LGSV vs LMS (Large Minus Small) Hedge Funds ) vs OLS

Hi David,

This one is another gud one, need to understand better for exam purpose.
Wud highly appreciate for your help.

You are being interviewed for the position of CRO for a large fund-of-funds. You are asked to comment on the risk management approach of the CRO you would replace, James Smith, and who just left to start his own fund-of-funds. To evaluate the risk of a hedge fund that invests in equities over the coming month, Smith proceeded as follows. Using monthly data, he would regress, using ordinary least-squares, the return of the fund over its history on the return of the S&P 500, the return of a portfolio long growth stocks and short value stocks, and the return of a portfolio long large firms and short small firms.He would then use an EWMA model to forecast volatilities and correlations for the risk factors. Using the exposures of the fund to the risk factors and the standard deviation of the residual of the regression, he would then forecast the volatility of the fund and use the parametric approach assuming normally distributed returns to estimate the one-month VaR of the fund. Which of the following statements are correct?

I. The well-known work of Fama and French tells us that Smith uses an appropriate risk model, so that the risk factors do not need to be changed. However, his approach makes the mistake of ignoring the fact that hedge fund returns are not normally distributed. The correct distribution of the residual should have higher kurtosis than the normal distribution.
II. The estimates of the exposures to the risk factors will often be biased since the returns of many hedge funds exhibit high serial correlation compared to the returns of mutual funds.
III. The model used by Smith will have low explanatory power because hedge funds change exposures to the risk factors he uses often, but the explanatory power could be improved by using additional asset based factors that have been developed in the literature.
IV. Since portfolio holdings for the typical hedge fund change so much, it is hopeless to hope to explain more than a trivial faction of the return of a typical hedge fund using a factor model.
Choose one answer.

a. Statements I and II are correct.
b. Only statement I is correct.
c. Statement IV is the only correct statement.
d. Statements II and III are the only correct statements

Regards,
Rahul
 
Hi david,
This look too big, but beleive me this are the real FRM question. This kind of lenght is normal of FRM Question.
The answer for the same are II & III. cud you plz explain why.

Thanks
Rahul
 

hsuwang

Member
Hello Rahul,

Can you kindly give me the source to this question because I've seen it before but can not recall from where. Thanks!
 

David Harper CFA FRM

David Harper CFA FRM
Subscriber
Rahul - I agree with Jack that sources would help, only because older questions often times are not relevant to the current candidate. This is an example, IMO, of a question that doesn't too much help the current candidate. It draws from a fixed income hedge fund replication paper that last appeared in 2007 (Fung & somebody). I see why (I) wants to be false (non-linear model) but since that time Andrew Lo's readings have been included and a candidate can fairly tolerate "linear clones" so (I) becomes debatable...

...similarly (III) is context-dependent on the Fung reading ... current candidates (infomed by Grinold & Lo) should understand hedge funds have non-linear exposures but they do not need to reject a linear regression model or a model that employs fewer (Fama-French) factors
...in short, a 2009 FRM candidate would be *appropriately* vexed by this question!

IMO, there are two useful ideas in the question:

1. Per II, hedge fund returns tend to exhibit positive autocorrelation; Lo has famously uses autocorrelation as a proxy for the relative illiquidity of hedge funds
2. The use of the Fama French factors (reference: Jaeger's hedge fund strategies): HML, SMB, & UMD

David
 
Top