Hedge Fund Databases

When a hedge-fund manager reports its returns to any database, this is a purely voluntary decision—managers are not required to disclose its returns to any data provider or regulator, and are free to stop reporting at any time. Therefore, a number of biases may arise among hedge-fund returns databases that are not present in other asset-pricing databases in which all securities of a given type are included, e.g., the University of Chicago’s Center for Research in Security Prices (CRSP) stock returns database.
First, given that the primary motivation for participating in a database is for market- ing purposes, funds generally seem to begin contributing their returns to a database after a period of outperformance. Because such funds are allowed to include prior returns upon their entry into the database, this practice leads to “backfill bias” or “instant history bias”, further boosting the average returns of funds in the database which are already inflated from the selection bias associated with the decision to be listed. Fung and Hsieh (2000) estimate a backfill bias of 1.4% per year for the Lipper TASS database from 1994 through 1998. Using the Managed Account Reports database (subsequently subsumed by Morn- ingstar Hedge/CISDM) from January 1990 to August 1998, Edwards and Caglayan (2001) estimate backfill bias of 1.2% per year.
Second, funds can choose to delist from the database at any time, and typically do so for one of two reasons: (1) fund managers decide to close their funds to new investments because they no longer have sufficient capacity; or (2) they shut down because of poor performance. These two motivations impart considerably different biases on the data—collectively known as “extinction bias”—the latter being more common and yielding a spurious favorable bias on average investment returns.
Third, some databases do not include extinct funds. This introduces “survivorship bias” and generally increases the average fund’s return.