This study investigates the benefit of investing in hedge funds for an uninformed and passive investor, holding stocks and bonds.

The database used, provided by MAR/Hedge, covers the period from January 1995 to November 2000. Over this period, the database is free of survivorship bias. The funds are classified into 7 categories, according their style: Event-Driven, Market Neutral, Global, Emerging Market, Sector, Short-Sellers and Fund of Funds.

In view of the non-normality of hedge fund returns, the author shows the usefulness of using an alpha with a power utility function as a performance measure, instead of a Jensen mean-variance alpha, which only regresses the hedge fund returns onto passive benchmark portfolios. However, the non-normality level varies among funds: for example, it is high for Event-Driven, Market Neutral, Emerging Market and Fund of Funds styles (with a proportion between 5% and 68%), but 59% of the hedge funds do not have negative skewness and excess kurtosis. In this case, the correction of the non-normality in the alpha is not useful. Only the alpha with a power utility function is computed in this paper. If the alpha is positive, the investor will add the hedge fund to his optimal passive portfolio.

For each hedge fund style, Kouwenberg compares the average alpha for two benchmark sets. The first model is based on the S&P 500 and the Nasdaq. The second model is an extended model which contains option selling strategies on the S&P 500. For Event-Driven, Emerging Market, Sector and Funds of Funds styles, alphas are significantly lower in the extended model. This may indicate that the option selling strategies are a powerful proxy for a part of the return variations in these strategies, by capturing the non-normality in the hedge fund returns.

He also computes the hedge fund models’ R², and the significance of the coefficients of the different factors. The Call Selling factor has a mainly negative significance, indicating that some funds prefer to buy call options instead of selling them. The Put Selling factor is positively significant, mainly for the Event-Driven, Emerging Market and Funds of Funds styles. Moreover, the second model provides a stronger average R² than the first model for each hedge fund style. It confirms that the inclusion of option strategies in a model allows the explanatory power of this model for the hedge fund return variations to be increased.

According to the extended model, 36% of the Event-Driven funds have a positively significant alpha (versus 35% with an insignificant alpha), Market-Neutral 40% (vs 38%), Global Funds 22% (vs 47%), Emerging Markets 6% (vs 27%), Sector 32% (vs 45%), Short-Sellers 35% (61%), and Fund of Funds 28% (vs 41%). This part of the hedge fund adds value to a passive portfolio.

It is interesting to consider whether the benefit of investing in a hedge fund with a positive alpha is liable to persist over a certain horizon. Kouwenberg tests the persistence of the hedge fund returns, from a selection period (January 1995-December 1997) to an out-of-sample period (January 1998-November 2000), thanks to a chi-squared test applied to a contingency table. Two performance measures are used: alpha (with a power utility function) and Sharpe ratio. The author adds a filter to the database, at the end of the selection period: a fund has to have a minimum track record. In the paper, the results presented are firstly drawn from the alpha of funds that have a minimum track record of 12 months and secondly from the Sharpe ratio of funds that have a a minimum track record of 36 months. In the contingency table, the effects of the survivorship bias are highlighted by distinguishing, over the out-of-sample period, between the funds that disappear, the funds with a positive alpha (or Sharpe ratio), and the funds with a negative alpha (or Sharpe ratio).

For the extended model, Kouwenberg finds evidence of performance persistence on the basis of the alpha and the Sharpe ratio, mainly for Event-Driven, Market-Neutral and Global funds, which constitute about 90% of the funds in the database. Sector funds show reverse performance persistence. Emerging markets do not report significant persistence. The results for Short Seller funds cannot be interpreted due to the low number of funds.

Nevertheless, these results should to be considered with care, because a chi-square test associated with a contingency table tends to overestimate the persistence level. Moreover, the proportion of disappearing funds is high in the out-of-sample period (29.6%), so it decreases the significance of the results.