Guns and Ice Cream

I’ve gotten some pushback on the line from my decarbonization piece that “wartime mobilization did not crowd out civilian production.” More than one person has told me they agree with the broader argument but don’t find that claim believable. Will Boisvert writes in comments:

Huh? The American war economy was an *austerity* economy. There was no civilian auto production or housing construction for the duration. There were severe housing shortages, and riots over housing shortages. Strikes were virtually banned. Millions of soldiers lived in barracks, tents or foxholes, on rations. So yeah, there were drastic trade-offs between guns and butter (which was rationed for civilians).

It’s true that there were no new cars produced during the war, and very little new housing.1 But this doesn’t tell us what happened to civilian output in general. For most of the war, wartime planning involved centralized allocation of a handful of key resources — steel, aluminum, rubber — that were the most important constraints on military production. This obviously ruled out making cars, but most civilian production wasn’t directly affected by wartime controls. 2 If we want to look at what happened to civilian production overall, we have to look at aggregate measures.

The most comprehensive discussions of this I’ve seen are in various pieces by Hugh Rockoff.3 Here’s the BEA data on real (inflation-adjusted) civilian and military production, as he presents it:

Civilian and military production in constant dollars. Source: H. Rockoff, ‘The United States: from ploughshares into swords’ in M. Harrison, ed, The Economics of World War II

As you can see, civilian and military production rose together in 1941, but civilian production fell in 1942, once the US was officially at war. So there does seem to be some crowding out. But looking at the big picture, I think my claim is defensible. From 1939 to its peak in 1944, annual military production increased by 80 percent of prewar GDP. The fall in real civilian production over this period was less than 4 percent of prewar GDP. So essentially none of the increase in military output came at the expense of civilian output; it was all additional to it. And civilian production began rising again before the end of the war; by 1945 it was well above 1939 levels.

Production is not the same as living standards. As it happens, civilian investment fell steeply during the war — in 1943-44, it was only about one third its prewar level. If we look at civilian consumption rather than output, we see a steady rise during the war. By the official numbers, real per-capita civilian consumption was 5 percent higher in 1944 – the peak of war production — than it had been in 1940. Rockoff believes that, although the BLS did try to correct for the distortions created by rationing and price controls, the official numbers still understate the inflation facing civilians. But even his preferred estimate shows a modest increase in per-capita civilian consumption over this period.

We can avoid the problems of aggregation if we look at physical quantities of particular goods. For example, shoes were rationed, but civilians nonetheless bought about 5 percent more shoes annually in 1942-1944 than they had in 1941. Civilian meat consumption increased by about 10 percent, from 142 pounds of meat per person in 1940 to 154 pounds per person in 1944. As it happens, butter seems to be one of the few categories of food where consumption declined during the war. Here’s Rockoff’s discussion:

Consumption of edible fats, particularly butter, was down somewhat during the war. Thus in a strict sense the United States did not have guns and butter. The reasons are not clear, but the long-term decline in butter consumption probably played a role. Ice cream consumption, which had been rising for a long time, continued to rise. Thus, the United States did have guns and ice cream. The decline in edible fat consumption was a major concern, and the meat rationing system was designed to provide each family with an adequate fat ration. The concern about fats aside, [civilian] food production held up well.

As this passage suggests, rationing in itself should not be seen as a sign of increased scarcity. It is, rather, an alternative to the price mechanism for the allocation of scarce goods. In the wartime setting, it was introduced where demand would exceed supply at current prices, and where higher prices were considered undesirable. In this sense, rationing is the flipside of price controls. Rationing can also be used to deliver a more equitable distribution than prices would — especially important where we are talking about a necessity like food or shoes.

The fundamental reason why rationing was necessary in the wartime US was not that civilian production had fallen, but because civilian incomes were rising so rapidly. Civilian consumption might have been 5 percent higher in 1944 than in 1940; but aggregate civilian wages and salaries were 170 percent higher. Prices rose somewhat during the war years; but without price controls and rationing inflation would undoubtedly have been much higher. Rockoff’s comment on meat probably applies to a wide range of civilian goods: “Wartime shortages … were the result of large increases in demand combined with price controls, rather than decreases in supply.”

