The War Production Board

In October 1943, Samuelson received a letter from Raymond Goldsmith, a German-Jewish economist who had immigrated to the United States in 1934, after which he had engaged in statistical work for a series of govern­ment agencies.

the overall magnitude of demand, prospects for boom, etc. I realize that the latter are important subjects but I feel that I have put in enough time on them already.⁴

Despite his workload, he could not let go of the opportunity provided by Goldsmith’s invitation. After further discussion, Samuelson suggested that his comparative advantage probably lay in “the use of critical rather than creative faculties.”⁵ Being offered the equivalent of $8,500 a year, a rate much higher than his MIT salary, would have added to the attraction of Goldsmith’s offer.

Despite this interest, Samuelson wrote to Goldsmith in January 1944, saying that he could not work for the WPB, a decision that Goldsmith regretted. Explaining that Samuelson’s letter had not arrived in time to stop the bureaucratic machinery, and that his appointment as an economic con­sultant had been approved by the Civil Service Commission, Goldsmith was clearly not going to accept no for an answer. He suggested that Samuelson could take the oath of office, but need do no work until his schedule allowed it. Samuelson protested that this missed the point, which was that he needed to be protected against himself: “because of my essential interest in the prob­lems at hand, it is the more important that I protect myself against myself, and not make it easy to take on responsibilities.”⁶ As proof of his interest, he enclosed a memorandum he had already written on whether the costs of converting industry back to peacetime production should be included in estimates of the cost of the war. His memorandum gave reasons why they should not be, though since writing it he had changed his mind. The cost of fighting a fire, he pointed out, should include the cost of getting the engine back to the depot afterwards.

On whatever terms—presumably what Samuelson jokingly called “a lend-lease basis”—Samuelson engaged in at least intermittent discussions with Goldsmith.⁷ He made proposals for alleviating the financial problems faced by businesses converting from military to civilian production, dis­cussing proposals ranging from expediting the termination of war contracts (which would make companies’ financial problems worse) to providing tax reliefs and “V loans” to war contractors and “government boon-doggling via private business”—subsidizing private companies to continue to provide jobs.⁸ He took it as a premise that there could be a recession if this were not handled correctly.

There was a dilemma in the government’s recouping as much as possible of the cost of the war from selling assets (factories and equipment) it no longer needed but ensuring that these assets got into the hands of those who could use them most productively.

Samuelson's discussions with Goldsmith did not stop after he started work­ing at the Radiation Laboratory? In June 1944, Goldsmith sent Samuelson some “additional data sheets” explaining how war outlays were calculated, and in July, Samuelson visited Washington and established that the WPB estimates agreed with his own.⁹ An economist at the WPB followed up this visit with a letter to Samuelson in which he endorsed Samuelson's view that $23 billion was the “rock bottom” estimate of the expected fall in national income—it being based on a multiplier of only 1.5, when actual multipli­ers were in the range 1.6 to ι.8.¹0 The paper in which Samuelson developed this estimate began by contrasting the rosy optimism of the economists who were forecasting a national income of $200 billion, with the view of busi­nessmen who could not see how their production could be higher than it had been before the war.¹¹ What the optimist forgets, Samuelson explained, was that high wartime levels of income were sustained by “the most colossal levels of deficit spending which any economy has ever experienced.”¹² He sounded like a traditional business cycle theorist (such as Schumpeter) when he argued that by most standards this was an artificial boom—justified by the national emergency, but a boom nonetheless.

Samuelson offered a nontechnical explanation of how the problem should be analyzed. Spending could be divided into “relatively stable magnitudes whose behavior has been and can be expected to be passively and predictably related to other magnitudes” and “relatively autonomous elements” that need to be explained in terms of “the unique circumstances expected to prevail in

a. See chapter 21 this volume. the period ahead.”¹³ Consumer expenditure on nondurable goods, Samuelson believed, fell into that category, each dollar of disposable income produc­ing a 60^“ rise in spending. Given taxes, $1 extra income would result in a rise of 70^“ or 80^“ in disposable income. In contrast, spending on consumer durable goods would be limited only by supply, because people had not been able to maintain their stocks of automobiles and other goods during the war. Federal spending and private investment were also relatively autonomous. These could not be forecast on the basis of past behavior or with variables like business saving, which was liable to enormous fluctuations. The skill of the forecaster lay in working out how these autonomous factors would change. Only after explaining this did he write down a mathematical model of the multiplier. Using these figures, he estimated that a $15 billion cut in federal spending would reduce national income by $25 billion, and cause unemploy­ment of 4 million people.

