Arnold Kling  

The Austerity of 1945-1947

PRINT
Drugs, Death, Censorship, and ... Lyrics Without Music...

Official data are somewhat sparse. I would appreciate pointers to any better data or additional information. I am also curious about what newspapers and magazines were reporting at the time about postwar conversion--how well it was going, what was working, what was not.

First, some characteristics of the civilian working-age population (then defined as age 14 and up), in thousands. Source: Bureau of Labor Statistics

category19451947
population94090106018
labor force5386060168
employed5282057812

If we divide the increase in employment by the increase in the size of the civilian labor force, the result is that at the margin 79 percent of the new entrants to the civilian force were employed. Ordinarily, I would say that is not good performance, but considering the size of the jump in the civilian labor force, it is not bad at all.

Next, I look at GDP. I prefer to use nominal GDP, because it allows me to put personal income and taxes in the same table. Source is the Department of Commerce. Figures are in billions.

Category19451947
GDP223.0244.1
Consumption120.0162.0
Investment10.835.0
Net Exports-0.810.8
Government Purchases93.036.3
Personal Income171.6190.9
Private wages and salaries82.6105.6
Government wages and salaries34.917.5
Transfer Payments5.610.8
Personal Income Taxes19.419.8
Disposable Personal Income152.2171.1

From 1945 to 1947, government spending declined by $56.7 billion, or about 25 percent of 1945 GDP. If you use Keynesian multiplier analysis, then you would expect to see a pretty sizable drop in GDP. Yet we did not observe that.

It turns out that in the postwar economy, the private sector found jobs for a huge number of people entering the labor force even as government spending plummeted. I am not suggesting that the same circumstances apply today. For one thing, there was a lot of pent-up demand after the war because of wartime rationing. But my point is that the laws of economics do not dictate that any decline in government spending must result in a contraction of the economy. Although the sort of macroeconometric model used by the CBO or private forecasting firms would say so.

[UPDATE: I should have pointed out that yesterday I participated in a conference in which David Henderson, by phone, suggested looking into this time period. In my vague plans to do a book on the evolution of ideas in macro, I was going to include this episode, because so many prominent Keynesians predicted a return to bad times. I'm actually wondering if any economists foresaw what was in fact quite a boom within the private sector.

I think this episode at the very least points out the difficulty of making empirical estimates in macro. It would seem to be an important episode because it is a "natural experiment" in the sense that the drop in government spending occurred for reasons having nothing to do with the economy, and it was a huge drop. That argues giving it a higher weight than other time periods in estimating the multiplier. On the other hand, other factors, including the fact that we started from full employment and that wartime rationing was lifted at the same time, suggest giving it a smaller weight. You will find that these same sorts of ambiguities permeate macroeconomic history, which is the main reason I am not willing to sign on to specific estimates for multipliers.]


Comments and Sharing





COMMENTS (10 to date)
Daniel Kuehn writes:

I'm having a tough time squaring this:

"If you use Keynesian multiplier analysis, then you would expect to see a pretty sizable drop in GDP. Yet we did not observe that."

with this:

"I am not suggesting that the same circumstances apply today. For one thing, there was a lot of pent-up demand after the war because of wartime rationing. But my point is that the laws of economics do not dictate that any decline in government spending must result in a contraction of the economy. Although the sort of macroeconometric model used by the CBO or private forecasting firms would say so."

Am I correct in understanding that you think that people who think in terms of multipliers think that the multiplier is the same in 2009 and 1999, or 1929 and 1949, for that matter? It's clear you don't think that. I'm having a tough time understanding why you think other people think that.

Do they? Maybe they're all dumber than I thought, but I never thought that was the argument or the assumption.

Daniel Kuehn writes:

It's really tough to follow the latest multiplier critique. The other day Russ Roberts was saying that it jumps around so much it's just an ex post excuse for spending. Now you're saying that it doesn't move enough.

Somebody has to be mischaracterizing how Keynesians use multipliers here.

My understanding is it's higher when demand is depressed and no greater than one - probably lower given bureaucratic inefficiency - when demand is not depressed. That seems to explain 1920-21, 1929-41, 1945-47, and 2008-2010 just fine to me. All these mystery inconsistencies really aren't that mysterious or damning.

Daniel Klein writes:

Higgs JEH 1999:

ABSTRACT: The orthodox view of the U.S. reconversion after World War II relies on unacceptable GDP figures for the wartime economy and misinterprets the low level of unemployment during the war. For the postwar transition, the emphasis on consumer demand financed by drawing down liquid assets accumulated during the war is inconsistent with the facts. The success of the transition depended on the reestablishment of “regime certainty,” which in turn depended on diminishing the influence of the more zealous New Dealers. Wartime and postwar political developments created sufficient regime certainty for the postwar market system to generate genuine prosperity.

The piece is a chapter in his later OUP book.

MD writes:

Daniel:
"Am I correct in understanding that you think that people who think in terms of multipliers think that the multiplier is the same in 2009 and 1999, or 1929 and 1949, for that matter? It's clear you don't think that. I'm having a tough time understanding why you think other people think that."

