Hedge fund databases are prone to: (1) self-selection bias (only good performers report); (2) backfill bias (only funds with good recent past performance retroactively report it); (3) survivorship bias (exclusion of dead fund performance); and; (4) liquidation bias (poor performers stop reporting but continue to operate for some period). Do hedge fund indexes therefore inaccurately portray industry performance? In the April 2011 revision of their paper entitled “Hedge Fund Biases After the Financial Crisis”, Dieter Kaiser and Florian Haberfelner estimate three of the four hedge fund database biases and explore how these biases evolved during the 2007-2009 financial crisis. They focus on liquidation bias, which leading commercial hedge fund databases do not attempt to control. Their principal analytic technique is to form hypothetical funds of hedge funds from actual single funds and compare the resulting hypothetical returns, after correcting for observable backfill and survivorship biases, with returns from real funds of hedge funds. Using data for 8,935 hedge funds (6,088 single funds and 2,847 funds of funds) for the period January 2002 through September 2010, they find that:
- During the 2007-2009 financial crisis, hedge funds overall lost about 20%, double the loss of their prior worst year (1998). The consecutive number of negative months was unparalleled. Many funds elected special restrictions on redemptions, and hedge fund liquidations outpaced new launches.
- However, hedge fund indexes reached pre-crisis high watermarks in 2009 and even outperformed equity indexes with double-digit returns in 2010.
- Over the entire sample period, survivorship bias is about 2.8% (0.9%) per year for single funds (funds of funds), while backfill bias is about 0.7% (0.00%) per year. The levels of these biases increase during and after the financial crisis.
- Over the entire sample period across all funds, liquidation bias is about 2.3% per year.
- Liquidation bias is about three times higher during and after the financial crisis than before it.
- Maximum liquidation bias across all funds for rolling 12-month intervals is about 11%.
- Across hedge fund styles, liquidation bias is strongest for Equity Hedge (maximum 12-month bias 19%) and Relative Value (maximum 12-month bias 12%) and weakest for Event Driven and Tactical Trading.
In summary, evidence indicates that hedge fund attrition spikes (such as during and after the financial crisis) elevate the liquidation bias in existing hedge fund indexes to as high as 10% per year. Investors should be skeptical of industry performance benchmarks derived from these indexes.
The authors provide lists of studies of hedge fund reporting biases.
One caution regarding findings is that the study sets aside self-selection bias in its analysis.