Another issue, which Rockoff touches on only in passing, is the great compression of incomes during the war. Per Piketty and co., the income share of the top 10 percent dropped from 45 percent in 1940 to 33 percent in 1945. If civilian consumption rose modestly in the aggregate, it must have risen by more for the non-wealthy majority. So I think it’s pretty clear that in the US, civilian living standards generally rose during the war, despite the vast expansion of military production.

You might argue that even if civilian consumption rose, it’s still wrong to say there was no crowding out, since it could have risen even more without the war. Of course one can’t know what would have happened; even speculation depends on what the counterfactual scenario is. But certainly it didn’t look this way at the time. Real per capita income in the US increased by less than 2 percent in total over the decade 1929-1939.  So the growth of civilian consumption during the war was actually faster than in the previous decade. There was a reason for the popular perception that “we’ve never had it so good.”

It is true that there was already some pickup in growth in 1940, before the US entered the war (but rearmament was already under way). But there was no reason to think that faster growth was fated to happen regardless of military production. If you read stuff written at the time, it’s clear that most people believed the 1930s represented, at least to some degree, a new normal; and no one believed that the huge increase in production of the war years would have happened on its own.

Will also writes:

War production itself was profoundly irrational. Expensive capital goods were produced, thousands of tanks and warplanes and warships, whose service lives spanned just a few hours. Factories and production lines were built knowing that in a year or two there would be no market at all for their products.

I agree that military production itself is profoundly irrational. Abolishing the military is a program I fully support. But I don’t think the last sentence follows. Much wartime capital investment could be, and was, rapidly turned to civilian purposes afterwards. One obvious piece of evidence for this is the huge increase in civilian output in 1946; there’s no way that production could increase by one third in a single year except by redirecting plant and equipment built for the military.

And of course much wartime investment was in basic industries for which reconversion wasn’t even necessary. The last chapter of Mark Wilson’s Destructive Creation makes a strong case that postwar privatization of factories built during the war was very valuable for postwar businesses, and that acquiring them was a top priority for business leaders in the reconversion period. 4 By one estimate, in the late 1940s around a quarter of private manufacturing capital consisted of plant and equipment built by the government during the war and subsequently transferred to private business. In 1947, for example, about half the nation’s aluminum came from plants built by the government during the war for aircraft production. All synthetic rubber — about half total rubber production — came from plants built for the military. And so on. While not all wartime investment was useful after the war, it’s clear that a great deal was.

I think people are attracted to the idea of wartime austerity because we’ve all been steeped in the idea of scarcity – that economic problems consist of the allocation of scarce means among alternative ends, in Lionel Robbins’ famous phrase. Aggregate demand is, in that sense, a profoundly subversive idea – it suggests that’s what’s really scarce isn’t our means but our wants. Most people are doing far less than they could be, given the basic constraints of the material world, to meet real human needs. And markets are a weak and unreliable tool for redirecting our energies to something better. World War II is the biggest experiment to date on the limits of boosting output through a combination of increased market demand and central planning. And it suggests that, altho supply constraints are real — wartime controls on rubber and steel were there for a reason – in general we are much, much farther from those constraints than we normally think.

 

 

 

  1. Since the housing stock is long-lived, the lack of new construction didn’t do much to reduce the supply of housing. Limits on construction were outweighed by the shift into military housing, so average persons per dwelling actually fell a bit during the war. That said, Will is right that there were acute housing shortages in some cities that were centers of wartime production.
  2. The best source on the mechanics of wartime planning is Paul Koistinen’s Arsenal of World War II.
  3. Rockoff himself has made the same comparison with the response tf climate change, and reached the same conclusion: The experience of wartime “mobilization suggests … that large scale projects involving the building of alternative energy sources or moving populations from low lying coastal areas could … probably could be accomplished without seriously endangering current living standards.” That said, he adds the caveat that a WWII-scale mobilization might require a similar political consensus that the country faces an “existential and immediate threat.”
  4. I reviewed the book here.

