In this thesis I examine both the in-sample and out-of-sample performance of mean-variance (MV) strategies and mean-lower partial moment (MLPM) strategies using three different models of expected returns within developed and emerging stock markets framework between April 1991 and October 1998. My results show that the MLPM portfolios outperform the MV portfolios. It may imply that traditional MV framework is no longer adequate to characterize investment decisions in emerging stock markets. It is because the emerging market returns cannot be characterized by a normal distribution but have significant skewness and kurtosis. The downside risk approach portfolio strategies that use conditioning information to predict market returns produce impressive out-of-sample performance over the 1991-...[ Read more ]

In this thesis I examine both the in-sample and out-of-sample performance of mean-variance (MV) strategies and mean-lower partial moment (MLPM) strategies using three different models of expected returns within developed and emerging stock markets framework between April 1991 and October 1998. My results show that the MLPM portfolios outperform the MV portfolios. It may imply that traditional MV framework is no longer adequate to characterize investment decisions in emerging stock markets. It is because the emerging market returns cannot be characterized by a normal distribution but have significant skewness and kurtosis. The downside risk approach portfolio strategies that use conditioning information to predict market returns produce impressive out-of-sample performance over the 1991-1998 period.