Forget about model-gaming, how about model-anointing?
December 20, 2012 9 Comments
I should note, though, as someone who has defended Brand Nate Silver™ in the context of his electoral-college simulation, that her post is sort of a revelation to people (like me) who had known nothing about actual Nate Silver. I had no idea, for example, that he had written something like this:
To suggest that Obama or Geithner are tools of Wall Street and are looking out for something other than the country’s best interest is freaking asinine. [...] This is neither the time nor the place for mass movements — this is the time for expert opinion. Once the experts (and I’m not one of them) have reached some kind of a consensus about what the best course of action is (and they haven’t yet), then figure out who is impeding that action for political or other disingenuous reasons and tackle them — do whatever you can to remove them from the playing field.
Straight out of the Anakin Skywalker school of public policy:
ANAKIN SKYWALKER: We need a system where the politicians sit down and discuss the problem. Agree what’s in the best interests of all the people, and do it.
Turning back to mathbabe’s points, the one deficiency I see in her diagnosis of model-misuse is that it ignores the role of government. Everything is the fault of ‘a corrupt and criminally fraudulent financial system’. Which, okay. But what caused that? What enables it? There is no attempt to take a step back and view the larger picture. Or, perhaps one step back is taken, but you really need two.
Let’s take her example of AAA securitizations. I think the OWS/intelligent lefty critique of that particular model-misuse in finance goes something like this:
- Evil banks noticed they could game rating-agency models and stuff deals full of bad shit while still achieving a given rating, which would make them money for some reason let’s not get into or think too much about.
- When honest rating-agency employees cried foul, they were corrupted (“we’ll take our business and the big fees elsewhere if you don’t”, and/or “there’s a job in it for you on my desk next Jan if you play ball”) or steamrolled (“you’ll get fired if you don’t play ball”). So, for institutional reasons that we might look to fix/address/regulate away, bad risk was illegitimately stamped AAA.
- The banks made money and the deals turned sour.
- Because, banks are evil (see 1).
Again: okay. Fair enough. But don’t you need a step 0?
0. Once upon a time, for some reason, there were these big places called ‘Rating Agencies’, and whatever tiny string of letters they slapped onto an investment was somehow a really big deal, on which, oddly, billions of dollars could ride. Like, if a thing had ‘AAA’ stamped on it then certain folks would almost unthinkingly buy it.
But if you try to start this story at step 0, and if you actually think about it, it doesn’t work either, because you start to ask questions. Questions like, well, why? Because, that’s strange. And odd. And not natural. Nor some kind of organic free-market outcome.
And once you notice that and dig into why, almost inevitably, you end up with this answer: because of some government decision and/or policy back in the day. Seriously! Try it at home! I’ve found that it basically works every time. At least, I’ve encountered no exceptions.
So it’s true that people in finance gamed those ‘rating’ models and used them cynically, to try to maximize profit. Of course. But none of that would have mattered one iota if ‘Rating Agencies’ hadn’t been anointed the Official Government Arbiters Of Credit in the first place. It’s no different than a regulation declaring how big eggs have to be to be ‘Grade A’, or how much milk fat egg nog must have for you to be allowed to call it nog, and whether you can put turmeric in it. Yes, duh, no shit Sherlock, market actors will find the cheapest ways possible to meet your arbitrary rules and thresholds. It’s all well and good to get mad about this sort of model-gaming, but where is the outrage at the model-anointing that incentivizes it?
More to the point, we still make all the same mistakes when it comes to model-anointing. For example, see the “Basel III” style approach to capital.
But this boils down to your views on human nature, I suppose. One sort of view a person could take might be:
Government policy is a given. Government policy is there to do good. Government policy has decreed that it’s a good idea to officially and legally link capital-requirements, among other things, to rating-agency ratings. And so, that’s a given and we should not question it; it’s just the way things are, like the landscape. Now in an ideal world, it would all work perfectly because no one would ever game anything or respond to the incentives created by the policy; it would just work as intended. So that’s what we need to aim for. Only Greed prevented it from working. Therefore, what we need is to clamp down on and stamp out Greed, and disallow Greed from influencing the use of models. Then everything’ll be fine.
Call this the lefty/OWS view. But there’s another view which goes like:
Human nature is a given. One aspect of human nature is that people are Greedy. Another, closely linked, is that they respond to incentives. In context this means that if you set up a government rule which artificially adds value to a product based on some agency’s model-based ‘rating’, you can be sure that people will try to figure out the cheapest way to deliver a given ‘rating’. So, knowing this, how about let’s not do that maybe, or if we do, do it with open eyes and not whine about the predictable results afterward?
I would call this the conservative view. It is, of course, the view that I hold.