Sunday, March 09, 2008

Does Warren Buffett have Investment Skill?

The outstanding performance of Berkshire Hathaway's stock price and book value over time is well known as is more anecdotally the performance of Berkshire's stock investments. It's very straight forward to compute an alpha and beta for the company's performance - I think alpha is around 10% - but much harder to examine the performance of Berkshire's equity investments. This is because Berkshire is mainly an operating business - insurance companies and non-insurance subsidiaries and only secondarily an investment vehicle.

A recent paper claims to be the first academic study to do so. They come up with an excess return of 6.5% to 12% depending on the benchmark. They do a whole bunch of other tests including comparing performance to a Monte Carlo simulation meant to model the differing levels of luck of different investment managers. Under reasonable assumptions, Buffett has done better than even the most lucky manager would assuming no-one has any excess investment skill. The authors also construct a mimicking portfolio which buys and sells the shares that BRK trades only when the trades are publicly disclosed. This portfolio does almost as well as the BRK portfolio. This means that copying Buffett is worthwhile and violates even a broad definition efficient market theory. Even if Buffett is rewarded for uncovering and interpreting profitable information the disclosure should result in that information being instantly priced into the market. But the mimicking portfolio assumes that you buy or sell the same stocks as Buffett only at the end of the calendar month in which the disclosure is made!

1 comment:

Anonymous said...

What a great post -- thanks for pointing me toward this paper. The part on mimicking Buffett so much after he makes his moves is really striking.