Economic Models and Economic Predictions

by Paul Krugman

Noah Smith has a further entry in the ongoing discussion of microfoundations and all that; I have some minor quibbles, but right now I just want to pursue a train of thought I’ve been having about the role of predictions in economic doctrine. Now, I don’t mean predictions of the type “third-quarter GDP will rise at an annual rate of 1.7 percent”; those things depend on a lot of detail, and a keen eye for the relevant details. What I mean instead are predictions of the effects of big shocks — and specifically cases in which different approaches make very different predictions. These “so-gravity-does-bend-light” moments are what can shift economic doctrine, and rightly so.

The initial triumph of Keynesianism had a lot to do with the fact that big increases in government spending as war approached did indeed lead to higher output and employment; so much for the Treasury View. Wartime spending is still the best evidence for fiscal effectiveness.

What happened in the 1960s and 70s was that economists who thought about microfoundations — who tried to understand price stickiness in terms of more or less rational behavior — made a big prediction: that the apparent tradeoff between unemployment and inflation would break down in the face of persistently higher inflation, and worsen. This prediction was right, and that had a big impact. I remember lunchroom discussions when I was in graduate school in the mid-70s, when the emergent freshwater school was being discussed; some of my classmates would say, with some worry, that those people had been right so far, so might they not be right now?

But they weren’t; there were no big further predictive successes, and in fact the extremely costly disinflation of the 1980s, which was more or less what modified Keynesianism predicted, was very much at odds with the freshwater view of the world.

So as I see it, the whole microfoundations crusade is based on one predictive success some 35 years ago; there have been no significant payoffs since.

And something else to consider: the events of the past few years, in which a tripling of the monetary base has failed to set off high inflation, and huge deficits have failed to send interest rates soaring, represent a predictive success for IS-LM-type macro fully comparable to the way stagflation represented a success for macro with thinking about microfoundations. It should be accompanied by a comparable rearrangement of doctrine.

The New York Times