Daniel, I could be mistaken about what Arnold's saying of course, but I believe his point is not about the magnitude of the multiplier, but rather the sign, which, according to Keynesian reasoning under the conditions Arnold describes, should not be negative. That is, if we consider the drop in G in the above figures alone, the effect on NGDP is supposed, according to typical Keynesian logic, to be negative [dGx(1/1-MPC)=dNGDP, where dG0]. His comment that "I am not suggesting that the same circumstances apply today" is likely merely an acknowledgement that ceteris non paribus est. Further, his point seems to be that the data indicate that there are forces guiding economic recovery that are of greater significance than the multiplier effect.

Now, Arnold is merely hinting at something here, as far as explaining what those forces are (though regular readers of this blog know he has a particular theory), so there is some ambiguity. However, his rhetorical point is pretty well on target. According to straight-up Keynesianism, the drop in G is in itself an unwise policy decision if one wishes to expand the economy -- it should have led to a drop in NGDP. Typical Keynesian logic only advocates lowering G when inflation is getting out of control and the economy is near "full employment". There are no data here on inflation, but it's likely Arnold brought up the employment figures precisely to establish that this condition didn't seem to apply. Hence, his point that government austerity is not necessarily an unwise policy.

Daniel Kuehn writes:

Daniel, I could be mistaken about what Arnold's saying of course, but I believe his point is not about the magnitude of the multiplier, but rather the sign, which, according to Keynesian reasoning under the conditions Arnold describes, should not be negative.

I don't know if this is what Arnold is saying - but I think this point of yours confuses the juxtaposition of the change in two variables with a mutliplier - which is a change in a variable relative to a counter-factual, holding all else constant.

You can't just look at changes in a data point and decide "I'm going to call that a multiplier" if you don't have some sort of counter-factual you're working off of. Presumably, all of us agree that aggregate demand was very strong during 1945-47? If we all take that as a baseline understanding of the situation, how is a negative multiplier demonstrated by the data? If we assumed there would be no change in GDP or any of Arnold's other figures, and we saw a decline in government spending accompanied by an increase in GDP, then you could call that a negative multiplier. But in the context of assumed strong demand completely unrelated to government spending, I'm not sure how you can just jump to the conclusion that this implies a negative multiplier at all.

Why did Robert Barro bend over backwards to try and figure out a convincing identification strategy to get his predictably low multipliers for most of the 20th century, if all he had to do was just compare datapoints?!?!

According to straight-up Keynesianism, the drop in G is in itself an unwise policy decision if one wishes to expand the economy -- it should have led to a drop in NGDP. Typical Keynesian logic only advocates lowering G when inflation is getting out of control and the economy is near "full employment".

Right - but my whole point is that this "straight up Keynesianism" is something of a misrepresentation of Keynesians, and probably more descriptive of what some politicians may be claiming. I understand who Arnold has in his sites. I applaud taking aim at those perspectives. What I'm concerned about is tying the very concept of the multiplier to them.

B.B. writes:

I wish some economist would turn "pent up demand" from an intuitive concept to an operational and meaningful concept rooted in utility and profit maximization and general equilibrium.

I do grant that wartime rationing and price-wage controls do create a special type of pent-up demand. There was demand repression during the war.

I also checked some curious numbers. In 1946, real GDP fell 11%, but nominal GDP was essentially flat. The GDP price index rose 11.5%

In 1947, real GDP fell 1%, but nominal GDP rose 10%. The GDP price idex rose 11%.

I think there were a lot of distortions to data from the end of rationing and controls. I do not trust the real numbers. Measured prices rose a lot, but actual prices may not have.

Adam writes:

Great point, Arnold. You need to write it up as an academic paper.

Dave writes:

Krugman seems to think that the rationing aspect is why this example is not relevant: see here and here. It certainly makes for a difficult comparison.

Try examining data on Japan for evidence against Keynesian stimulus ideas.

david writes:

Besides the point about rationing, there's also the restoration of global trade under Bretton Woods and technological developments following the war and the US being the only undamaged industrial nation and thus enjoying massive world demand for industrial goods.

Ted writes:

I guess one could point out output decline over 12% in 1945 ...

But, anyway, New Keynesian analysis doesn't say the fiscal consolidation necessarily leads to a "recession." It leads to a fall in output, relative to what it would have been, which is basically true by definition since a component of output is government consumption. That however doesn't mean a massive recession ensues. It just means output and employment are lower than they otherwise would have been because a big component of aggregate demand (the government) has exited the buying realm. The logic of now is that in depressed times where AD is already incredibly low it would be misguided to start consolidation since it would only add more damage to a hurt economy. Of course, this may not be true either because you are wrong that NK analysis even requires any damage to the economy from austerity. That's just one possible outcome. However, let's turn the story on its head. Let's say our government announced and passed massive cuts to social security, medicare, and defense. What would happen? Well, instead of generating a recession it would be slightly expansionary in the short-run, and very much expansionary in the long-run. If the cuts to spending are viewed as credible and long-term commitments to reduction government spending as a percentage of GDP, relative to what it would have been, then in expectations of a future lower tax liability consumers would revise estimates of their permanent income upwards which would increase both future AND current consumption. In the short-run, the Fed would largely neutralize it's expansionary effects (except right now it would probably be very expansionary because the Fed isn't doing their job and would allow the expansion I suspect), and in the long-run it would obviously be expansionary.

So, even in the NK model it isn't clear that austerity has to cause recessions, it all depends. Though short-term gimmick austerity like 50% discretionary spending cuts for 5 years or something stupid would likely just hurt the economy now, but credible, long-term reforms would likely be highly expansionary.

Comments for this entry have been closed
Return to top