Hedge Funds

Hedge Fund Returns – survivorship bias and future return forecasts

Hedge Fund Returns – an overview of research

Or go directly to our library of articles and white papers on hedge funds

Forecasting future hedge fund returns is difficult. It doesn’t help that investment professionals and academics can’t even agree over historic returns, given the problems of survivorship bias and backfill bias.

Survivorship bias and Backfill bias

“Survivorship bias” is the tendency for dead funds to be excluded from measures of historic performance. Failed funds have, on average exhibited much lower returns than surviving funds. Depending on which hedge fund studies you place most faith in, this means that historic hedge fund index returns are overstated by 3-8% pa. See the Lyxor paper below for a fuller discussion.

“Backfill bias” is similar, in that it will tend only to be successful hedge funds that join an index, at which point they will provide the index provider with data of their historic returns.

Hedge fund distributions

For most hedge fund strategies, returns do not follow a normal distribution. Equity hedge and equity driven strategies, for instance, are negatively skewed with fat left-tails, due in part to the use of options or dynamic trading strategies.

The drivers of hedge fund returns

In their July 2015 research report, “A new era for Hedge Funds” Lyxor ran some simple regressions to examine hedge fund returns using the HFRI Composite Index. This suggested that, over the last 25 years, equity beta has been around 35% and bond yield beta around 60%. Since 2009, equity beta has fallen and bond beta has risen, with the two variables in combination explaining 73% of hedge fund returns. They estimate that alpha generations was running at 4.5% pa from 1990-2009, but fell to 1.2% pa from 2009-2014.

The authors readily admit that, decomposing returns into just two betas, and describing the residual excess return as alpha, is a huge oversimplification. Some academics believe there is a whole “zoo” of risk factors, or alternative betas, which explain hedge fund returns.

BlackRock, for instance, in their paper “Dissecting and differentiating hedge fund returns” describe the components of hedge fund returns as comprising not just old-fashioned market beta and alpha, but also “smart beta” and “exotic beta”. BlackRock describe smart beta in the conventional way; exposure to risk factors such as value, momentum, size or quality which have demonstrated added value over time. BlackRock’s analysis of returns from 2006-2013 suggests that for equity-related hedge funds, around 35-40% of excess return is explained by equity beta, with 10-15% due to smart beta. In fact, the better performing funds were markedly more dependent on these beta factors.

From smart beta to exotic beta

BlackRock describes the main sources of exotic beta as two-fold; insurance provision and liquidity provision. Providers of insurance include those willing to take on catastrophe (bond) risk, event-driven risk, merger arbitrage risk, distressed asset risk, or short volatility risk. While providers of liquidity generate return from statistical arbitrage, new issue arbitrage and passive purchases from forced/distressed sellers.

Long-term Forecasts of Hedge Fund Returns

The quick and dirty way to construct hedge fund return estimates is to decompose expected returns into risk factors; for instance to make assumptions about market beta returns, to add estimates for alpha and alternative beta and to subtract an allowance for hedge fund fees.

So taking an equity long-short strategy which averages 70/30 equity/cash: if equity returns 6% pa and cash 2% pa, with alpha (plus alternative beta) of 1.7% pa and fees of 2.0% pa, then the net long term expected return is:

Annual return = ( 0.7 x 6 ) + ( 0.3 x 2 ) + 1.7 - 2.0 = 4.5% pa

Of course, this is hopelessly over-simplistic, but it provides a starting point for constructing rational expectations of hedge fund returns.

Lyxor, in their “new era” document, posit that in the future economic environment, alpha (and alternative beta) generation is likely to add around 3-4% pa, which when added to their assumptions on equity and bond markets, generates returns of 5-6% pa in excess of the risk-free rate. This is perhaps on the bullish side of reasonable expectations.

Forecasts of Hedge Fund Returns

For comparison, here are the latest total return estimates from JP Morgan and Northern Trust for broad hedge fund categories, compared to their estimates for cash and equities.