By October 1944, the pressure on Samuelson’s time was less, presumably because his work in the Radiation Laboratory had eased. Anticipating that it would reduce even further, he told Goldsmith that he would now like to do some consulting work.¹⁴ One of the reasons he gave was that he had real­ized it was “hopeless to work in isolation” when analyzing contemporary economic events, a situation that doing consultancy work would remedy.

This was not to be, and the bureaucratic formalities, including getting approval from Samuelson’s draft board, were not completed until March 1945. Formal notification of the appointment came on May 7, and he was sworn in the next day—the day when war officially ended in Europe—by which time the Bowman report on science policy had been sent to Bush.¹⁵ His salary, the daily equivalent of $6,500 a year (raised to $7,175 two months later), was significantly less than Samuelson had told Goldsmith other agen­cies had offered him, but was still much higher than his MIT salary and what he had been receiving for the NRPB work.¹⁶ It seems likely that one of the bureaucratic hurdles that Goldsmith was trying clear was to get Samuelson assigned a higher grade than the Civil Service Commission believed appro­priate for someone of his age and experience.

Goldsmith wanted Samuelson not as a mathematical specialist but as a general economist.¹⁷ Goldsmith had to prepare a report on reconverting industry to peacetime activities, and he proposed to do this by reviewing what had happened to industry in the six years since 1939, covering the uti­lization of plants, the distribution of industrial production between indus­tries, and changes in the distribution of industry across regions.¹⁸ This was the basis for assessing the size and structure of the economy after recon­version. The transition would be considered in two stages—from V-E Day (May 8, 1945) to V-J Day (still unknown), and from V-J Day to “postwar normality.^,'^19,b The report would also assess the arguments for continued gov­ernment control over industry during this transition.

The way Samuelson proposed to tackle the problem involved estimating the effects on industry of different levels of income (on which he had worked at the NRPB), but it was also necessary to take into account how this income was being generated. Estimating full employment income at $170 billion in 1943 dollars, he explained that this level of income could be achieved through high private investment, high consumption, or government sup­port: each of these would produce different patterns of industrial output. Samuelson also attached importance to regional imbalances, in that wartime industrial expansion had caused certain cities to reach sizes that might be unsustainable in the long term. Interestingly, he also argued that supply-side factors—bottlenecks affecting resources needed for recovery—could have adverse effects on aggregate demand and that there were financial problems that needed attention. His was a world in which aggregate demand might fail to generate the requisite level of output and one in which money and finance mattered.

A week after his appointment, Samuelson spent two days in Washington. A week later, Goldsmith sent him forecasts of GDP and its components through 1947, asking for a reaction by telephone, so urgent was it to get his input quickly.²¹ The method Goldsmith and his colleagues had used was consistent with the views Samuelson had expressed in his earlier memoran­dum. Nondurable consumption was modeled using estimates of income and assumptions about tax rates and the propensity to consume. Nondurable spending was assumed to be constrained by productive capacity, in the belief

b. V-E and V-J refers to the Allied Forces' victory in Europe and later in Japan. that accumulated savings would generate sufficient demand. These assump­tions meant that their forecasts, of a fall from $206 billion in 1946 to around $156 to $160 billion in 1947, changed little across the different scenarios for reductions in military spending that they were considering.

Samuelson’s response to these figures is not on record, but a few days later he wrote a memorandum to Goldsmith on how to estimate industrial pro- duction.²² The 1941 ratio of industrial production to national income would probably provide an upper limit; plotting the ratio over recent years and extrapolating would probably give a low figure. Alternatively, Samuelson argued that it might be possible to make use of studies undertaken by the Bureau of Labor Statistics (BLS) and Wassily Leontief’s input-output tables for 1939. Input—output tables made it possible to work out how much industrial production would be generated by a given level of consumption, and BLS data could be used to work out the relationship of consumption of different types to national income. If productivity increased and work­ing hours could be forecast, this would give estimates for employment. At the very end of his memorandum, Samuelson noted that he was assuming there would be sufficient outlets for investment, and that a way of checking would be to compare the estimates of productive capacity in different sec­tors, derived by using Leontief’s input—output table with those produced by the WPB.

