Very bad economic theory (historiography)
By Mike Kimel
Cross posted at the Presimetrics blog.
One of the enduring myths about the US economy, which like most such myths is used to support a lot of very bad economic theory, is the notion that there was some sort of big economic boom at the end of World War 2. Case in point: Tyler Cowen links to a working paper called The U.S. Postwar Miracle by David R. Henderson. Henderson also mentions the paper in a blog post of his own.
The download page for the paper includes a few paragraphs from the introduction which are reproduced below:
We often hear that big cuts in government spending over a short time are a bad idea. The case against big cuts, typically made by Keynesian economists, is twofold. First, large cuts in government spending, with no offsetting tax cuts, would lead to a large drop in aggregate demand for goods and services, thus causing a recession or even a depression. Second, with a major shift in demand (fewer government goods and services and more private ones), the economy will experience a wrenching readjustment, during which people will be unemployed and the economy will slow.
Yet, this scenario has already occurred in the United States, and the result was an astonishing boom. In the four years from peak World War II spending in 1944 to 1948, the U.S. government cut spending by $72 billion—a 75-percent reduction. It brought federal spending down from a peak of 44 percent of gross national product (GNP) in 1944 to only 8.9 percent in 1948, a drop of over 35 percentage points of GNP.
Update: Dan here…see spencer’s post as well.
While government spending fell like a stone, federal tax revenues fell only a little, from a peak of $44.4 billion in 1945 to $39.7 billion in 1947 and $41.4 billion in 1948. In other words, from peak to trough, tax revenues fell by only $4.7 billion, or 10.6 percent. Yet, the economy boomed. The unemployment rate, which was artificially low at the end of the war because many millions of workers had been drafted into the U.S. armed services, did increase. But during the years from 1945 to 1948, it reached its peak at only 3.9 percent in 1946, and, for the months from September 1945 to December 1948, the average unemployment rate was only 3.5 percent.
Ask people who lived through that period as young adults what economic conditions were like, and you will inevitably get the answer that they experienced an economic boom. The U.S. economy during the post-World War II years is exhibit A against the Keynesian view that economies will necessarily suffer high unemployment and slow growth when governments make big cuts in government spending. Why did the U.S. economy do so well in the years following World War II given how badly it had done in the years preceding America’s entry into the war? The answer, in a nutshell, is that dramatically reducing government spending and deregulating an economy can take that economy from sickness to health. In short, one of the main things a government can do to help a weak economy recover is to step aside.
The paper goes on to provide more supporting material, etc., but essentially, Henderson brings up the big postwar economic boom, a boom so impressive he terms it a miracle, and then goes on to point out its proof that a big cut in government spending over a short period of time won’t cause an economic mess.
The following graph shows real GDP per capita (data from the BEA’s NIPA table 7.1) beginning in 1944:
Now, recall, World War 2 ended in 1945, at which point spending dropped a lot. But what happened to the economy? As the graph shows, what Henderson refers to as a boom or a miracle actually looks like a big decrease in GDP per capita. Furthermore, it doesn’t create the conditions for rapid growth either. The 1944 peak wasn’t reached again until 1953, and it didn’t stay permanently above that level until 1955, though it threatened to fall below that level as late as 1958!
Moving on, in the last paragraph I quoted from the working paper, Henderson compares the postwar economy favorably to the pre-war economy. We can compare the period from 1944 to 1960 to the period from 1929 to 1940 (beginning of the Great Recession to a year before the US’ entry into the war) by indexing. The following graph shows the percentage change in real GDP per capita from the 1944 peak for every year from 1944 to 1960. Similarly, it shows percentage change in real GDP per capita from 1929 to each year until 1940.
Notice… the economic collapse during the Great Depression was much sharper – almost twice as deep as the one that followed the end of World War 2. However, the recovery following the 1929 collapse was also a lot faster, and it began precisely (drum-roll) when the New Deal began. Whereas it took about fifteen years for the economy to move appreciably beyond where it had been in 1944, the 1929 peak was surpassed ten years later, and the economy by then was growing extremely quickly. By 1939 and 1940, real GDP per capita was expanding at over 7% a year. In fact, real GDP per capita growth rates exceeded 7% in five of FDR’s first eight years in office, and yet, the economy was poised to grow even faster during WW2. For comparison, since GDP started getting reported, there has never been a single year outside of FDR’s presidency when real GDP per capita growth rates reached 7% a year.