 JP Morgan (long-term)

 Northern Trust (5yr forecasts)

 Equity Hedge Strategies

5.5%

3.8%

 Equity Driven Strategies

6.0%

5.0%

 Relative Value Strategies

5.25%

5.0%

 Macro Hedge Fund Strategies

5.0%

4.2%

 Hedge Fund Composite

4.25%

4.4%

 Cash

2.0%

1.5%

 US Equities

6.5%

5.6%

Source:

Northern Trust (2015) Five Year Outlook, 2015 Edition

JP Morgan (2015) Long Term Capital Market Return Assumptions 2016

Hedge Funds and Manager Skill

In practise, there is a wide dispersion of hedge fund styles and strategies, and an even wider dispersion of manager skill. Within strategies, this dispersion of course narrows, but research from Morgan Stanley shows that over the period from 2000-2013, most strategies shows a dispersion of 3-7% pa between the top and bottom decile performers – a huge difference when accumulated over a fourteen-year period.

Recommended reading – white papers on hedge fund returns

"Alpha, Beta and Costs: The Hedge Fund ABC”  Ibbotson, Chen and Zhu (2011)

The authors break down historic hedge fund returns into three component parts - alpha, market beta and costs. In this analysis, the traditional definition of beta is used (any “alternative betas” are defined here as “alpha”). After dissecting the results from 1995-2009, the authors estimate that the equally-weighted return from hedge funds over the period was made up of Alpha - 3.0% pa, Beta - 4.7% pa and Costs (3.4% pa. The post-fees return of 7.7% pa after fees would have been 14.9% before adjustment for backfill and survivorship bias.

”Hedge Funds: A Dynamic Industry in Transition” Getmansky, Lee and Lo (2015)

This papers takes a broad overview of academic research and also provides updated empirical evidence on hedge fund returns. One of the conclusions reached is that survivorship and backfill bias cut hedge fund returns in the Lipper database by half.

”Relax, Hedge Funds Do Just as Well as You Thought”  Eqira (2015)

This note from quant analysts Equira seeks to refute the claims in the paper above that hedge fund index returns are halved by backfill and survivorship bias. They observe that hedge fund indices are already adjusted for such bias, and support their arguments by examining both FOHF returns and those from endowments.

”Hedge funds in strategic asset allocation”  Lyxor Asset Management (2014)

This wide-ranging 63 page report from Lyxor suggests a new framework for hedge fund classification, which is based upon their sensitivity to common risk factors. This comprehensive report analyses a variety of statistics relating to hedge fund returns, risks and correlations, citing numerous academic studies. The authors also decompose the excess return from each strategy type into traditional beta, alternative beta and alpha. Finally, the authors examine the equity beta and bond beta in each strategy and propose a framework for strategic asset allocation, whereby funds should be considered either “equity substitutes” or “bond substitutes”.

”An Outcomes-Oriented Approach to Alternatives”  Morgan Stanley (2014)

Morgan Stanley’s 20-page paper, “An Outcomes-Oriented Approach to Alternatives” sets out a framework for institutional investors to use when allocating to hedge funds. The papers analyses a number of hedge fund strategies, describing the correlation of their returns with macro risk factors. The paper is particularly useful for helping investors consider their diversification objectives, and how to meet them with an appropriate selection of funds.

”Passive Hedge Funds”  Tupitsyn and Lajbcygier (2015)

This study argues that for two-thirds of hedge funds, returns are derived only from linear exposure to systematic risk factors, and therefore should be considered “passive” rather than “active”. Furthermore, “passive” managers tend to outperform “active”. This echoes the arguments in favour of hedge fund replication as a cheaper way of achieving “hedge fund” type returns.

”Hedge Fund returns - a new era?”  Lyxor Asset Management (2015)

“Dissecting and differentiating hedge fund returns”  BlackRock (2014)

”Hedge Funds: Value Proposition, Fees, and Future”  Cambridge Associates (2013)

”Chasing Hedge Fund Winners: The Appeal and the Risk” Commonfund Institute (2015)

”Hedge Fund Replication” Hewitt EnnisKrupp (2013)

”Hedge Fund Returns vs. Traditional Asset Classes in Interest Rate Tightenings” HIMCO (2014)

”Hedge Funds Returns After CalPERS - The Case for the Defense” Cliff Asness (2014)

related content

Feedback