Hedge Fund Performance During and After the Financial Crisis
In this research, a deductive approach was used. Quantitative data on hedge fund performance was collected in the form of monthly returns of the major hedge fund strategies, between June 2007 to January 2017. The performance of each strategy, value-wise, was first visually inspected and then return data were analysed. The correlations of hedge fund strategies with the benchmark S and P 500 was also studied.
To ascertain explanatory factors for the performance of the various hedge fund strategies and the extent to which the strategies incorporate market risk, the strategies were tested via Carhart's four-factor model (1997). The four-factor model was chosen as it can provide additional explanations for performance over the traditional CAPM and Fama and French (1993) three-factor model, explained earlier. The four-factor model was an extension of the Fama and French (1993) model and included four risk factors to explain returns (excess market return, excess return of small cap over big cap stocks (SMB), excess return of value stocks over growth stocks (HML), and the excess return of the stocks that went up minus the stocks that went down (UMD).
Regressions were estimated as follows:
To obtain further insights into the strategies which hold value for investors, tests for persistence and risk-reward rankings were also conducted. The traditional Sharpe ratio had been a useful metric to rank fund strategies, on the basis of reward and risk. Given asymmetricities in the returns of hedge funds, evaluation of performance was also made in a downside-risk framework, with reward-risk ratios, such as the Upper Potential Ratio and the Sortino Ratio.
Investors look at persistence in performance at two levels – consistent positive performance and consistent performance better than the market. While a lot of work has been done by various authors on the subject of persistence in performance of Hedge Funds, this research relied on the work of Agarwal and Naik (2000), who after extensive analysis of quarterly, half yearly and yearly returns, found maximum persistence at the quarterly horizon.
Taking these considerations into account, persistence tests were devised as follows:
(i) To test for persistence, t-tests were conducted, over a three-month rolling window.
(ii) The total number of statistically significant negative performance windows were subtracted from the total number of statistically significant positive performance windows, and the different strategies were ranked for persistent positive performance on this basis, then tested against the ranked total performance of the various strategies, to test for any relationship between persistence in positive returns and actual performance. A non-parametric test was used for this test.
(iii) The procedures in (i) and (ii) were repeated for excess performance, over the return of the S and P 500, for the whole period, during and after the crisis.
(iv) To test for any relationship between total performance and alternative reward-risk ratios, the Sharpe ratio, Upside Potential ratio (UPR), and Sortino ratios were computed for the different strategies, then ranked and tested in comparison with actual performance. The t-bill rate was taken as the minimum acceptable return (MAR) in these calculations. A non-parametric test was used for this test.