25 thoughts on “Guns and Ice Cream”

  1. This is pretty compatible with an MMT worldview; you can print money and give it to people to buy things until we hit 1% unemployment, but you’d have to place some sort of rationing on scarce goods: land, natural resources, energy. All those can be covered with Georgist taxes (land value taxes, mineral taxes, BTU taxes, etc.).

    One thing that’s always bothered me about this model, however, is that it doesn’t seem to work as well for non-advanced countries; the Soviets did it, rapidly advancing from a backwards country to an industrial superpower in a very short period, but it began to sputter in the 1950s. Or, Venezuela today (I know there’s a major problem with graft).

    1. It’s not simply about rationing finite resources, the other big element is to actively fund the development of alternatives (at the time artificial rubber and other such elements but could also be different “recipes) to those finite resources. In other words the active funding of resource creation is a big element of spending driven transformation.

      There’s a lot that could be said about the soviet union and much too much to analyze there but one element I would highlight is i don’t think their planning apparatus was good at funding resource transformation rather than expanding scale with existing technologies.

      The big problem for Venezuela is how impoverished their productive structure was and is while politically doing what was necessary to transform that structure, like land reform, was extremely difficult (peasant assassinations were and are a thing). They’re food dependency has always been incredible. MMT would also highlight the appropriate technology literature and how places like Venezuela urbanized very intensely without a productive structure that could keep up with it (and this vastly predates the Chavez).

  2. Shouldn’t really compare 42-45 to 40/41, should you? But to what those years would have been like without conversion of production to war materials? Thinking you’d see a hell of a lot more “crowding out.“

  3. Since we are on the subject of planned economy it would be interesting to read your thoughts on the Soviet experiment along with any references you feel are useful.

    The USSR did manage to triple its economic output between 1950 and 1973 while it essentially stagnated in the following two decades until the fall of the Soviet Union in 1990.

  4. How much of the consumption can be attributed to changes in the labor market as the depression lifted? Or is that the real point, crowding out is something you would see near a production frontier, not coming out of a depression.

  5. “From 1939 to its peak in 1944, annual military production increased by 80 percent of prewar GDP. The fall in real civilian production over this period was less than 4 percent of prewar GDP. So essentially none of the increase in military output came at the expense of civilian output; it was all additional to it. And civilian production began rising again before the end of the war; by 1945 it was well above 1939 levels.”
    But why pick 1939 as the baseline? From 1941—a peacetime year until December 7—the 1944 civilian production index is down 15 percent, and 1943 was down 18 percent. That’s not just crowding out, it’s the equivalent of a severe depression in civilian output. And 1941’s civilian production index was itself probably depressed a bit from what it would have been without the crowding-out effect of that year’s rearmament.
    “By the official numbers, real per-capita civilian consumption was 5 percent higher in 1944 – the peak of war production — than it had been in 1940. Rockoff believes that, although the BLS did try to correct for the distortions created by rationing and price controls, the official numbers still understate the inflation facing civilians. But even his preferred estimate shows a modest increase in per-capita civilian consumption over this period.”
    So, you’re saying the consumption data is markedly different from the production data in Rockoff’s graph, because production includes, I guess, both consumer goods and capital goods, so while the latter were way down the former held steady and didn’t show the big dip in Rockoff’s production data? Can you show the consumption data? How does 1944 consumption compare to a 1941 baseline?

  6. “You might argue that even if civilian consumption rose, it’s still wrong to say there was no crowding out, since it could have risen even more without the war. Of course one can’t know what would have happened; even speculation depends on what the counterfactual scenario is. But certainly it didn’t look this way at the time. Real per capita income in the US increased by less than 2 percent in total over the decade 1929-1939. So the growth of civilian consumption during the war was actually faster than in the previous decade. There was a reason for the popular perception that “we’ve never had it so good.” It is true that there was already some pickup in growth in 1940, before the US entered the war (but rearmament was already under way). But there was no reason to think that faster growth was fated to happen regardless of military production.”

