performs better for the BDTI. For the backtesting results, the VaR models with the skewed Student-t distributed innovation are preferred for both left tail and right tail. For the container shipping industry, the densities of the HRCI and CCFI exhibit the long memory property of volatility and the fat-tail phenomenon. Besides, the FIAPARCH model performs better for the HRCI, whereas the HYGARCH model performs better for the CCFI. For the backtesting results, for the HRCI, the skewed Student-t VaR models perform correctly in all of the cases for both left tail and right tail. For the CCFI, the Student-t VaR models perform correctly in all of the cases for both right tail and left tail. The above results suggest that precise VaR estimates may be acquired from a long memory volatility structure with the Student-t and skewed Student-t distributions. Such models improve the long-term volatility forecast and provide more precise pricing of freight contracts. These findings provide a more accurate estimation of VaR for shipping freight indices and could be applied in trading FFAs and dealing with portfolios of freight derivatives, which includes freight options.