None of this should not be a surprise to anyone who read my book, or regular readers of this blog. After all, I had a post some time back looking at the recovery from every recession beginning in 1929, broken out by whether the government had decreased spending, increased spending a little, or increased it a lot. There weren’t that many recessions when the government cut spending, but in those where it did expansions were slower and shorter than when the government increased spending. The relationship between expansions and tax cuts is more clearcut; tax hikes were associated with faster, more durable expansions than tax cuts.
I would also note that Henderson also makes a number of comments about how the unemployment rate was very low during the immediate postwar period. I’ve noted this in posts before: Bureau of Labor Statistics figures begin in 1948. Any unemployment figures from before that time are estimates produced later, and usually using a different methodology or assumptions than those produced today, making them not entirely comparable to figures from 1948 on. But even ignoring that, people in school, in the military, and folks not looking for a job are not considered unemployed. The GI Bill put a number of people in school. As to military enrollment, according to Table 248 of the 1951 Statistical Abstract of the United States, it fell from 12 million in 1945 to 3 million in 1946, and then stayed around 1.5 million for the rest of the decade. By contrast, in the 1930s, the number of personnel in the military rose from about a quarter million in 1930 to 335,000 in 1939. Put another way, by putting people in school or keeping them in uniform, government government spending forced unemployment rates down by simply by keeping a lot of people out of the labor force. (Note that the estimates for unemployment during the 1930s actually go the other way – for reasons that are not obvious, people who worked for the various New Deal programs are often counted as unemployed.)
Finally, it is worth noting – some of the commentators to Tyler Cowen’s post also seemed to incorrectly believe that there was a post WW2 boom, though they tended to attribute that non-existent boom to the fact that the US came out of WW2 intact and went out building up other participants of the war. The fact that there are a variety of incorrect views about what happened in the past is not important. The fact that people believe in things that are demonstrably (and easily demonstrable, at that) not true is vital and unfortunate. As Michael Kanell and I point out in Presimetrics, theorizing based on incorrect facts leads to very poor theory, poor theory leads to abysmal policies, and abysmal policies lead to very unfortunate outcomes that negatively impact the lives of all of us.
—
Added on November 7, 7 AM:
Henderson notes in comments that his paper provides two reasons why GNP data (which he uses, or GDP data for that matter) was wrong during the World War 2 years, or at least not comparable with the post-War era. I had meant to mention them but forgot. Apologies.
Henderson’s reason 1: An “index-number problem.” As Henderson puts it:
When price controls were removed after the war, prices shot up. Therefore, the prices used to convert nominal GNP into real GNP made real GNP look lower than it actually was.
Henderson’s reason 2:
Second, the GNP and GDP data, which are supposed to measure the value of production, instead measure government spending on goods and services at their cost—that is, at the price the government paid for them. But we no have no idea what the value of all those goods and services bought by the government during the war was worth. So we can’t compare GNP during the war with GNP after.
Both of these issues are dealt with by something made explicit in the post – real GDP per capita was essentially flat from 1947, the year the economy bottomed out according to real GDP per capita figures, to 1949. And yet, price controls were lifted in 1946 and rationing was long gone. The production of battleships and tanks and whatnot had also run into a wall by then, in deference to the production of nylons, automobiles and meat.
Additionally, notice the respective trajectories of the two curves in Figure 2 after each bottoms. Notice that when growth finally began in the post-War period, it was much slower than the growth in the New Deal era. Additionally, it may not be obvious from the graph, but after 1938 growth was accelarating!. As noted above, real GDP per capita grew at more than 7% a year in 1939 and 1940 (!!!) , and at a rate of 15.9% in 1941. Realistically, we cannot attribute 1941 figures to World War 2 – the US only entered the war, woefully unprepared, on December 7th, after all. Additionally, 1941 growth, though lower than 1942, exceeded those of 1943, 1944 and 1945, when the war effort really took off. If the war was really the cause of rapid growth, 1941 would not have been that impressive, and certainly it would have been less impressive than 1943 and 1944.