    In retrospect there is good reason to think that.

    The 1929-39 growth trend line is the wrong baseline for assessing the war economy. It’s more appropriate to look at the Rockoff graph and extend the 1938-41 trend line through the war years. That trend shows that the Depression was over and growth was strong during the pre-war period. Civilian production grew at an average compounded rate of 7 percent per year from 1938 through 1941 (despite probably some crowding out by rearmament in 1941). Had there been no war and that pace continued past 1941, the civilian production index would have reached 328 in 1946, well above the actual figure of 287. Even if average growth rates after 1941 had fallen to 4 percent per year, civilian production would still have reached 296 by 1947, equaling the actual value. In the meantime, without the war civilian production and consumption would have been markedly higher during 1942-5 on a 4 percent trend line.

    From Rockoff’s graph you can make a plausible argument that the war economy had no long-term impact on growth at all to compensate for its suppression of civilian production and consumption during the war.

  7. Re the irrationality of war production:

    The point I was groping for is that government war-time investment was fundamentally irrational from an economic standpoint because of its extremely short time horizon and indifference to returns.

    This point is really a response to your review of Wilson, where you write:

    “The large-scale investment in plants and equipment required by both military mobilization and industrialization is often unattractive to private wealth-holders, who put a steep discount on returns far off in an uncertain future.”

    I think this misconstrues the nature of war-time investment. The government was not investing in an uncertain long term during the war, it was investing for a particular goal in the extreme short term, namely VJ Day. Even that overstates the time horizon; the sole purpose of government war investment was to maintain a margin of production over attrition sufficient to keep the troops supplied.

    The Liberty ships are a case in point. Much of the productivity gain on the Liberties came from simplified designs and manufacturing techniques, but the changes made the ships less functional, slower and flimsier. They were designed for an average service life of just five years; some continued in service after the war, but many were just scrapped. That was fine with war planners: they reckoned the war would be over in 5 years and couldn’t care less whether all the Liberties sank after that. And if the war wasn’t over in 5 years and the Liberties sank, they would just build more; the margin was all that mattered. (Almost: they did introduce a slightly better-quality line of Victory ships that could outrun submarines late in the war.)

    Yes, some war investment did survive and get repurposed for civilian use, but much of it was wrecked or scrapped. The important counterfactual is whether normal, economically rational investment would have on average produced longer-lived and more productive assets than war planning did. There’s a good case it would have.

    So the problem with war production is not that private wealth-holders refuse to invest in industrial assets for an uncertain future—they do that all the time–it’s that they refuse to undertake long-term investment in industrial assets that are intended to have an extremely short life-span that cannot possibly repay the investment. Only the state can afford such a short time-horizon and to be so indifferent to long-term returns, which is why only the state can finance war. But that’s also why war-time investment is a dubious model for a civilian economy or the energy sector, which should think hard about the long-term viability of investments and assets.

  8. The Soviet Union is an instructive case. They had a successful war economy too, churning out lots of good weapons under trying circumstances (albeit with a crucial assist from American and British aid).

    But the central planning that beat Hitler worked badly for civilian production after the war. The Soviets remained a military superpower but never became an industrial superpower, and could only keep up with the Pentagon by devoting a huge portion of the economy to the military (going on one third to one half, IIRC).

    Meanwhile, the Soviet civilian economy was a Hayekian cautionary tale, marked by inefficiency, gluts and bottlenecks, poor quality, low innovation, severe housing shortages, periodic food crises, the egalitarianism of the cabbage line.

    It was also a study in the politicized economic distortions that are rife in central planning in war or peace. The politically favored sectors of armaments, Olympic sports, ballet and chess flourished because of an outsized allotment of resources, while housing, agriculture and autos lagged amid a generalized economic torpor and chaos.

    Any argument for a planned war economy as a model for a civilian economy has to reckon with the Soviet example (and other communist economies).

