Double-index VaR model and skewed distribution of indices

• Main Authors: Chiam, Yee Hong, Yos, Virin, Zhou, Yuan
• Other Authors: Low Chan Kee
• Format: Final Year Project
• Language: English
• Published: 2009
• Subjects: DRNTU::Business::Finance::Risk management
• Online Access: http://hdl.handle.net/10356/15156
• Institution: Nanyang Technological University

Value at Risk (VaR) is widely used in many financial institutions to measure portfolio risk. In our project, we examine if the single-index model under RM methodology that assumes normally distributed returns can be improved on. We try using—1) a skew-normal or skew-t distribution for the index returns instead of the normal assumption; 2) GARCH(1,1) to model volatility of the indices; and 3) a double-index model using two indices. We find that skew t distribution outperforms the normal and skew normal distribution in VaR estimates. The skew normal does not necessarily give better VaR estimates than the normal distribution, except for the double-index case. GARCH outperforms RM for in-sample data, but not for out-sample predictions. Finally, the double-index model performs better than the single-index model under skew normal assumptions, but not under normal assumptions.