Monday, October 10, 2011

And Now Back To Our Regularly Sponsored Bad Research Discussion

I'm back to picking through the wreckage of this research by Pertrac on the relationship between performance of hedge funds and their size or age.

In my first post, I explain that two very simple theories may explain a link between smaller, younger funds and better performance.  However, just as importantly, an incredibly common data problem (survivorship bias, where we only have data on better funds) could also explain the result.

In my second post, after having actually looked at some of the paper, I expressed shock and horror that the Pertrac authors actually imposed extra survivorship bias in their data.  I must admit, I really haven't read the paper that carefully.  I just cannot bring myself to care, since it is fundamentally so flawed.

However, I did leave readers hanging...sort of...saying I'd address the four potentially interesting comments in the paper.  The fact is, I was very generous.  Two of them a redundant.  So, what are the two comments that might matter?

Here they are:
  • Young funds have produced better returns with less volatility since 1996...[blah, blah, blah, a bunch of numbers...] (p. 6)
  • Small funds have produced better returns, but with more volatility...[blah, blah, blah,...more numbers...] (p. 7)
These statements might be interesting because they at least attempt to compare performance across different characteristics on a level playing field: They implicitly adjust for observed risk.  They don't actually adjust for observed risk.  

Why not?  Survivorship bias!  They cannot adjust for observed risk because they have very likely (a) included some of the riskiest funds for which risk paid off, and (b) excluded most of the riskiest funds for which risk taking did not pay off.  (Remember, they went out of business because they took lots of risk, and it went bad on them!)

So, while it might be interesting to say young funds have produced better returns with less volatility, I am 100% certain the authors have underestimated volatility.

Just to keep you on the edge of your seat, we'll save one more for the next post.  Can a very simple theory explain why the risk a manager takes might decrease as the fund ages, or grows?  I think you know the answer will be yes!