Hedge Fund Conference Kellogg School of Mgmt

1
Hedge Fund ConferenceKellogg School of Mgmt

• Roundtable Discussion
• Hedge Fund Performance Evaluation
• Lawrence R. Glosten
• Columbia University

2
The academic approach

• Portfolio manager provides excess return (over
  the risk free rate) Xpt in period t
• These returns are modeled as
• Xpt a ß1X1t ßnXnt et
• Where the Xs are excess returns that capture the
  risks that investors care about
• a is the value provided by the portfolio
  manager which in general depends upon the factors
  that are included

3
Performance evaluation problems

• There is some suggestion that the excess returns
  may be measured with error
• There is some (though certainly not universal)
  agreement on what Xs should be used for modeling
  simple equity returns
• Fama/French three factors market, small minus
  large, high minus low (book/mkt)
• Less agreement on how to model more complicated
  returns from hedge funds
• Worry about performance evaluation affecting
  investment strategies

4
The returns

• Self-reported returns, including back-fill
• Back fill data identified as such
• Some odd statistical properties
• In a market model regression with lags, lagged
  market returns have significant coefficients
  (Asness, Krail and Liew)
• Hedge fund returns exhibit more serial
  correlation than one would expect if returns were
  calculated from market prices (Getmansky, Lo and
  Makarov)

5
Modifications to standard performance evaluation

• Non-linear termsoptions on the market (Glosten,
  Jagannathan)
• Merger Arb writing puts (Mitchell, Pulvino)
• Trend following buying straddles (Fung and Hsieh)
• Lagged market returns to correct for smoothed
  hedge fund returns (Asness, et al)
• Other portfoliosbonds, currencies, commodity and
  security straddles (Fung and Hsieh)

6
Hedge fund alphas and betas

7
Problems

• As factors increase in number, run into too few
  data points per parameter
• Things (betas, alphas) change through time (Fung,
  Hsieh, Naik, Ramadorai)
• But, may not need so many factorsfor consistent
  estimation of alpha, need to include only risks
  that investors care about

8
Finally

• Even if we knew all alphas, betas and R2s ,
  this does not say how much an individual
  (organization) should invest in a hedge fundthis
  depends upon preferences
• At least knowing the betas gives an idea of how
  hedge funds fit into rest of portfolio and how to
  manage risk with other securities
• Quantitative estimation might be helped by
  qualitative information from hedge funds

9
Questions I

• For Steve Nesbitt
• A quantitative analysis of performance provides,
  in addition to alpha, betas with respect to
  various risks. How do you use this information
  to advise clients on hedge fund allocations?
  Perhaps you could also talk a little about
  portable alpha.
• For Don Fehrs
• What information do you get from hedge funds in
  addition to past returns? What are the important
  qualitative inputs to deciding to carry (or drop)
  a hedge fund? Has there been much change in your
  portfolio of funds over time?

10
Questions II

• For Gary Brinson
• It appears that the contract between investors
  and hedge funds has been fairly standard, and is
  based on returns rather than on performance as
  we have been discussing the term today. Do you
  see this practice as persisting, or might we see
  alpha compensation?
• For Girish Reddy
• Fund of Funds (FoF) growth has been rather
  remarkable (Steve Nesbitt reports that FoFs held
  36 of hedge fund assets at year end 05, compared
  to 17 in 00) Three developments could lead to a
  reduction in the attractiveness of paying a fee
  to FoFs increased investor sophistication,
  intermediation by consultants, development of
  multi-strategy hedge funds. What do you think?