Homogenized Hedge Funds

by: Christopher Holt

Critics of hedge funds often point to their use of leverage and exotic financial instruments, their relative illiquidity, and their concentration and counterparty risk. If only hedge funds were forced to adhere to constraints, the argument goes, they would pose less risk to investors and to the financial system. If only hedge funds could be offered with the usual protections associated with mutual funds, all would be well. But regular readers may recall this 2007 post about a paper by Vikas Agarwal, Nicole Boyson, and Narayan Naik that concluded “hedged mutual funds” actually underperform hedge funds. Why? Because the hedge fund fee structure incentivizes managers to perform.

A new study of Europe’s answer to “hedged mutual funds”, UCITS-compliant hedge funds, finds that they perform well enough – but are far less “heterogeneous” than hedge funds proper. The authors of this study, Nils Tuchschmid, Erik Wallerstein and Louis Zanolin say their analysis of UCITS funds has at least one major benefit over those conducted on “hedged mutual funds”: their database is way bigger. There are simply way more UCITS funds than there are SEC-registered mutual funds that use a hedge fund-type strategy. In fact, UCITS funds are responsible for more than 100% of the growth of the hedge fund industry since 2006. Yes, more than 100%. The UCITS database used by the authors of this study grew by around 70 billion USD while they report the total AUM of the HFR Composite index components grew by a total of only around 20 billion USD. The chart below from the paper shows the number (grey line) and AUM (black) of “alternative” UCITS funds (in EUR).

UCITS databases also have advantages over regular hedge fund databases, say the authors. The usual biases found in self-reported data don’t affect UCITS data. For example, “instant history biases” arise from the fact that hedge funds tend to report results only after they have a solid track record. However, UCITS managers are forced to report returns from the moment they launch a fund. “Survivorship bias”, as regular readers know, is the result of databases only containing funds that have survived to the present – not a huge help when one is trying to use that data to predict how a sample of existing funds will perform in the future. Finally, “selection bias” results from a database vendor itself seeking out and extracting data from a sub-set of the overall hedge fund universe. Many hedge funds don’t report to a database and the constituents in a database may not be a truly random sample of all hedge funds. As a result, the authors suggest that extrapolation of results across all databases is dangerous.

UCITS data is free of these biases. But what about the impact of the UCITS rules on the funds themselves. Obviously, UCITS regulations mean that UCITS database constituents must materially differ from, say, HFRI constituents. Tuchschmid, Wallerstein and Zanolin figure there are three things that might lead to this difference…

  1. “restrictions on the level of risk and leverage”,
  2. “limitations on eligible investment instruments”, and,
  3. “an opportunity set which is less prone to contain funds with extreme returns”

Okay, so when you have a database devoid of the usual biases and whose constituents face a set of investment constraints, what do you find? For starters, you find that volatility is way lower. The chart below from the paper shows the cumulative return of the HFRI vs. alternative UCITS funds and the S&P500.

You might wonder if the constraints listed above might affect the investment strategies undertaken by funds and therefore the resulting factor exposures generated by those strategies. The following chart created with data from Table 7 compares the factor loadings of the 7 Fung & Hsieh factors for the HFRI Composite and alternative UCITS funds (hedged into USD to allow comparison).

Definitions (from original Fung & Hsieh paper):

  • S&P: Standard & Poor's 500 stock return.
  • SML: Wilshire 1750 Small Cap – Wilshire 750 Large Cap return.
  • Bond: month end-to-month end change in the Federal Reserve’s ten year constant maturity yield.
  • Credit: month end-to-month end change in the difference between Moody’s Baa yield and the Federal Reserve’s ten year constant maturity yield.
  • Bd Opt: return of a portfolio of lookback straddles on bond futures.
  • FX Opt: return of a portfolio of lookback straddles on currency futures.
  • Com Opt: return of a portfolio of lookback straddles on commodity futures.

Although the UCITS-compliant alternative funds are generally thought to be akin to hedge funds proper, you can clearly see that their factors loadings belie significant differences. Perhaps because of the constraints placed on the types of securities they can hold, the UCITS funds seem to have a higher equity and fixed income loading. But the trio of researchers also finds that the cross-sectional dispersion of a sample of UCITS funds is lower than that of a sample of HFRI constituents. The chart below, created with data in Table 10 shows the range of mean returns and standard deviations in their samples (over the December 2006 to July 2009 time period).

Note that the range of mean returns and standard deviations is much lower among the sample of alternative UCITS funds – even when you compare across the same strategies. It appears from this data that UCITS rules remove the lion’s share of cross-sectional dispersion from fixed income strategies.

In conclusion, Tuchschmid, Wallerstein and Zanolin hedge their bets when it comes to the critical question: which is better? Report the authors:

…for the time being both alternative Ucits funds and hedge funds will remain because they have distinctly different characteristics which bring value to different investment objectives. Alternative Ucits as a group offers better liquidity terms, show lower attrition rate, and exhibit lower dispersion of return. Yet, none of these characteristics can guarantee high performance.

Sure, homogenized milk is easy to digest. But alas, one of the major drawbacks of the homogenization process is that it doesn’t allow the cream to rise to the top…