Integral
Can't you hear it?
Oh, 'tis a great story, with heroes and villains, tales of bravery and disgrace.
Well, maybe not quite.
Start in the 1960s. You have these wonderful large-scale quantitative models used for policy evaluation.
Then Lucas comes along in '73 and digs into those models. He realizes that they have a particular feature: people can be fooled by the government, constantly, consistently, year-in and year-out. If the central bank says "we're gonna increase the money supply 10% this year, just like every other year," people believe it -- even if the historical record is only, say, 5% money growth per year.
So Lucas writes down an alternative: expectations about the future have to be consistent with the model itself. People are "asymptotically correct" -- everyone knows the true model of the economy and bases their forecasts on that model, at least in the very long-run. People base their forecasts on what the government's actually done rather than their proclamations, and the government has to build up credibility for its proclamations to be anything other than cheap talk.
And I do think that market participants are rational, in the long-run, with enough information and time to learn about the economy. Maybe not super-rational in the sense of "knowing the true model" (hell, nobody knows the true model, that's why we still have economists!) but, you know, close enough.
The problem comes when trying to use the long-run rationality of expectations in a short-run context. In the short run everyone (especially in financial markets) is trying to make a buck, and people can be fooled, and you can find a sucker if you look hard enough. Trouble is when the other guy thinks you're the sucker....
One of the research directions I'm going in explores "rational learning", where market participants know *basically* what the economy looks like but learn more about it over time. There you can get bubbles and short-run fluctuations until everyone settles down and learns the long-run equilibrium.
Well, maybe not quite.
Start in the 1960s. You have these wonderful large-scale quantitative models used for policy evaluation.
Then Lucas comes along in '73 and digs into those models. He realizes that they have a particular feature: people can be fooled by the government, constantly, consistently, year-in and year-out. If the central bank says "we're gonna increase the money supply 10% this year, just like every other year," people believe it -- even if the historical record is only, say, 5% money growth per year.
So Lucas writes down an alternative: expectations about the future have to be consistent with the model itself. People are "asymptotically correct" -- everyone knows the true model of the economy and bases their forecasts on that model, at least in the very long-run. People base their forecasts on what the government's actually done rather than their proclamations, and the government has to build up credibility for its proclamations to be anything other than cheap talk.
And I do think that market participants are rational, in the long-run, with enough information and time to learn about the economy. Maybe not super-rational in the sense of "knowing the true model" (hell, nobody knows the true model, that's why we still have economists!) but, you know, close enough.
The problem comes when trying to use the long-run rationality of expectations in a short-run context. In the short run everyone (especially in financial markets) is trying to make a buck, and people can be fooled, and you can find a sucker if you look hard enough. Trouble is when the other guy thinks you're the sucker....
One of the research directions I'm going in explores "rational learning", where market participants know *basically* what the economy looks like but learn more about it over time. There you can get bubbles and short-run fluctuations until everyone settles down and learns the long-run equilibrium.