Thursday, February 2, 2012

Negative stimulus, 1946

I ran across a fascinating article, "A Post-Mortem on Transition Predictions of National Product,"  in the 1946 Journal of Political Economy, by Lawrence Klein. Klein, who would go on to create the main macroeconomic forecasting models and a Nobel Prize, was  confronting one of the first great failures of Keynesian economics:

We all recall clearly the headlines in last Autumn's press, declaring that "Government economists predict 8 million unemployed by 1946." ...We now find ourselves in the first half of 1946 with about three million unemployed and facing one of the greatest inflationary pressures that we have ever experienced. The economists who were warning us of a deflationary danger during the early months of the postwar transition period should have been stressing precisely opposite economic policy
Yes indeed. Government spending, or "stimulus" in the modern lingo, was certainly going to fall like a stone at the end of the war. 8.5 million young men would be dumped on the labor market. With a Keynesian mindset the forecast was obvious: we're going right back to the depression. Instead, we got sharp inflation and a boom.Where did they go wrong? This event has been covered in more depth, but let's keep reading Klein's analysis...
The customary model for prediction can be called the simplest Keynesian model...The model can be written as GNP=C(GNP)+I, where GNP = gross antional product, the function C(GNP) is the consumption schedule, and I = autonomous investment....

It is immediately obvious where this forecast failed-in the prediction of consumer expenditures,...The order of magnitude of the error involved is great, and, what is more serious, it is great enough to lead to disastrous policy recommendations. The predicted GNP of $164.5 billion should call for an inflationary policy, but this is just the opposite of the policy that was needed...

Most critics will agree that the consumption function is incorrect...At least as important as the statistical errors, however, are the errors of economic theory...[the consumption function].may be incorrect from an economic theoretical point of view and from a statistical point of view.
And after this poor Klein dithers around with statistics, the special effects of  "bottlenecks," wartime, liquidity and so on. With the benfit of hindsight you can see him circling around the elephant in the room, the permanent income theory of consumption.  No, returning soldiers did not follow some "schedule" relating this year's consumption to this year's income. He's so close, he can smell the elephant, but he never quite touches it. That would wait 15 years and Milton Friedman's permanent income, to be followed by Lucas, Sargent and the others who tore apart this Keynesian castle.

Why are we reading this? It's fun to go back and see how really smart people understood things at the time. Maybe it should give us some humility -- so much policy debate seems based on the idea that we know everything so well. If we understood things as well as we now see Klein understood things, would we still want to spend trillions on our best guesses? Klein is happy that the Government didn't follow the widely-recommended "inflationary" or stimulative policy advocated in 1945! If you were transported back to to 1790 and got sick, would you want to be treated by the best doctors of the day?

On the other hand, many of Tom Sargent's writings suggest our ancestors understood nuances of policy even better than we do -- or at least before the brief interlude in which  Keynesians forgot everythng their ancestors knew. (Start with Tom's Nobel Prize speech. The point is not that hard to see between the lines.)

As a research economists, it's a bit frightening. There are surely such elephants in our room that we're missing, and we will kick ourselves for not really noticing them. Try not to get so close and miss the big point!

I got to Klein on a different mission. 1946 is a great data point to understand -- the great negative stimulus which coincided with a boom and inflation. I also got to this paper while digging in to the long set of historical writings that found budget constraints missing in Keynesian models. This is a big point in my writings about stimulus, but I'm not the first to notice that. More later...

Notice a Journal of Political Economy article published in August 1946 about how summer 1945 forecasts for 1946 turned out!   That could never happen now, in any academic journal.

Update: David Henderson and Russ Roberts on the postwar depression that wasn't, with good Samuelson quotes.