These communications show that Goldsmith was relying on Samuelson not only as an authority on the multiplier and the theory of income deter­mination but also as a source of technical advice on empirical methods. For example, he asked for his views on of some estimates of the capital stock— “industrial facilities,” or capital plant and equipment for manufacturing and processing—produced by the Twentieth Century Fund. The problem was that regression analysis did not produce equations that fit the historical data very well (the correlation between capital stock and GNP was not high).²³ There was also uncertainty about how much of the capital investment made during the war would be useful in peacetime, and the effects of direct con­trols on industry were hard to estimate. Samuelson’s view was that in an economy with full employment, such problems made it difficult to be sure about the value of the multiplier.²⁴ There was no reason to believe that the multiplier would be the same when military spending was reduced as it had been during the expansion of 1943—44. Such analysis, taking account of problems on the supply side, was a long way from the simplistic focus on aggregate demand of which Keynesians such as Samuelson were later accused.

Over the summer, Goldsmith, having problems getting the report done in time, turned to Samuelson for something else:

Meanwhile, there is one thing which I should like you to do. I want to include in the report a fairly short description, running to something like five pages, of the basic mechanism through which, under present conditions, inflation or deflation (or more generally changes in aggre­gate demand) operates and of the main forces involves [sic].

I would regard such a section mostly as a venture in the economic education of the top administrators. Therefore, it would have to be written in non-technical style. However, this piece should go well beyond the over-simplifying and elementary statements that are gener­ally given to the lay public. Some simple algebra might well be used. What is more important, the limitations to which the broad state­ments usually made are subject and the time lags involved in the pro­cess should be brought out. Similarly, some indication should be given which processes are cumulative and which are not. In other words, what I am after is a statement on the one hand understandable to administra­tors who have a reasonable knowledge of economic facts though not of theory, but on the other hand fully consistent with the results of up to date economic analysts. This may be an order almost impossible to fill but I think you are as likely to handle it successfully as anybody.²⁵

Whereas he had previously called upon Samuelson to advise them on how to handle data, he was now asking him to draw on his pedagogic skills.

Goldsmith continued to turn to Samuelson for technical advice, asking for comments on forecasts of GDP up to 1950 and estimates of GDP for 1945— 47 based on two very different assumptions about the end of the war in the Pacific—September 30, 1945, and December 31, 1946; the possibility that the war might end in 1945 was now being taken seriously.²⁶ Samuelson sug­gested that rather than “substantially full employment” during the transition, it would be better to assume the most optimistic figures for the minimum level of unemployment (probably 4 percent of the labor force) and to allow for some shrinkage of the labor force as “extraordinary war workers” stop being available for paid work. He also paid great attention to how workers would respond to unemployment. High unemployment would induce some people to enter the labor force and would discourage others, and it was hard to tell which effect would dominate. Many women would be pushed out of the labor market and would not seek employment, unless their husbands and fathers were unemployed. There were also problems relating to hours worked that the forecasts had not taken into account. Samuelson also expressed views on the estimates of agricultural output and productivity in the service sector.²⁷ He was expressing views on all aspects of the modeling process.

In the meantime, Goldsmith had raised a further point concerning the expository pages he wanted Samuelson to write.²⁸ It had become fashionable to argue that inflation and deflation could occur at the same time, mak­ing it easier to reconcile the views of different government officials.^c He thought this sufficiently important that he wanted Samuelson to say some­thing about it—whatever conclusion he came to—that could be included in his own introductory chapter.²⁹ The pages that Samuelson sent, headed “Determinants of National Income and Inflation,” explained that while econ­omists were not agreed on their forecasts, there was widespread agreement for the need to focus on saving and investment. This was not a new method of analysis, and it could explain why there was a boom during the First World War, why there was a boom of limited duration in the 1920s, and why the Great Depression gave way to wartime prosperity. He then tackled the objec­tions that were raised to this method of analysis. It was right to argue that production was what mattered, but as most business people realized, “the factors which determine over-all production are to a large extent financial.”³⁰ He then explained why government had to be involved. Business people had no authority to use their shareholders’ money to stem the tides of prosperity and depression, and in any case they did not have the “bottomless purse” that was needed. Samuelson hammered away at the idea that both conservatives and liberals agreed the free enterprise system would not automatically gener­ate the right level of effective demand—too much demand and there would be inflation, too little demand and there would be depression. After a digres­sion on “the mysterious multiplier,” which he suggested could be omitted on a first reading, he explained why saving had to be matched by some form of spending. If this did not happen, income would not flow back to businesses, which would then have to cut back on production. The “offsets to saving,” as he called the required spending, could include private investment, but also foreign lending (exports minus imports), autonomous increases in consump­tion, or government spending. Because private investment was irregular, government fiscal policy had to act as “a stabilizing fly-wheel.”³¹

