Last week's reminder that the collective investment performance of the hedge fund industry has been poor for a decade, as represented by HFR's indices, prompted some familiar objections to the analysis in the comments.
Our approach - comparing hedge fund investment gains to those of a simple stock and bond portfolio of index funds - is too simplistic, the critique goes. It isn't possible to invest in the hedge fund index, so lumping all the strategies together makes no sense, is another one.
By way of answer, we'll highlight an academic paper published in December which took an intellectually sharper blade to the question of investing in hedge funds, and skewered the whole premise for doing so.
Download the dry-sounding Hedge Fund Performance Prediction here, a piece by Nicolas Bollen, Juha Joenväärä and Mikko Kauppila.
Their work looked at what predicts when hedge funds will beat the market, by growing an investor's capital at a faster rate than an appropriate benchmark, and stemmed from a debate in the literature. A landmark 2011 study which assessed the returns of investors in 11,000 hedge funds from 1980 to 2008 found those clients might as well have invested in Treasury bonds. Hedge fund investors take more risk, for less return.
Other work has found there were plenty of individual funds which delivered “abnormal” (ie. good) investment returns, but also that the mayfly-like existence of many funds made them dangerous - the average life of a hedge fund is about five years.
So, having argued that “the ability to predict performance is critical to justifying an allocation to hedge funds”, the authors set out to “study a comprehensive set of the performance predictors developed by existing research to determine whether any have sufficient power to enable investors to reliably outperform a passive benchmark.”
After all, it is not much use trying to invest in a set of hedge funds without some sense of what makes a good one. Otherwise, what would be the point of paying investment consultants to advise on hedge fund portfolios?
Spoiler: none of the predictors “can select funds that add value following the market bottom of March 2009”.
Here's the claim to the weightiness of the study:
Our contribution to the literature is two-fold. First, our analysis is as comprehensive as possible along several dimensions. We measure the information content of 26 predictors developed in published research in a common framework to shed light on their relative ability. In contrast, most other papers focus on just one or two predictors. While several surveys summarise the results of multiple studies, to the best of our knowledge ours is the first to examine this large of a cross section of predictors side-by-side. Furthermore, motivated by the results of Agarwal et al. (2009) and Joenväärä et al. (2015), we consolidate six hedge fund databases to ensure a wide coverage of funds over a long time period (January 1994 through December 2016), which mitigates the potential data mining concerns raised by Fama (1998) and Harvey et al. (2016).
After they broke investment strategies into quintiles (the top 20 per cent, next 20 per cent, etc), they also didn't content themselves with averages. They took thousands of smaller subsets from those quintiles chosen at random, to try to replicate the investor experience.
The benchmark chosen for comparison was a portfolio split 50:50 between the S&P 500 index and Vanguard's Total Bond Fund. (Other comparisons are available.)
The full list of statistical predictors can be enjoyed here (right-click to open a larger version in a new tab):
As for what was found, first the glimmer of hope. A common measure used to assess investors is the Sharpe Ratio of their portfolio, a calculation which captures the risk taken, in terms of price volatility, as well as the investment profit or return. The dumb portfolio had a Sharpe ratio of 0.65, and four of the 26 predictors pointed to top-portfolios which were significantly higher.
In pure profit terms, ignoring the risk, 17 of the predictors led to portfolios which were better than the benchmark. Practitioners might call the difference “Alpha”.
Then the rub: it isn't enough to choose from the top quintile.
This gets directly at the point that an investor can't invest in the hedge fund average, which likely overstates the investor experience. A pension fund is highly unlikely to put money into every hedge fund in a top quintile, but choosing a smaller set of funds worsens the odds of getting the average.
Random samples of portfolios of 30 top-quintile hedge funds were better than the benchmark only 41 per cent of the time. Choose just five top-quintile hedge funds at random, and a mere 14 per cent of such fund portfolios were better than the passive benchmark, the study found.
Remember too that there are biases in hedge fund databases - in a death spiral, sending in the monthly numbers becomes less of a priority.
Also consider what it means to focus on the top-quintile set of managers, assuming it is possible to predict them: 80 per cent of hedge funds aren't good enough. Intuitively, playing for the best one in every five managers is going to be very hard.
Indeed, here is what the paper concluded, and remember that their data runs to 2016:
no matter which predictors are used, and even when using raw returns, we find that [hedge fund] investors could not avoid dramatically underperforming the passive benchmark following the market nadir in March 2009. There likely are a number of superior managers that succeeded during this period, but identifying them ex-ante would have required a different type of insight than measures based on past performance and fund attributes.
There is an array of financial publications that would be delighted to profile the consultant, fund-of-fund manager or pension overseer who has a stellar track record of investing in hedge funds, us included. Managers of pension schemes might want to ponder why they never read about them.
