The Martingale’s Song

Lots of pundits have been trying, apropos the current financial crisis, to point out how those evil private equity funds have rigged the game. The basic argument is that if I run a fund which makes a series of bets, each of which has a very high probability of a small positive payoff and a very small probability of losing all the money in my fund, then I’ll very likely show great returns for many years until one terrible day when I lose everything in the fund. During those years, as the fund manager I’ve pocketed a management fee plus profit-sharing which I don’t have to give back. The investors on the other hand have lost all of their money.

You know, it would be a lot easier to run a successful fund if only institutional investors were this gullible. Unfortunately, there are a few problems with the theory.

First, the investment structures for the vast majority of private equity funds I’ve seen have this nasty feature called a “clawback”, which is petty much exactly what it sounds like. In fact, the general partner usually does have to give back the interim profits that it took if the fund loses money (or underperforms some benchmark) prior to its close and the distribution of all money to the investors. The reason that clawbacks exist is exactly that investors don’t want to be stuck in this situation. It’s not as if this investment issue is a new discovery that is dawning on us only now. In concept, this is the same basic problem that was understood vis-à-vis the martingale betting strategy 200 years ago.

Another problem is the fact that unless you planned to retire early upon the closing of that fund, you would have a very hard time raising more money when the IRR on your prior fund was negative infinity. Believe it or not, guys who run a series of successful funds usually retire richer than guys who wipe out their investors.

The more fundamental problem is that the hypothetical fails to distinguish between risk and uncertainty. We have learned in retrospect that the actual risk of various mortgage securities was greater than market participants (who have generally lost a lot of money as a result of their misjudgments) believed at the time. But contrary to the hypothetical that is often proposed, there was not some “kind of gambling machine somewhere that works kinda sorta like an enormous roulette wheel”. That is, the probability of failure was not known in advance the way it is on a roulette wheel – that probability is exactly what the markets were trying to estimate. They were wrong, as market participants often are, but that’s very different from being a con artist.

(cross-posted at The Corner)