Dutch Economist Henry Kat, profiled in a recent New Yorker feature, was skeptical that hedge funds produce alpha after 2 and 20 (or much higher for the average fund of funds (3 and 30) or funds run by a secret cabal of international quants and a chain smoker), which doesn’t make us feel that bad for thinking the same.
Kat was former head of the equity-derivatives desk at Bank of America but wasn’t down with ‘waking up at 5am, getting on the train and spending all day in the office for 25 years.’ This means that he’s not a complete nutjob, which is reassuring. Kat now settles for less than a hundred thousand pounds a year as a professor.
Kat followed through on his hedge fund skepticism by conducting two hedge fund related studies. The first, published in the June 2003 Journal of Financial and Quantitative Analysis, looked at the fee-adjusted returns of 77 funds from 1990-2000 in relation to returns generated by market benchmarks with similar risk profiles. The result – 72 of 77 funds failed to outperform the benchmark.
The second, posted online as a working paper in 2006, looked at more than 1,900 funds and generated a similar result. Only 18% of funds beat the designated benchmark, and the most successful funds had declining returns over time. The after-fee alpha was negative in the vast majority of cases.
How do hedge funds convince rich investors otherwise? For starters, they exaggerate, demonstrated in a 2005 paper by Malkiel and Saha (a Princeton Prof and a NY investment analyst). The study showed that funds are usually telling fish tales when talking about past performance and that hedge fund returns in aggregate are skewed by the mysterious disappearance of imploded funds from databases. Factoring dead or missing funds into the picture, Malkiel and Saha found that hedge funds made an average return of 9.32% from 1996-2003, instead of the 13.74% average return of funds in published databases. Another study (Brown, Goetzman and Liang) suggested that fund of fund fees negate what is generated in above market returns.
More after the jump...
Hedge fund managers also have a tendency to ride the market with a particular strategy, and look like geniuses until the trend breaks. Several research teams (Fung and Hsieh, Hasanhodzic and Lo) have demonstrated that a large degree of hedge fund return variation is attributable to broad market movements in the price of securities, opposed to the purported genius of computer algorithms or individual investors.
Kat and fellow researcher Helder Palaro (who helped Kat with his second hedge fund study), developed FundCreator, a software program that conducts trades designed to produce the return of designated funds within a similar risk profile, and without the high fees or mystique. The software is designed so that users can customize the management of their “fund” to correlate with other market benchmarks, like the S&P.
Kat and Palaro launched FundCreator as a business in 2006 at the behest of several investors. To have funds managed by the program, investors pay 1/3 of 1% plus the cost of executing all trades, which amounts to less than 2 and 20, and offers what many consider to be a cost-efficient dynamic futures trading strategy with a comparable risk profile. Tracker funds have launched at many of the bulge bracket banks that use a similar methodology as FundCreator.
Several finance gurus and researchers are skeptical that hedge fund performance is so easily mirrored, and argue that new lower cost asset management strategies are just computer algorithms as well. Computer algorithms based on delicate mathematical assumptions that could be wrong. Also, there is still a dearth of publicly available info on hedge funds, especially ones that are 2000 vintage or later.
The buy side, and the media coverage of it, has always been primarily a tale of the top quartile. It’s not so much the fund, it’s choosing one that’s the problem. Top quartile hedge and private equity funds have outperformed the market and produced warmth and fuzziness all around, while the rest have dragged down the pack. There are a lot of garbage funds that create a lot of asset value destruction (observed within the portfolios of some of the more cronyistic and ineptly managed pension funds, and that info is all public in some cases). The other quartiles, or ‘the majority,’ produce the data in the aforementioned studies that lead to conclusions like “the majority of hedge funds underperformed after fees in a certain time threshold.” The top – the Quantums and RenTechs of the world, get the publicity, and continue to give investors a giant erection.
Market saturation with the asset class (or sorry, fee structure), and the fact that so many hedge funds are directly opposed to each other (Arnold vs. Amaranth, Meriwether’s subprime bets vs. the Dillon Reads of the world, etc), eventually causes that attractive top quartile to contract into the top 10%, and so on, so picking a fund isn’t getting any easier.
Hedge Clipping [New Yorker]
Posted by anon, Jun 27, 2007 11:13AM
Hedge fund "replication" is a dumb idea. More recent academic research shows that Kat's method is plain vanilla main-variance. The method is also highly unstable because it depends on historical covariance, which is unstable in financial markets. Finally, the method is also unreliable because Kat's hedge fund replica does NOT match the returns of the real hedge fund, and even the correlation of the replica to other market factors does not match the hedge fund's correlation to the same market factors. You can read the full paper at http://allaboutalpha.com/blog/2007/05/31/allaboutalpha-exclusive-new-northwater-study-finds-hf-replication-techniques-to-be-limited/
Interesting that Professor Kat is not trying to sell his services for a "mere" fee of 33 basis points per year (33 bps of billions of dollars adds up very quickly). Supposedly Kat's value added or market edge is the computer program that his grad student hacked together in order to test the theory which Kat published in academic journals. What kind of idiot would pay for this service when the research is publicly available and the software to implement it could easily be developed at the minimal cost of one starving grad student?