New Paradigms for Evaluating Performance and Performance Persistency of Domestic and Globally Diversified Portfolios

This manuscript reexamines performance evaluation of managed portfolios. Past measures of portfolio evaluation such as the Sharpe, Treynor, and Jensen measures are subject either to the inability to rank performance based on statistical significance, or are dependent on both a single factor CAPM return generating process and the selected market portfolio. Recent studies show performance ranking is sensitive to the selection of the market proxy when the security market line is used to evaluate performance. Additionally, CAPM based measures that appeared to work well in the 1960\u27s no longer appear to function effectively. Many anomalies to CAPM have been documented since the 1970\u27s and recently, Fama and French (1992) declared the CAPM beta to be dead. To date, no agreement among scholars has been reached on the appropriate return generating process or the appropriate market proxy. Therefore, performance evaluation of managed portfolios is laden with ambiguity. What is needed is a measure that can rank performance on a statistically significant basis, is applicable to a variety of return generation processes, is free of the requirement to observe the market portfolio or the true risk free proxy, and can bridge the gap between theory and practice. This manuscript provides such a measure. The economic and legal rationale for portfolio manager behavior is reviewed, a method to detect if the differences in performance are statistically significant is provided, and a method to rank portfolios with statistically significant performance differences is provided. The proposed methodology is tested empirically on open-end mutual funds for the period September 1981-94 and results of the new measure are compared with those of the traditional measures. Further, tests of performance persistency are conducted to detect if past relative performance can be a guide for predicting future relative performance. Finally, persistency test results for the four methods of evaluating performance are compared. Results indicate that while performance and performance persistency exists for traditional measures, it is not evident with the proposed procedure. This supports the efficient market hypothesis and suggests that performance detected with traditional measures is merely a CAPM anomaly, not true differential performance