But, let us ignore the fact that the economy really took off after 1938, and assume that the growth rate from 1932 to 1938 (the one dip during the New Deal era) had prevailed from 1938 to 1947. Since the annualized growth rate from 1932 to 1938 was (an extremely impressive) 4.7%, real GDP per capita would have been 31% higher in 1947 than in 1941. I bring that up because Henderson has this paragraph:
As noted, there is no good way to compare output after the war with output during the war because about 40 percent of wartime output was not sold in a market. But it does make sense to compare postwar output with prewar output. We can take 1941 as the last year before America’s official entry into the war because the United States did not enter the war until December 8, 1941. In that year, real GNP (in 1964 dollars) was $287.1 billion. In 1946, the first full year after the war ended, real GNP was $337.9 billion, and in 1947, it was $336.8 billion (both in 1964 dollars). Thus, real GNP in the first two transition years after the war was more than 17 percent higher than before the war.
If you use real GDP per capita rather than real GNP, the difference between 1941 and 1947 figures is 20%. But… that 20% difference is only impressive if you start from the assumption that growth would have been mediocre without the war. That might be a reasonable assumption if you’re a libertarian or a conservative and have bought into the prevailing mythology about the New Deal era, but as I noted just two paragraphs up, simply extending the pattern from 1932 to 1938 would have resulted in a much bigger change between 1941 and 1947 than the one that was actually observed. (i.e., 31% as opposed to 20%) Thus, Henderson may not realize it, but he is inadvertently pointing out that at some point between 1938 and 1947, economic growth really decreased by a lot.
So when did that happen and how big was it? Well, we know it didn’t happen in 1939, 1940 or 1941 (growth rates exceeding 7% for the first two years and almost 16% in 1941). We know it didn’t happen from 1942 to 1944. Which leaves…. precisely the time period where the official data shows a big drop – the end of the war and the immediate war period. Additionally, since every year from 1939 to 1944 had a growth rate significantly in excess of the 1932 – 1938 growth rate that would imply 1947 real GDP per capita being 50% higher than 1941 real GDP per capita, a simple slow-down in economic growth won’t make the math add up – you need some fairly big negative numbers… and that again is precisely what the data shows.
—
Added on November 7, 9 AM:
Henderson also seems to implicitly assume that just because items the government spending goes into GDP at cost, as opposed to at some market price, it must imply that GDP was lower than reported. But if anything, it might indicate the exact opposite. How much would the market have valued another B-17 or rifle, without knowing whether that marginal unit was the difference between losing the war to the Germans and Japanese or not? In fact, many Americans were willing to give up huge amounts of their lives to the war effort. My own grandfather was too old to be drafted, and with two children, would have been exempt from the draft if he wasn’t too old, yet he gave up his business and took a massive paycut to tour Europe as an enlisted man from the driver’s seat of a Sherman tank. And clearly, my grandfather wasn’t alone out there. I would argue if anything, given the alternative to victory was the sort of occupation viewed in Europe and Asia, and given what Americans gave up to the war effort, the amount Americans collectively valued what the government was buying was understated by the cost the government paid.
Typo alert 1848 appears once. It should be 1948. I am surprised that Cowen linked to this without denouncing it as nonsense. I thought he was well informed. Has Henderson ever heard about this guy called Robert Barro ? Barro estimated a government spending multiplier of 0.7 mostly using WWII. Barro is not a Keynesian. How does Henderson reconcile his claims with that bit of the literature ?
Now Barro is being silly too, since he assumes that rationing had no effect on private consumption and legal restrictions on investment had no effect on private investment, but at least he got the sign right.
Got it…thanks.
Oh Dan the title should be “Very Bad Economic History” or, for clarity, “Very Bad Economic Historiography.” History and the study of history are not the same, but neither is theory.
