Investment performance appraisal and asset pricing models

Abstract

In today's increasingly volatile financial markets and uncertain global economic environment, stock markets are becoming more and more risky and managed funds' performance is rather disappointing. Therefore, investors' demand for accurate fund rating and security/portfolio return prediction is at an all time high. This thesis addresses three important related issues: (i) rating Australian mutual funds' efficiencies using data envelopment analysis (DEA), (ii) assessing the significance of various risks related to higher-order moments in pricing securities/portfolios and (iii) investigating the beta-return relationship across various market volatility regimes.Because of its ability to incorporate many input factors affecting the mutual fund's performance, the DBA technique has advantages over other performance measures proposed in the finance literature. Currently, fund managers are being criticised for rating their funds based on short-term performances only. In this study, many input-output factors capturing short-term and long-term performances are used. Further, an input variable capturing management strategy is constructed and a pleasing result is that this factor turns out to be very important in explaining the variation in inefficiency across mutual funds. When the analysis includes long-term performance characteristics more and more funds become DEA-efficient. Further, a simulation study is conducted to assess the sensitivity of efficiency rating to variable selection and the retuns-to-scale assumption. When the DEA model does not include all the variables deemed relevant, the variable returns to-scale assumption appears to be a safer option than the constant returns-to-scale assumption.The second issue addressed in this thesis concerns the importance of higher-order co-moments in pricing securities/portfolios. Many previous studies that examined unconditional asset pricing models with higher-order co-moments produced mixed results. To our knowledge, this is the first study to consider these models conditional on up and down market movements. For a sample of Australian securities, the systematic variance and systematic skewness are found to be priced, and these significant results were not uncovered in the unconditional model.Finally, in order to investigate the stylised fact that high market volatility leads to high returns, the CAPM beta is allowed to vary across the three market volatility regimes, namely, low, usual and high. Using thirty Dow Jones industrial stocks and a sample of Australian industrial portfolios, this conditional model was investigated as to whether the three-beta pricing model is useful in explaining cross-sectional security/portfolio returns with the result that no statistically significant evidence was found.