  9. 1) You still can’t use the war economy to argue for a Keynesian magic wand. The data from Rockoff show that, even if meat and shoes and ice cream were in good supply, there was still serious crowding out of civilian production and trade-offs between cars and tanks, ships and housing, guns and butter. Further, even if consumption held up or grew during the war (I’m skeptical) there were still opportunity costs in diverting rising output into military goods instead of raising civilian consumption and living standards even more. Those opportunity costs and trade-offs were very much the subject of political conflict during and after the war.

    Even if Keynesian stimulus works great in the future, there will still be bitter arguments over which sectors it should feed: Housing? Agriculture? Hospitals? Sewage treatment? Transport? Desal plants and air conditioners to cope with drought and warming? All these sectors meet urgent needs and can make a plausible claim to investment priority over clean energy.

    2) The example of the American war economy does not show that central planning will succeed in a civilian economy, and there’s much evidence to the contrary that you haven’t engaged with.

    The deeper problem here is that you take a generic view of “planning,” where in fact specific planning processes can be radically different and incommensurable. The way a government plans for a war is different from the way private firms plan for markets. Considerations of time horizons, return on investment, cost and resource constraints, access to financing, consumer preferences and market competition mean that the planning processes of government planners will be drastically different from those of businessmen. To simply assume that a central-planning framework will work as well in a civilian context as the business-planning framework is indefensible. You have to demonstrate that in detail.

    4) I’m not against central planning in every civilian context. In the clean energy sector it has a mixed record. But it’s important to avoid magical thinking. Keynesianism and central planning in the energy sector can be useful, but they aren’t panaceas, and they won’t do away with trade-offs, opportunity costs, resource constraints or political controversies over how to allocate rising output. World War II shows that in spades.

    1. I will write a proper reply later. But where I’m hoping we can end up is somewhere between the idea — which I really do think is widespread — that real reources are fixed and that anything devoted to climate change is a one for one subtraction from meeting other needs; and the idea that there are no real constraints at all.

  10. Two other factors to consider:

    – the war was a huge investment in human capital (most military people are not actually fighting, but doing technical jobs – mechanics, radio techs, machinists etc – and even fighting is an affair demanding close coordination and skilled management). This was a large and lasting investment

    – the war also invested in a range of new technologies and – crucially – the infrastructure and expertise to support them. Oil and airplanes are two examples, but one could add many more. Private enterprise is rather bad at this sort of transition – all that I can think of have been assisted by extensive government direction.

    Re Liberty ships – rather than retiring after five years, many soldiered on for 20 or more. Over 500 were bought by Greek and Italian interests and formed the core of their merchant marines.

    1. Right, some of the 2500 + Liberties stayed in service after the war. The 5-year service life was an average, and could of course be extended by repairs and retrofits. Still, the consensus is that the Liberties were low-quality ships, thought they did the job they were designed for well enough. Design is always a compromise between quality and cost, and the short-term purpose of the Liberties put the emphasis on the latter.

    2. Poking further into the post-war fate of Liberty ships, so it does look like very few of them were scrapped immediately after the war. Most went into the National Defense Reserve Fleet, lying permanently at anchor awaiting call-up in an emergency. About 1500 are listed in the NDRF in 1956. Several hundred of them would have been in service during the Korean War.
      Most of them were gradually scrapped over the next 15-20 years.

      Most of the rest of the 2400 or so that survive were bought by the Greeks and Italians, and some stayed in regular service into the 1970s.

      I can’t find stats on the average service lives of those that were not mothballed in the NDRF and continued in merchant service, but I guess they were not as flimsy as I assumed.

    3. Issues with the Liberties:

      1) They were built with a cheaper and faster all-welded construction method instead of riveted construction. In conjunction with subtle architectural defects and an inferior grade of steel, the result was large cracks sometimes opening in the hulls; in three cases ships split in two without warning at sea. This flaw was unplanned but was due to corner-cutting to reduce costs and speed construction.

      2) The second major problem with Liberties was the reciprocating steam engine, which was obsolete compared to state-of-the-art steam turbine engines but was again cheaper and faster to build. The engine limited the speed to 11 knots, which made the ship more vulnerable to submarines and less valuable for commercial shipping compared to the 15-knot industry standard. That was a deliberate design choice, driven by the desire to get tonnage afloat as fast as possible.

