Economics in Times of Crisis
Cliometric Sessions at 1990 ASSA Meetings--December 28, 8:00 AM


Wartime Prosperity during World War II?

Robert Higgs
Seattle University

For nearly half a century economists and economic historians, almost without exception, have misinterpreted the performance of the U.S. economy in the 1940s. The reigning view has two related aspects, one pertaining to the conceptualization and measurement of the economy's performance and one pertaining to the explanation of that performance in the light of macroeconomic theory. The two aspects are encapsulated in the title and subtitle of the chapter on World War II in Jonathan Hughes' textbook: "War Prosperity: The Keynesian Message Illustrated."

I shall challenge every part of the consensus. My claim is that the accepted profile of the economy's performance during the 1940s--peak "prosperity" during 1943-45, followed by substantially worse performance during 1946-49--is exactly the opposite of the profile most defensible from a welfare standpoint. Further, universally accepted macroeconomic explanations of the events of the war years are unwarranted.

The decade of the 1940s was a watershed in American economic history, and economic historians need to understand it properly; but they also need to understand a more general matter, namely, that the standard concepts and measures of national income accounting and the explanations derived from orthodox macroeconomic theories lose their meaning in certain institutional contexts, chief among which is the command economy. The prevailing misinterpretations of the performance of the U.S. economy during the 1940s have arisen because economists and economic historians have failed to appreciate that the U.S. economy during the war was a command economy.

The Consensus

According to the orthodox account, the war "got the economy out of the depression." Evidence for this claim usually includes (1) great decline in the standard measure of the unemployment rate, (2) great increase in the standard measure of real GNP, (3) slight increase in the standard measure of real private consumption. The entire episode of apparent business cycle expansion during the war years is understood by most writers as an obvious validation of the simple Keynesian model: enormous government spending, financed mainly by issuing debt, spurred the military economy itself and had multiplier effects on the civilian economy, the upshot being increased real output and employment and decreased unemployment. Some analysts, recognizing the enormous increase of the money stock during the war, blend Keynesian and monetarist explanations, treating them as complements rather than substitutes.

The authors of economic history textbooks rely on data taken from Historical Statistics to document their accounts of economic performance during the 1940s. None of them expresses any awareness that those data-- for example, GNP measures based on the Commerce Department's concept of national product--might be problematic. The standard numbers receive universal acceptance at face value.

Employment and Unemployment

The standard measure of the unemployment rate (the number of persons officially unemployed as a percent of civilian labor force) falls greatly between 1940 and 1944, from 14.6 percent to 1.2 percent. Michael Darby's measure of the unemployment rate, which does not count those in "emergency government employment" as unemployed, falls from 9.5 percent to 1.2 percent. Either measure signals a virtual disappearance of unemployment during the war, but in the circumstances neither measure means what it is commonly taken to mean.

The buildup of an armed force of more than 12 million persons by mid-1945 made an enormous decline of the standard unemployment rate inevitable. But the welfare significance of this decline is far from the usual one. Of the 16 million persons who served in the uniformed armed forces at some time during the war, 10 million were conscripted, and many of those who volunteered did so only to avoid the draft and the consequent likelihood of assignment to the infantry. The civilian labor force during 1940-45 was 54-56 million. Therefore the 12 millions serving in the armed forces in 1944-45, most of them under duress, constituted about 18 percent of the total (civilian plus military) labor force, itself much enlarged during the war.

In short, the country started in 1940 with an unemployment rate (Darby concept) of 9.5 percent; the government then pulled the equivalent of 22 percent of the prewar labor force into the armed forces; and, voila, the unemployment rate dropped to a very low level. No one needs a macroeconomic model to understand these events. No plausible view of the economy, given the facts of the draft, is incompatible with the observed decline of the standard unemployment rate. Whether the government ran deficits or not. whether the money stock increased or not, massive military conscription was sure to decrease dramatically the standard rate of unemployment. For those who insist on a macroeconomic framework, the model estimated by Paul Evans (Journal of Political Economy, Oct. 1982) can be used to confirm that labor market changes during the war were overwhelmingly driven by the draft.

So, the tight civilian labor market during the war reflected the creation of huge military employment, but military "jobs" differed categorically. They ranged (sometimes within the same job) from the abjectly disgusting to the intolerably goring to the unspeakably horrifying. Whatever their qualities, they lasted for the"duration." Often they entailed substantial risks of death, dismemberment, and other physical and psychological injury; sustained involvement in combat drove many men insane. Physical casualties included 405,399 dead and 670,846 wounded. To make the military jobs commensurable with the civilian jobs betrays a monumental obtuseness to the underlying realities.

