Friday, June 19, 2009

The Trouble of Rationality

. Friday, June 19, 2009

Yglesias has been reading the new Justin Fox book and mulling over the Efficient Markets Hypothesis:

The “weak” EMH says that stock market motions are unpredictable and a person without access to private information can’t “beat the market” in a consistent way. The “strong” EMH says that stock market prices are in fact correct and, therefore, that liquid financial markets allocate capital perfectly. ...

The strong EMH certainly doesn’t seem true—market history is full of crashes and bubbles. But the EMH is a nice, solid totalizing theory that can serve as the basis of a research program. By contrast, while you can poke tons of holes in strong EMH, you can’t really produce an alternative totalizing theory. Such a theory would, after all, violate weak EMH. And if you had such a theory, you would use it to make money and in doing so restore the market to equilibrium. But then you circle back ’round to the fact that the market exhibits all these huge irrational swings.

I'm going to shoot from the hip a little bit here. Why can't anybody produce an "alternative totalizing theory" to EMH? Because to do so you have to assume that investors are irrational. Which, judging by the title (The Myth of the Rational Market), is what what Justin Fox does (N.B.: I haven't read the book so this may not be true). And a lot of behavioral economic work has found that investors actually are irrational, at least in the Homo Economicus sense of the term. Contrary to popular belief, these views have not been heterodox for a very long time, if they ever were. As Yglesias notes, chapter 12 of Keynes' General Theory basically spells this out. Alan Greenspan is certainly not considered anti-market, but his most famous saying is about "irrational exuberance".

So okay: why don't we just model economic actors as irrational, then? Because we can't, at least if by "irrational" we mean erratic and unpredictable. If peoples' behaviors are random, then what are we trying to model? This is clearly a false view of the world: people often make mistakes, but they broadly act in their self-interest as they understand it. And when people do deviate from textbook rationality, they tend to do so in systematic ways, which implies that they are trying to make rational decisions but face informational and emotional constraints.

For example, the "disposition effect" shows that investors systematically sell assets that have increased in value and hold assets that have lost value. This an example of a larger "irrationality" whereby people are risk-averse toward potential losses and risk-loving toward potential gains. Such behavior does violate the assumption that agents maximize utility -- built into much economics and political science -- but only if risk-neutrality is part of that assumption. If it is not, if we accept that actors place greater weight on the status quo than any other point on the utility continuum, then it becomes much easier to interpret investor behavior. If we assume that the first instinct of humans is to preserve what we have, and the second instinct is to be risk-averse towards losses and risk-loving towards gains, then manias and panics become perfectly natural side-effects of normal human behavior. Bubbles and bank runs should be expected. Moreover, this is perfectly consistent with a "weak" version of EMH.

This begs the question: if the status quo bias and endowment effect are common and predictable, perhaps even instinctual in an evolutionary sense, then perhaps we should refine our definition of rationality to reflect that fact. Doing so would require adding some uncertainty to our models and accepting less precision, but it would also prevent us from making extraordinary mistakes like accepting "strong" EMH as gospel truth.

In fact, social scientists have done something similar to that: "bounded" rationality allows individuals to possess imperfect information and face other constraints, but to then make decisions as rationally as they can under those constraints. Most cutting-edge research uses some version of bounded rationality rather than perfect rationality, and such models are consistent with "weak" EMH. This isn't perfect -- even if you know that risk-neutrality is rational you may not be capable of forcing yourself to act that way 100% the time -- but it gets us closer to reality.

All to say that we don't have to reinvent the wheel to understand things like financial crises or the political responses to them. We have the framework at our disposal. There is room for improvement, naturally, but we aren't flying blind.


The Trouble of Rationality




Add to Technorati Favorites