A report from Ibbotson Associates indicates that the returns reported in hedge fund databases are often much higher than they should be because of backfill bias and survivorship bias.
Backfill bias occurs because many hedge funds include unreported performance to data collectors when they first start reporting returns; they typically report only favourable early returns, says the report.
Survivorship bias occurs because when a fund fails it is removed from a database along with its performance history. This means databases tend to track only the successful funds.
The report shows that the compounded annual return of the Tass database of hedge funds from 1995 to March 2004 was 16.64%. Without backfill, this figure was 14.74%. If dead funds are included, the figure was just 9.06%. The authors add that value-weighted indices are likely to have less severe biases. The HFRI Index and the CSFB/Tremont index returned roughly 13% over this period.
