Wednesday, March 11, 2009

The Problem with Rational Expectations in Macro

I'm not much of a macroeconomist, which is why I haven't had much to say about the current financial crisis and recession. But I did have to take the standard macro courses in grad school, and I remember having one major methodological objection to the rational-expectations models we were learning at the time. Arnold Kling nails it:
Economist X has one model of the economy. Economist Y has another model of the economy. In X's model, people believe in X's theory. In Y's model, people believe in Y's theory. It is logically impossible for economist X and economist Y to inhabit the same universe! Yet they do. This tells me that the axiom of rational expectations is too strong.
Exactly. But notice that this is not a criticism of the rationality assumption. Nor is it a criticism of incorporating expectations into an economic model. Neither component of the label "rational expectations" was the problem. The real problem was the additional assumption, rarely stated explicitly but applied repeatedly when solving the models mathematically, that all economic actors shared the economist's model of the world. And that's manifestly false if even economists don't all share the same model.

I sometimes wonder (idly, with no intention of doing the research to find out) if any macroeconomist has constructed a model in which the modeled agents make rational decisions based on assumptions that differ from those of the model.


Anonymous said...

See "Robustness" by Sargent and Hanson

Anonymous said...


It's unfortunate that all the non-Keynesian schools really have is this theory. They don't even teach rational expectations here, and my professor is, without a doubt, a Keynesian in every sense of the word.

The problem in macro is that the ad-infinitum of Keynes' arguments goes straight to central planning, because as long as k > 0, government spending is "superior" to tax cuts as a means of raising output.

We will likely have a whole lot of data to refute Keynes after this "crisis" is over, but until then, government is squandering trillions of dollars on failed programs to raise employment and make people happy.

Tim said...

Thanks for the inspiration, Glen.

August said...

It's so absurd I don't really believe it. No one runs around trying to get me to believe in a 'gravitational model.' If people have to believe in the damn model, then then it can't possibly explain reality.
So these guys must know they are magicians, not scientists.

And it seems they are succeeding at convincing large numbers of people that economics isn't a science due to their chicanery

Kevin Scobey said...

Another problem with Rational Expectations, as it relates to the current financial crisis, is something that's probably quite obvious to any good trader or financial investor: actors in financial markets aren't rational at all. Whereas in a goods market, consumers may see orange juice on sale, and stock up, expecting prices to go back up; in financial markets, if stocks "go on sale," it just makes people want to sell even more. If stocks go through the roof, people want to buy even more (generally speaking).

Even though financial actors appear to be quite irrational, the general behavior of financial markets overall, is actually predictable and quite explainable. It's just that you can't fit financial markets neatly into economic theories that have their roots in explaining goods markets. This is why the "new" science of Socionomics came about, which is based on the Wave Principle. (

Jeff Brown said...

Hi Glen,

There's a strand of literature involving models of bubbles based on overconfidence, started by Jose Scheinkman and Wei Xiong, in which some agents put more confidence in the meaning of a signal than others do, and prices as a result become "wrong".

Other dude:

> If people have to believe in the damn
> model, then then it can't possibly
> explain reality.

Economists don't ask you to believe their models. The purpose of a model is not to convince the reader that no other factors are relevant, but to determine what sort of effects might arise out of a particular, tractable subset of the characteristics of reality. When other details matter too, we argue over which details to worry about. It might not feel satisfactory but it's the best we can hope for. We could sit back and mope that everyone's irrational and prediction is hopeless, but enough people are already doing that.

Greg Mankiw has made this point much more elegantly than I can. Somewhere. His principles book, probably.

Anonymous said...

It's not clear what 'rational expectations' would even mean aside from 'subjective expectations'. Only individuals have rationality!