Two essays on portfolio management under non-normality

This dissertation focuses on portfolio management issues under non-normality. The first essay, "Multivariate GARCH-modeling and Mean-VaR Analysis under Non-normality: an Empirical Investigation of Emerging Stock Markets", examines the impact of non-normal returns on correlation modeling using AG-DCC model and optimal asset allocation under mean-VaR framework. We assume that the innovation terms follow normal, student and skew-student distributions. We find that AG-DCC model based on skew-student distribution provides better out-of-sample correlation forecast. Using estimated covariance matrices from the three distributional assumptions, we obtain optimal portfolio weights with and without transaction costs within the mean-VaR allocation framework. We find that in both cases AG-DCC model based on Student distribution outperforms the other two models at all significance levels in terms of failure rates. At lower VaR level (0.01 and 0.025), AG-DCC model based on Student distribution provides more conservative allocation than the other two specifications; at higher VaR level, AG-DCC model based on Student distribution produces a failure rate that is very close to the specified VaR level.;The second essay, "Integrated Risk Management of a Corporate Bond Portfolio" measures the risk of a corporate bond portfolio by taking into account the non-normality of the loss distribution. We build the loss distribution of the bond portfolio by Monte Carlo simulation where market risk and credit risk are integrated through copula functions. Risk measures, such as VaR and Expected Shortfall as well as Economic Capital, are calculated for simulated portfolio loss. We find the integrated expected loss is highest for Clayton copula. Clayton copula produces the lowest downside risk in terms of VaR, Expected Shortfall and Economic Capital. We also find that in all three cases, credit risk clearly dominates market risk, and that market risk has a negative contribution to the integrated portfolio risk. These results hold for both investment grade and non-investment grade groups. Integrated risk management also generates diversification benefits in terms of reduction in Economic Capital needed to back up portfolio risk and the reduction is most significant for non-investment grade group.