Building Efficient Frontier for Non-normal Asset Returns using Copulas

The aim of this work is to provide a comprehensive empirical study of the evolution of dependence structure and tail dependence for a large set of developed and emerging markets. Three methodological contributions are seriously developed. First and based on well documented literature on empirical regularities, a novel Mean-CVaR-Copula approach is introduced to allow for non linearity dependence structure and deviation from normality hypothesis. In order to do so, an alternative measure of diversification is involved to classical linear correlation model, which is suitable under normality hypothesis. Indeed, the joint distribution is modeled using Copulas functions to capt non linear dependence across market. Second, an extension of Kapil Agrawal¿s approach (2008) is proposed to provide more reliable results based on Copula theory. Third, a Copula theory methodology is developed to link long memory parameter (long range time dependence parameter) and Copula parameter (dependence parameter) in order to control financial crisis persistence and reduce their effect on assets allocation.