To see more clearly what happened to the labor force, one can examine the part of the total (civilian plus military) labor force composing the labor "residuum," that is, all those outside nondefense employment. This includes the civilian unemployed plus uniformed members of the armed forces plus civilian employees of the armed forces plus everyone employed in the military supply industries. This measure rises from 17.6 percent in the fiscal year 1940 to more than 40 percent during fiscal years 1943-45, then drops abruptly to about 10 percent during fiscal years 1946-49. the extraordinarily high level of the labor residuum during 1942-45 signals that the "prosperous" condition of the labor force during the war was spurious: official unemployment was virtually nonexistent, but four-tenths of the labor force was not being used to produce consumer goods or capital capable of yielding consumer goods in the future. The sharp drop of the labor residuum between fiscal years 1945 and 1946 marks the genuine return of prosperity.

Real Output

To find out what happened to real output during World War II, economic historians usually reach for Historical Statistics. Economists typically reach for the latest issue of the Annual Report of the Council of Economic Advisers. Which source one consults makes a difference. Although the two series show roughly the same profile of real GNP during the 1940s, the latest Commerce Department version indicates (when indexed as 1939=100) a peak value of 192.7 in 1944, versus a peak value of 172.5 in 1944 in the series taken from HSUS. The twelve percent difference is hardly negligible, even though it reflects only statistical as opposed to conceptual revisions. Both series show a sharp drop of real GNP for 1945 to 1946: 12 percent in one case, 19 percent in the other.

Another familiar source in John Kendrick's Productivity Trends (1961). Kendrick's series for real GNP ("national security" variant) moves similarly but displays some discrepancies, especially vis-a-vis the most recent Commerce Department version. From 1945 to 1946, the Kendrick estimate drops by just 9 percent. By 1949 it has fallen nearly into agreement with the other two, the difference being little more than 1 percent.

Economists and economic historians who employ the standard real GNP series seem generally unaware that the series may be conceptually problematic. By contrast, Simon Kuznets, a leading figure in the development of the national income accounts, expressed many concerns. In his National Product in Wartime (1945), Kuznets discussed a number of issues that analysts must consider when deciding how to construct national product measures for a nation at war. Noting that the "complexity of observable reality compels the investigator to select one set of assumptions from among many concerning the purpose, value, and scope of the economic activity," he observed that "a major war magnifies these conceptual difficulties, raising questions concerning the ends economic activity is made to pursue...[and] the distinction between intermediate and final products." He noted that "war and peace type products...cannot be added into a national product total until the difference in the valuation due to differences in the institutional mechanisms that determine their respective market prices are corrected for." Convinced that the problem had no unique solution. Kuznets constructed several variants, each based on distinctive assumptions.

One of the variants Kuznets developed in his 1945 book ("wartime concept a," his most preferred) does not differ greatly from the standard Commerce variant. Discrepancies for 1942 and 1943 reflect the difficulty Kuznets experienced during the war in getting complete data on war production and constructing appropriate deflators for munitions.

After the war, Kuznets refined his wartime estimates. The later estimate differs substantially "partly because of the allowance for overpricing of certain types of war production, partly because of the exclusion of nondurable war output (essentially pay and subsistence of armed forces)." Kuznets contrasted his estimates with those of the Commerce Department, remarking that "the Department of Commerce totals yield results difficult to accept." He wrote that "the adjustment for over-pricing of war production. arbitrary as it may be, is preferable to no adjustment, or to one which, like that of the Department of Commerce, fails to meet the issue."

The problem Kuznets had in mind was that the Commerce Department made no correction for the decline in the relative price of munitions during the war, thereby overestimating the increase of real GNP. The Commerce Department itself had little confidence in its estimate of real munitions output during the war. In the Department's 1954 supplement on national income. Commerce economists admitted the validity of Kuznets' criticism, but they failed to make the necessary corrections.

As a result, the Commerce variant of GNP shows a much larger increase of real GNP during the war. As of 1944, the Commerce index stands at 192.7, Kuznets' at 135.8--an enormous difference, enough to alter substantially one's understanding of the path of real GNP during the 1940s. The difference carries over into the postwar period. Whereas the official Commerce estimate of real GNP drops precipitously in 1946 and remains at that relatively low level for the rest of the decade, the Kuznets variant Increases by about 8 percent in 1946, then rises slightly higher during the subsequent three years. Although the two series stand at almost identical levels in 1949, their profiles over the decade look completely different.

Besides making an adjustment for the change in munitions prices, Kuznets also deleted nondurable war output (pay and subsistence of armed forces) from his estimate. Obviously, this change also contributes to the discrepancy with the Commerce variant. He might have gone even further, however, to delete all government outlays for war purposes. The crucial question is: Does government war expenditure purchase a final good, and hence belong in GNP, or an intermediate good, and hence not belong?