      So the ships were okay for military transport and found post-war civilian buyers at drastically reduced military-surplus prices, but they would not have been designed or built as they were for a civilian market.

      1. Turbines , with their need for advanced metallurgy, high-precision machining and large reduction gears, were very expensive. Hence the continued use of reciprocating engines (including in eg, the British corvettes that were the mainstay of convoy escort).

        The crack issues (combination of steel and welding) were a problem only in very cold waters. Not good for north Atlantic, but ok elsewhere.

        Given the enormous merchant marine losses of the war (eg 96% of the Italian fleet) – which made tonnage a critical factor in decisions until the early 50s – mass production of a cheap design would have made sense anyway – for any market. Remember that these were built essentially for civilian needs – the transport of food, fuel and raw materials.

        1. We don’t know that the Italians would have chosen to invest in a Liberty-style ship rather than a better design. The Liberties happened to be there at fire sale prices–in 1946 they sold for about $600,000. Would they have been bought by the civilian market at their production cost of about $2 million–given the likelihood of their requiring expensive repairs as well? Maybe not.

  11. Robert Gordon’s dissertation on the jump in fixed capital measurements is worth noting here in regards to the value of privatized government plant after the war.

  12. Note to Mr. Boisvert:
    The Liberty ships were built of AISI (American Iron & Steel Institute) A-7 structural steel, which was the ‘run of the mill’ for US steel producers of that era. That is the stuf that still holds up the Empire State, the Golden Gate, and many things in between. Problem was that the specs for it were finalized in 1900, before much was known about ‘brittle cracking in cold weather conditions’

    The Skinner Unaflow reciprocating engines were powerful enough to keep up with the ‘fast convoy’ speeds, were kept running by men who had never seen a marine engine four months previous, and the parts could be built by the foundaries and machine shops in small-town America and assembled by Henry J.

    In the winter and spring of ’42-’43 – the worst time of the U-Boat war – losses in some transatlantic convoys were about 20% of the ships that sailed. At that rate, building a ship that can stay together for three round trips makes some sense.

    1. It’s not just the cold-temperature brittleness of the steel. The all-welded construction of Liberty hulls allowed cracks to propagate that would not have with the usual riveted construction. Also, the design of the ship’s hatches contributed to cracking problems.

      And there’s rushed construction. A study I saw suggested that the fraction of ships with cracking problems rose steeply as construction times fell. There’s a trade-off between soaring shipyard productivity and quality.

      Sources say “fast convoy” speed was anywhere between 9 and 13 knots, putting the Liberty’s top speed of 11.5 knots in the middle. U-boats cruised at about 10 knots, with a top speed of 17 knots. Speed really matters in evading a wolf-pack, which is one reason the Victory ships upgraded the engine to 15-17 knots.

      Right, the Liberties were expected to have high losses, one reason they weren’t built to last.

  13. From Wikipedia: (The HX series of) “fast convoys made up of ships that could make 9-13 knots. A parallel series of slow convoys, the SC series, was run for ships making 8 knots or less,”

    Speed, of course, requires horsepower, which means more, or more sophisticated engines, and better-trained engine room crews.

    A reasonable job of welding can be done by newly-trained men and women ‘fresh off the farm’. A productive riveting gang is a half-dozen men, some quite physically strong and agile, used to working together.

    (This is the fourth or fifth time I’ve had this same argument. It seems there is a book of ‘right-wing talking points’ which lists partial truths about the Liberty ships for anyone disputing the economic lessons of US World War II productivity.)

  14. The problem with Mr. Boisvert’s arguments is that ‘we could have done better than the Liberty ships – with unlimited supplies of: premium-quality steel, machinists in the steam-turbine plants, riveting gangs in the shipyards, and experienced marine engineers to sail them.

    As Churchill said “Generals only enjoy such comforts in Heaven, and those that demand them do not always get there.”

Comments are closed.