Stephen Laniel’s Unspecified Bunker

Attention conservation notice: 1,500 words, trying to put together in a messy way all my guesses about the subprime collapse, mixed in with the little bit I know about economics and statistics. Likely to be valuable only to those who enjoy watching elephants draw with crayons.

Paper Of The Moment is the above-named, by Brad DeLong et al. I picked up the recommendation via Richard Thaler’s excellent and very readable paper on “The Law of One Price in Financial Markets”. Thaler has made something of a career out of exploring — and poking good fun at — what economics considers paradoxical or anomalous. The big anomaly in “One Price” is that sometimes financial markets get into a weird state where the market assigns facially ridiculous prices to securities and holds them there for a long while. Shares of 3Com, for instance, were for more than a month being assigned a negative price. The big law in economics is that under reasonable assumptions about market structure, this can’t happen, or at least not for long — and the more fluid the capital market is, the more quickly this should disappear. The story goes that rational people will step in when they see a systematic irrationality and profit from it. So long as irrationality persists, where the price of a stock is different from what the fundamentals say it should be, rational people will move in and make quick money. They will buy shares of 3Com at the implicit negative price, for instance, then wait a while for the market to correct itself, then sell and make a quick profit. This will continue until all the irrationalities have been driven out of the market.

Why does that fail? Thaler goes through a few possibilities, then finds cases where none of those possibilities hold and yet the market’s irrationality continues.

Hence we come to DeLong’s paper. The abstract is clear enough:

We present a simple overlapping generations model of an asset market in which irrational noise traders with erroneous stochastic beliefs both affect prices and earn higher expected returns. The unpredictability of noise traders’ beliefs creates a risk in the price of the asset that deters rational arbitrageurs from aggressively betting against them. As a result, prices can diverge significantly from fundamental values even in the absence of fundamental risk. Moreover, bearing a disproportionate amount of risk that they themselves create enables noise traders to earn a higher expected return than do rational investors. The model sheds light on a number of financial anomalies, including the excess volatility of asset prices, the mean reversion of stock returns, the underpricing of closed end mutual funds, and the Mehra-Prescott equity premium puzzle.

I’m looking forward to digging into it.

As I’m sure a lot of people are doing right now, I’m trying to get my head around the subprime bubble. Again assuming a lot of different stories, the bubble can be made to look paradoxical. Why were Moody’s et al. asleep at the switch? (”Standard and Poors cut the credit rating of ACA Financial Guaranty Corp from AAA (the best possible) to CCC (just about the worst kind of junk) in one move.”) And does their failure here say anything broader about market irrationality? A lot rides on the safety grade that the ratings agencies give to individual mortgages. Surely it’s unreasonable to expect any single institution to handle this properly; normally we hope for an entire market to weigh in on something like this in its collective, Central Limit Theoremy way. So first off: why do we rely on Moody’s, S&P and a few others to do this big job?

I assume there’s some good reason for offloading risk measurement onto them. Assuming that we’re stuck with a few large institutional risk measurers, smart regulation would seem to be vital; we need the government here because, for whatever reason, we’ve decided that the market itself can’t measure this kind of risk.

Financial companies are sitting on the top of a giant pile of abstractions, but Cosma has convinced me that there’s nothing particularly concrete about the lowest levels of this pile of economic objects: much as we might wish for it, the phrase “the value of a house” is itself a massive abstraction, resting on the same shifting sands of supply and demand. Besides: if you actually had some surefire way to measure “the value of a house” that didn’t involve just asking the market what it would pay for that house, you could make virtually unlimited money (again assuming some things about how long it takes the market to realize the error of its ways and price the house in line with your understanding, how much it costs you to hold onto the house in the meantime, and how willing banks are to loan you money). So while there does seem to be a problem of insane abstraction, it’s not clear that we can get rid of that abstraction by (somehow) forcing companies to buy and sell only mortgages rather than collateralized debt obligations.

If I’m not mistaken, my picture of the whole problem took a turn for the surreal today, upon reading a fascinating interview with a hedge-fund manager (via Chris Young, via Jason Kottke, via snarkmarket, via Points of Note). Arbitrageurs execute many trades automatically — algorithmically. They do this to subtract the human element of fear and panic from their trades: while everyone else is freaking out, goes the story, your algorithm is patiently examining fundamentals and buying securities that the market has underpriced; it’s snapping up 3Com when the market says the stock has a negative value, for instance. Again, this is supposed to lead to greater efficiency: once all the irrationality has been swept out of the market, what will be left are stock prices that reflect the true value of the underlying asset.

But again, this seems to rest on some assumptions about many independent economic actors all moving in an uncoordinated way. And the point of the interview with a hedge-fund manager is that they’re not moving in an uncoordinated way: when you open up the black boxes, all the algorithms are basically behaving the same way, because the people who programmed them all share basically the same assumptions about the world, all got their MBAs from the University of Chicago or CMU or Harvard, etc.

So then one quick question is: how has market volatility changed over time? Have robot traders moving in lockstep made the market more volatile? Doesn’t the Efficient Market Hypothesis (our story about the disappearing arbitrage opportunities, above) imply something about decreasing volatility? If the market’s ruthless discipline is always forcing the prices of assets to their true values, wouldn’t you expect those true values not to change very rapidly? My suspicion is that a truly rational market wouldn’t be a very volatile one. I recall reading somewhere (probably in “Market force, ecology, and evolution”) that price shifts are quantifiably far more volatile than the new information in the market would justify.

There’s a curious algorithmic question around the black boxes, too: why weren’t the various interacting black boxes prepared for the possibility of other black boxes moving in the same way? The study of how to behave in the presence of other agents who are pursuing their own ends is called game theory, and some of the world’s smartest people have been working on it for at least 64 years. Or were the black boxes indeed modeling the other agents, just very poorly? Economics is like any other mathematics-based discipline: garbage in, garbage out. If you don’t model others properly, don’t blame the math when you fail to predict how they’ll act.

I’d love to dig into those black boxes. One wonders if this is another case of people designing exceedingly complex models when something simpler but more robust in the face of an uncertain world would have done a better job.

I’m hoping to get some clarity about the broader market by understanding the subprime collapse; I think it’ll be very instructive. The collapse seems to have laid bare how a lot of the parts interact: government regulators, hedge-fund traders, the auditors who are supposed to be watching over us while we sleep, the ostensible rationality of the actors, and the ostensible super-rationality of the computers.