Tuesday, January 12, 2010

Too Correlated to Save

Another thing that came out clearly in reading Too Big To Fail was the remarkable way people anchor on how the world used to be, and their remarkable inability to envision the possibility of change. The investment banks had been around for years and years, the whole time gradually leveraging up with overnight debt to fund illiquid long term investments. Despite the blatantly obvious instability of this business model, nobody could seem to imagine that it would really collapse at some point, just as nobody now can seem to imagine that a government guarantee of this model will cause a return to the same instability it generated before. Unsustainable things don't go on forever, though people's myopic time horizons and ignorance of history mean that they do seem to repeat forever. Once again, assuming that there are individual human decision makers in this system is just going to cause you to misunderstand it -- if there's anything like collective consciousness, you still have to account for how it got built, a particularly daunting task given that you don't even know how the individual got built.

At any rate, it's obvious that banking is an inherently unstable business model. And apparently there's even a fancy mathematical model with two Nash equilibria if you need to crack the scientific egg on your forehead. In all this, I completely agree with Paul Krugman. But then he Blue the dog.

My basic view is that banking, left to its own devices, inherently poses risks of destabilizing runs; I'm a Diamond-Dybvig guy. To contain banking crises, the government ends up stepping in to protect bank creditors. This in turn means that you have to regulate banks in normal times, both to reduce the need for rescues and to limit the moral hazard posed by the rescues when they happen.

And here's the key point: it's not at all clear that the size of individual banks makes much difference to this argument. It's true that the big losses in mortgage-backed securities seem to have been concentrated at the big financial institutions. But the losses on commercial real estate, which look likely to be even worse per dollar lent, have been largely among smaller banks.

Remember, the great bank runs of the early 1930s began with a run on the Bank of the United States, which was only the 28th largest bank in the country at the time.

The point is that breaking up the big players is neither necessary nor sufficient to protect us against financial crises. That's why my focus is on reducing leverage.

Reducing leverage is a good idea. In fact, I agree, it's the best idea. Though I do think it would slow down economic growth. Not because banking is so innovative or anything, but just because of it's good old fashioned fascism -- everybody keeps working their ass off for the promise that in the future they will be able to retire and take over the world from their arm-chair. I loan it to you and you loan it to me, and we both have to bust a nut trying to pay it off, instead of just making the amount of stuff we are individually content with.

But he errs in suggesting that the size and concentration of baking doesn't matter. In theory, this is true, but there ends the usefulness of the mathematical model. In practice, it's missing a sort of biodiversity argument. The more banks there are, the more possibility that not all of them will lever up and make the same dumb loans at the same time. It's still possible of course, as the run-up to GD I proves. But I do believe that this ecosystem is generally more stable with greater diversity. Though now that I'm thinking about it, I remember reading somewhere that this simplistic idea of diversity=stability and stability is good has some ecological caveats. I seem to remember reading something about increasing diversity destabilizing certain environments. But I would never let the facts get in the way of a good theory.


verdaccio said...

if only everyone was a banker

Clark said...

Why stop there? I've got neurons that would love to get in on the latest sub-prime securitization.