Can Hedge-Fund Returns Be Replicated?: The Linear Case
Nov 1st, 2006 | Filed under: Alternative Beta & Hedge Fund ReplicationBy: Jasmina Hasanhodzic & Andrew Lo, MIT
Published: August 16, 2006
This is a much-cited paper that aims to explain the returns of hedge funds, in aggregate, using several risk factors. The conclusion:
“This raises the possibility of creating passive replicating portfolios or clones using liquid exchange-traded instruments that provide similar risk exposures at lower cost and with greater transparency...the performance of linear clones is often inferior to their hedge-fund counterparts, they perform well enough to warrant serious consideration.”
To their credit, the authors point out to questionable practicality of using highly complex risk factors to replicate hedge funds:
“…some of the derivatives-based replication strategies may be more complex than the hedge-fund strategies they intend to replicate, defeating the very purpose of replication…”
As a result, they stick with easy-to-buy factors: ”the stock market”, “the bond market”, “currencies”, “commodities”, “credit”, and “volatility.”
The media has cited this study as evidence that:
To continue reading this article please login (at the right) or click here to learn more about accessing our archives.
[…] Can Hedge-Fund Returns Be Replicated?: The Linear Case […]
[…] MIT’s Andrew Lo (see research) agrees, telling FT his cloning process aims “to threaten existing hedge funds that disguise beta as alpha.” […]
[…] Hasanhodzic, J. and A. Lo, Can Hedge Fund Returns Be Replicated? The Linear Case, Working Paper MIT Laboratory for Financial Engineering, 2006. (ed: link to AllAboutAlpha posting on this paper) […]
[…] The first half of the report is a comprehensive and concise review of the last ten years of academic research on the topic of hedge fund alpha and (later) hedge fund replication. For example, the report’s contention that factor models are less than ”satifying” is backed up by a table showing r-squares in the 30% to 70% range (in-sample), with out-of-sample data being much worse. The authors then re-run the approach used by Andrew Lo (see related posting) and find similarly “unsatisfying” results. […]