Portfolio Optimization in a Downside Risk Framework
Author | : Lars Huelin |
Publisher | : LAP Lambert Academic Publishing |
Total Pages | : 136 |
Release | : 2011-04 |
ISBN-10 | : 3844301577 |
ISBN-13 | : 9783844301571 |
Rating | : 4/5 (77 Downloads) |
Book excerpt: The present study examines how downside risk measures perform in an investment management context compared to variance or standard deviation. To our knowledge, this paper is the first to include several acknowledged downside risk measures in a thorough analysis where their different properties are compared with those of variance Risk is an essential factor to consider when investing in the capital markets. The question of how one should define and manage risk is one that has gained a lot of attention and remains a popular topic in both the academic and professional world. This study considers six different downside risk measures and tests their relationship with the cross-section of returns as well as their performance in portfolio optimization compared to variance. The first part of the analysis suggests that the conditional drawdown-at-risk explains the cross-section of returns the best across methodologies and data frequency. Conditional valueat- risk explains the daily returns the best but the worst in monthly returns. Variance, together with semivariance, perform average in both data frequencies. The second part of the analysis concludes that conditional value-at-risk and conditional drawdown-at-risk are the two superior risk measures whereas semivariance is the worst performing risk measure - mainly caused by the poor performance during bull markets. Again, variance performs average compared to the downside risk measures in most aspects of this analysis. Overall, this thesis shows that the choice of risk measure has a significant effect on the portfolio optimization process. The analysis suggests that some downside risk measures outperform variance while others fail to do so. This suggest that downside risk can be a better tool in investment management than variance.