On unemployment, the key issue is that employment can be higher than full employment. Full employment doesn’t mean everyone works 24/7 (or 16/7 or whatever is the biological limit). It means people work as much as they want to work. During WWII many people were drafted. Many others worked in partly out of public spirit. Then when the war ended, the desired total number of hours to work fell sharply as, with the war over, the country was much richer. At the micro level, the huge taxes (including the very huge tax called the draft) cause labor supply to be very high due to the huge income effect. Yes I just asserted that high taxes cause high labor supply. It is very clear from long term trends in weekly hours that income effects outweigh substitution effects.
The perception of a boom was not irrational. People in the USA were much better off after Germany, Japan, Italy, Romania and Hungary were defeated. That was exactly as predicted by Roosevelt and Truman. The war was constly. It was also necessary. No one but David Broder argues that war is good, because it causes increased public spending which is good. The perception is based on the huge increase in consumption and leisure.
The problem, however, wasn’t that Roosevelt and Truman were demanding huge stocks of weapons but that the Axis powers kept destroying the weapons.
The open question is whether government purchases crowd out consumption, true leisure or involuntary unemployment. Obvouisly during WWII true leisure was crowded out. Also, while consumption did not fall (or fall much) the huge increase in productive capacity indirectly caused by wartime spending (and by rationing so the huge product wasn’t consumed) made much higher consumption possible after the war.
There’s an awful lot here. It will probably take me days to slog through what Henderson has written, plus all of this.
One caution. GDP/Cap can be misleading, most especially in the decade following WW II.
I was born in December, 1946, and always thought I was at the leading edge of the first wave of the post war baby boom. My first grade class was about 1.75 times the size of that year’s 2nd grade class.
This can really distort GDP/Cap, as I discovered here.
http://jazzbumpa.blogspot.com/2010/10/us-economy-is-dying.html
Population growth was at a far greater rate up to the late 50’s than at any time since.
A growing denominater can distort the ratio, and even cause positive GDP growth to be a negative GDP/cap change.
Cheers!
JzB
Sorry to keep commenting, but I think you should describe what happened to total employment (not just civilian employment) during the period. This takes more than 5 minutes of googling (I checked and I do mean 5 — I punted befoer investing a 6th) since they always want to give civilian employment. The Civilian population jumped up after the war and it wasn’t just the baby boom.
Thanks jazz. . . .this was the question I had for Mike as well. The technical question of GDP/captia and whether it the “capita” is entire population or working age population only.
Mike, with all due respect, I’ve pointed out more than once that you need to consider demographics more frequently in your analyses, and this seems to be another case in point.
Eeek!
Be very, very, careful about using GDP figures from wartime. You know this thing we have with GDP and pollution? The eternal cry that we include the cleaning up of pollution as an addition to GDP but we don’t include the original creation of the pollution as a decrease in GDP?
And yes, this is a problem with GDP as a measure of economic health. For it implies that if we just had another Macondo well blowing out every few weeks then we’d stimulate ourselves into huge, fat, wealth.
Obviously insane: we’d be producing a higher GDP for sure, but that’s only because we don’t include destruction of the environment in our GDP figures (that comes in Net DP and other measures, where we look at depreciation etc….or at least should, although we don’t really include environmental depreciation in national accounting figures yet).
Now, use this same logic to look at a war economy. Yup, we’ve a high GDP because we’re producing lots of things, AS RW notes we’re at well above full employment and so on.
But what are we doing with what we produce? Blowing it up. Sending Marines on involuntary tours of Pacific islands. Making GIs crawl from Normandy to the Rhine. none of these are what we would usually call “adding value”. Indeed, most of them we would call the destruction of value.
It’s a fault of GDP figures: just the way we define what they are. and the end result of this is that you really cannot compare, at least with any certainty, that war time and peace time GDP figures are comparable. There’s just too much error in the way we define GDP for this to really work.
On GDP: there is something like (total wages and other income), there is something called (total production), there is something like (total expenses). GDP is the keystone unifying these three variables. It does describe something in the world ‘out there’. The National Accounts will teach you how much of production went to butter and how much to guns – it’s exactly the same system which you despise which gives this information!