In his studies of long-term economic growth, Kuznets always insisted on a "peacetime concept" of GNP. In this, government spending counts only if it pays for a flow of goods to consumers or a flow into capital formation. Military spending enters only to the extent that it finances additions to the military capital stock, the justification being that even though military durable and construction are used for military purposes, they represent capital that could be employed for nonmilitary purposes--a justification that seems far-fetched with regard to may forms of military capital.

Application of this approach in estimating GNP yields the series that Kuznets (Capital in the American Economy, 1961) designated Variant III. This looks completely different over the 1940s. Whereas alternative estimates show more or less large increases of real GNP during the war years, Variant III reached a peak in 1941, stalls throughout the war period, then surges with the demobilization and reconversion. Real GNP jumps by nearly 17 percent between 1945 and 1946, remaining at the new, higher level for the rest of the decade. Although this variant shows almost the same 10-year gain as the alternative indexes show for 1939-49, it follows a completely different course in getting to the common end state. If one were to accept Variant III as the most justified concept, the perception of "wartime prosperity" would be undermined.

William Nordhaus and James Tobin, in their monograph "Is Growth Obsolete?" (1972) made numerous adjustments to the standard GNP concept to transform it into what they call a measure of economic welfare (MEW). They aimed to eliminate from GNP "activities that are evidently not directly sources of utility themselves but are regrettable necessary inputs to activities that may yield utility"; that is, they sought to eliminate spending that is "only instrumental." Accordingly, they deleted, inter alia, all national defense spending. The implication of accepting the Nordhaus- Tobin position on the treatment of military output is to move at least to Kuznets' Variant III as the most meaningful of the available measures of real GNP during the 1940s.

If one makes that move, one's understanding of how the U.S. economy performed during the 1940s changes completely. Now one sees no wartime prosperity. The real GNP in 1944, which the Commerce variant shows at a cyclical peak, was no greater than it had been in 1941 according to Variant III. Only with the end of the war did the economy at last break out of its 15-year era of substandard performance. The conclusion emerges even more starkly if one adopts fully the Nordhaus-Tobin position on the treatment of military outlays.

Finally, one can make an even stronger argument for rejecting the orthodox account of changes in real GNP during the war. One can simply argue that outside a more or less competitive market framework, prices become meaningless; all presumption that price equals marginal cost vanishes, and therefore no theoretically justified estimate of real national product is possible. Although I consider this argument sound, I do not expect many of my fellow economist to accept it.

Real Consumption

Some writers who recognize that the expansion of real GNP during the war consisted overwhelmingly of military outputs nevertheless insist that real private consumption also increased. In Seymour Melman's colorful but otherwise representative portrayal, "the economy [was] producing more guns and more butter..Americans had never had it so good."

This view is wrong. It fails to take sufficiently into account (1) the understatement of actual inflation by the official price indexes, (2) the deterioration of quality and disappearance form the market of many consumer goods, (3) the effects of the rationing of many widely consumed items, and (4) the additional transactions costs borne and other sacrifices made by consumers in getting the goods that were available. When one corrects the data to provide a more defensible estimate of real consumer well-being during the war, one finds that it almost certainly declined. How much it declined cannot be established with precision, because the estimates require assumptions, and plausible alternative assumptions give rise to differing conclusions.

If one uses the price index of Milton Friedman and Anna Schwartz (Monetary Trends, 1982) to deflate personal consumption spending per capita, one finds that real per capita consumption reached a prewar peak in 1941, nearly 9 percent above the 1939 level; it then declined by more than 6 percent during 1941-1943 and rose during 1943-45; still, in 1945, it had yet to recover to the level of 1941. In short, according to this index, there was no wartime prosperity. The war years passed with the average consumer below the level of well-being enjoyed in the last prewar year. In 1946, however, the index jumped by 18 percent, and it remained at about the same high level for the rest of the decade.

During the war years consumers suffered extraordinary welfare- diminishing changes. To get the goods that were available, millions of people had to move, many of them long distances, to centers of war production. After bearing substantial costs of relocation, they often found themselves crowded into poorer housing. Because of the disincentives created by rent controls, the housing got worse each year, as landlords reduced or eliminated maintenance and repairs. Transportation, even commuting to work, became difficult for many workers. No new cars were being produced; used cars were hard to come by because of rationing and sold on the black market at enormous prices; gasoline and tires were rationed; and public transportation was crowded and inconvenient for many, as well as frequently preempted by the military authorities. Of course, in literally thousands of particular ways, consumers lost their freedom of choice.

One must also recognize that while consumers were actually getting less, they were working much harder and longer to get the available goods: the ration of civilian employment to population increased from 47.6 percent in 1940 to 57.9 percent in 1944, as many teenagers left school, women left their homes, and older people left retirement to work. The average workweek in manufacturing, where most of the new jobs were, increased from 38.1 hours in 1940 to 45.2 hours in 1944; and the average workweek increased in most other industries as well, in bituminous coal mining by more than 50 percent. It is difficult to understand how working harder and longer for a diminished flow of consumer goods comports with the description that "economically speaking, Americans had never had it so good."

