Does the choice of performance measure influence the evaluation of hedge funds, страница 7

In Table 3, we report the results of various robustness tests. Instead of presenting the rank correlation between all pairs of performance measures for each robustness test (as in Table 2), we report only the (average) rank correlation between the Sharpe ratio and all other performance measures. We found that the main result is robust with respect to

(a)  variations of the database (instead of the ehedge database, we used the Credit SuisseFirst Boston/Tremont hedge fund indices, the Hennessee hedge fund indices, and the Center for International Securities and Derivatives Markets (CISDM) hedge fund database),

(b)  variations of the investigation period (we broke down the period from 1985 to 2004 into five equally long time periods and we investigated separately the time period from 2000 to 2004 to account for the backfilling bias as explained in Footnote 9),

(c)  variations of the exogenously fixed parameters (for the LPM-based measures, theminimal acceptable return was varied between 0% and 1%, for the drawdown-based measures, the number of drawdowns was varied between 1 and 10, and for the VaRbased measures, the significance level was varied between 0.01 and 0.20),

(d)  an elimination of outliers (we eliminated between 1 and 10 of the highest and lowestreturns from the time series),

(e)  a separate consideration of surviving funds and dissolved funds (to account for thesurvivorship bias as explained in Footnote 9), and

(f)  a separate consideration of 23 different hedge fund strategies.

Table 3

Results of robustness tests