On jobs: 1946 was the wonder year with a job growth (USA) of 8% (never seen before, never since) – though this followed a 6% decline in 1945. It was ‘only’ in 1950 that the 1943 maximum was surpassed in a very significant way (http://www.bls.gov). In Europe, it was only after about 1953 that 1929 levels of GDP per capita were surpassed. That’s about a quarter of a century!
the per capita thing is a nonissue. as noted in one of the additions to the post, growth rates are almost the same whether you use real gdp per cap or simply real gnp as Henderson does. The issue is whether the data is wrong. As noted in the post the data fits the known facts. Henderson, on the other hand gets his facts wrong. That real gnp is twenty percent higher in forty seven than in forty one does not support his case . That the economy was about the same size in forty nine as forty seven, ditto.
Simply put, he is making an arguent based on a libertarian fantasy ratherthan on data.
Just a minor note, the US started mobilizing for war in 1940, even though we only declared war at the end of ’41. Lend-lease, the protective mobilization act, and the escorting of Atlantic convoys, all pre-dated the formal entry of the US into the war, although the costs and economic effects of these efforts are dwarfed by post declaration-of-war efforts.
I suggest we all go back and look at the data I posted on real civilian GDP In WW II I posted here.
http://www.angrybearblog.com/2010/07/ww-ii-civilian-real-gdp.html
From 1933 to 1950 real civilian GDP grew at a 5.6% growth rate even though it fell -18.4% and -14.8% in
1943 and 1944 — a decline about the same as the drop in 1929-33.
The analysis that the economy boomed after WW II by Henderson is really bad analysis, largely because it omits so much information, some of which Mike has pointed out. But mainly it omits the information on “forced savings” during WW II that provided people the means to finance the post war increase in spending. To say the post war economy happened without government policy playing a significant role is just blatantly misleading and dishonest. Second, no one brought up the question of what happened to all the women who had been drawn into the labor force during WW II like by mother who was a Rosy the Riveter in the local aluminum plant in eastern Tennessee. Like most of her fellow female workers she silently went home and started having babies. In other words they dropped out of the labor force and this created openings for the returning servicemen. This movement in and out of the labor force of female wartime workers massively distorted the data but of course no is pointing this out in the analysis. That is why the labor force contracted in 1946. Note Henderson did not say anything about this.
Moreover, his analysis of Germany is just as blatantly misleading in that it says nothing about the Marshall Plan or the German currency reform. The German recovery did not really start until after the currency reform of 1948.
m.jed
It’s total population.
http://www.investorwords.com/6827/GDP_per_capita.html
My point is a caution about fine-tuning. Henderson points to 1946 a lot, and there couldn’t have been much of an effect on total population at that early date. I’m sure it takes several years of pumping lots of kids throught the pipleline before it makes a difference.
“Ask people who lived through that period as young adults what economic conditions were like, and you will inevitably get the answer that they experienced an economic boom.”
Don’t suppose that the fact that the growth in average income per capita translated into median income per capita had anything to do with it, do you? If you survey the people in the top 1% of income brackets, the post-1999 era has been sooooweeet!
Robert,
I’ll put the employment post on my to do list. It may have to wait a couple of weeks until I have the semblance of time – there’s two things ahead of it.
Of course civilian employment jumped. A whole lot of people returned to the country or got discharged from the military immediately after the War, my grandfather among them, and there were a lot of jobs created to fill the pent up demand from four years of war.
The point is not whether there was a one time jump in employment, which seems to be where Henderson’s hat is hanging when all is said and done. The point is, is there sustainable growth? Does everyone feel happy they’re finally home? Of course. Are they finding jobs? Sure. But by 1949, have they produced a whole heck of a lot more than in 1947? (I picked two years after the war to avoid whatever distortions he seems to think are there.) If the answer to that is no, and it is, Henderson’s boom is awfully hard to spot.
However, it gets worse. I wrote a quick post (up on Presimetrics now, up on AB later today) which has a graph which I think nails this coffin shut once and for all.