Inappropriate Macro Models

None of the standard macroeconomic theories employed to account for the wartime experience of the U.S. economy provides an acceptable explanation. The standard models cannot do the job because none is a model of a command economy, and the United States during 1942-1945 was a command economy. Regardless of the peculiarities of their assumptions about certain price rigidities, persistent disequilibria, transaction costs, informational asymmetries, fundamental uncertainty, and the like, all of the standard macro models presume the existence of genuine markets for commodities, for the services of the factors of production, for bonds and other financial securities.

None of these assumptions even approximates the conditions that prevailed during the war. Commodity markets were pervasively subject to controls: Price controls, rationing, and in some case outright prohibition in the consumer goods markets; price controls, prohibitions, priorities, conservation and limitation orders, quotas, set-asides, scheduling, allocations, and other restrictions in the markets for raw materials, components, and capital equipment. while taxes were raised enormously, many forms of production received subsidies so that the price controls at the retail level would no drive suppliers from the market. Factor markets were no freer, and in some respects were less free, than the commodity markets. Conscription forced 10 million men into the armed forces. Credit markets experienced total control, as the Federal Reserve undertook to reduce and allocate consumer credit and pegged the nominal interest rate on government bonds at a barely positive level. Two-thirds of the investment in manufacturing plants and equipment during the period from July 1940 through June 1945 was financed by the government, and most of the remainder came forth in response to tax concessions and other de facto subsidies authorized in 1940 to stimulate the rearmament.

In sum, the U.S. economy during 1942-1945 was the exact opposite of a free market system. Every part of the economy was either directly controlled by the authorities or subject to drastic distortion by virtue of its relations with suppliers and customers who were tightly controlled. To suppose that the economy allocated resources in response to prices set by the free play of demands and supplies in underlying markets for commodities, labor services, and loanable funds is to suppose a complete fiction. So, the assumptions that underlie standard macro models are unsatisfied by the empirical reality of the wartime economy.

So What Did Happen?

Apart from economists and economic historians who have been misled by inappropriate and inaccurate statistical constructs, why did Americans--evidently almost all of them--think that prosperity had returned during the war years?

The question has several answers. Fists, everybody with a desire to work was working. After more than ten years of persistently high unemployment and its associated insecurities (even for those who were working), full employment relieved a lot of anxieties. Second, the national solidarity of the war effort, although decaying as time elapsed after the initial upsurge of December 7, 1941, helped to sustain the spirits of many who otherwise would have felt themselves aggrieved by the controls, shortages, and other irritations. Third, some individuals (e.g., employment in northern and western industries) were better off, even though the average person was not. Wartime reduction of the variance of person income--and hence of personal consumption--along with rationing, meant that many people at the bottom of the consumption distribution could improve their absolute position despite a reduction of the mean. Fourth, although people could not buy many of the things they wanted at the time, they were earning unprecedented amounts of money. Perhaps money illusion, fostered by price controls, made the money earnings look bigger than they really were. In any event, people were building up bank accounts and bond holdings; even if they were actually living worse than before, they were feeling wealthier.

Which brings us to what many be the most important factor of all: namely, that the performance of the wartime economy, notwithstanding its command-and-control aspect, broke the back of the pessimistic expectations almost everybody had come to hold during the seemingly endless Great Depression. In the 1930s, especially the latter half of the decade, many people had come to believe that the economic machine was irremediably broken, that old-fashioned prosperity would never return. The frenetic activity of war production--never mind that it was just a lot of guns and ammunition--dispelled the hopelessness. People began to congratulate themselves on their productive abilities. They began to think: if we can produce all these planed, ships, and bombs, we should be even better at turning out prodigious quantities of cars and refrigerators.

When the controls began to come off and the war ended earlier than anticipated in 1945, optimistic consumers and producers launched eagerly into carrying out plans based on rosy scenarios and, by so doing, made their expectations a reality. Of course, the ability to draw down the accumulations of financial assets build up by "forced saving" during the war was important. But obviously the presence of those assets alone could not have turned the trick. (If such tricks were possible depressions would be easy to cure.)

One sees, then, that World War II did get the economy out of the Great Depression, but not in the manner described by the orthodox story. The war Itself did not get the economy out of the depression. The war production surge brought forth massive amounts of munitions, leaving the private economy in general no better off, and in many respects worse off, than it had been in 1941. But certain events of the war period--the buildup of financial wealth and especially the transformation of expectations--justify an interpretation that views the war as an event that recreated the possibility of genuine economic recovery. The year 1946 witnessed the real economic miracle.