A copula approach to modelling dependence structures between Asian equity markets.

• Main Authors: Chin, Zhuo Song., Teo, Chin Seng., Tham, Eugene Poh Keong.
• Other Authors: Li Ka Ki Jackie
• Format: Final Year Project
• Language: English
• Published: 2011
• Subjects: DRNTU::Business::Finance::Portfolio management
• Online Access: http://hdl.handle.net/10356/46357
• Institution: Nanyang Technological University

In recent times, increased dependence between markets and asset classes has rendered traditional techniques such as linear correlation incapable of adequately modelling dependence structures. This has led to growing interest amongst academics and portfolio managers in new and more effective tools to analyse inter-market and inter-asset dependences. Copulas in financial applications have emerged as one of the more promising methods to measuring dependences. By allowing flexibility to model marginal distributions and dependence structures separately, copulas enable practitioners to more comprehensively understand and describe the complex dependence dynamics between markets. More recently, the use of the Normal copula to price Collateralized Debt Obligations (CDOs) was blamed for global financial crisis in 2008. The focus has since shifted to examining copula families which can better model tail dependence. This research paper seeks to enhance the understanding of cross border dependence between equity markets through the use of copulas. First, we illustrate how copulas can better capture dependence structures between the Hong Kong Hang Seng Index and the Singapore Straits Times Index. Second, the properties of different copulas are compared, and the most appropriate copula is selected to model the different dependence structures and tail dependences in different time periods. Third, we show how the best fitting copula varies across the different time periods, tail threshold levels, and time horizon. We provide persuasive evidence that the risk appetite of portfolio managers, which affects the width of tail selection, will have important implications on the copula selection, ultimately leading to mis-estimation of dependence risks.