Does the choice of performance measure influence the evaluation of hedge funds, страница 6

We use two test statistics to check the significance of the rank correlations. First, statistical significance is tested using a standardized version of the Hotelling-Pabst statistic. In this test, the hypothesis of independence of the two related rankings is checked for all correlation coefficients. However, even at the significance level of a = 0.01, there is no case in which the hypothesis of independence can be confirmed. Therefore, the hypothesis of independence of the measurement series must be rejected for all correlation coefficients.

Instead of testing whether the rankings are independent (in other words, the rank correlationiszero),itispossibletocheckthehypothesisthattherankcorrelationissmaller thana certain given rank correlation x. We did this using the Fisher transformation and found for a significance level of a = 0.01 that the hypothesis that the rank correlation is smaller than x is rejected for all x smaller than 0.917 (see Rees, 1987, p. 383, for the test statistic).^[11]

In conclusion, on the basis of our data, none of the new performance measures results in significant changes in the evaluation of hedge funds as compared to that found using the Sharpe ratio. Thus, it does not much matter which of the numerous measures is used to assess the performance of hedge funds. Because the newer performance measures result in rankings that are practically the same and thus each gives a similar assessments of hedge funds, use of the Sharpe ratio (even if it displays some undesirable features) is justified, at least from a practical perspective.

3.3. Robustness of the findings