The inflationary gap was based on the argument that “[i]f ‘full employ­ment’ were as sharply marked a condition as the edge of a billiard table,”

c. As Samuelson was to point out, this remark raises questions about how inflation and deflation are defined.

rises in demand would have no effect on prices until full employment was reached. If demand rose past the point of full employment, output would not be able to rise any further and there would be a gap between demand and output that could only be reduced by rising prices. However, the story would not end here, for rising prices would raise corporate profits, and eventually wages would rise, resulting in an inflationary spiral. This would have no end, because it involved competition for a limited quantity of goods.

This can be put in an even simpler way. At full employment, there is 100 percent of output to be distributed. But if total effective demand is very high, consumers may want 90 percent of national product, govern­ment 20 percent, and capital producers, 40 percent. This adds up to a total of 150 percent, which is an impossible situation. Each group can­not get all that it wishes. But it can bid prices up in the attempt. The only thing that keeps prices from rising infinitely fast, is the fact that there is necessarily some delay between the receipt of income and its expenditure. By the time people go to spend their money, they will find it will buy a good deal less than 90 percent of the national income.³²

To reinforce the point, he explained that “the attempt of each group to better itself during an inflation by raising prices has been compared to the comic spectacle of a fat man, who in bending down to pick up his derby hat, only succeeds in giving it another kick in front of him.”^d

Responding to Goldsmith’s request to write about the problem, he explained that it would be possible to see simultaneous inflation and defla­tion, for prices did not have to move in the same direction. Equally impor­tant, whereas inflation always meant rising prices, deflation was used to refer both to falling prices and to depression; that is, prices could rise even when there was considerable unemployment, as had happened in 1937.

On August 6 and 9, 1945, atomic bombs were dropped on Hiroshima and Nagasaki, respectively, bringing the war to an abrupt end. Goldsmith wrote to Samuelson on the day of the Nagasaki bombing, saying that fin­ishing the report had become even more urgent and asking him come to Washington to discuss it.³³ The WPB was wound down and Samuelson’s appointment was terminated at the end of September, but Goldsmith con­tinued to approach him for advice. Goldsmith had written a nontechnical

d. Given the widespread accusation that Keynesian economics, up to the 1960s, ignored the effects of expected inflation, and feedback from inflation onto wage increases, it is important to note that Samuelson was thinking in this way at this time. account of the topic that was based purely on Samuelson’s memorandum (he explained he had been too busy to read anything else), but he feared it made demands on the reader that were as great as those made by Samuelson’s paper.³⁴

An interesting feature of the draft for which Goldsmith sought Samuelson’s advice was that he avoided talking in terms of saving and investment, which he thought potentially misleading, and adopted his own fourfold classification of transactions.⁶ Goldsmith’s alternative clas­sification reflects the thinking of someone who had worked extensively on the national accounts, wanting an exhaustive scheme for classifying expen­ditures, and seeking a clear rationale for putting particular items in one category rather than another. He took many ideas from Samuelson, and in places used his wording, though he made extensive modifications of his own, such as paying much more attention to the distinction between plans and what ended up being recorded in the national accounts. He was not just recycling Samuelson’s material; rather, he was trying to develop it in theoretically innovative ways, which was why he wanted Samuelson’s reassurance that it was “unobjectionable from the point of view of the present status of professional discussion.”³⁵ This shows that, although the Keynesian terminology of saving and investment was becoming widely used, it was still not taken for granted. Unfortunately, Goldsmith needed a quick response and Samuelson’s response, presumably over the telephone, is not recorded.³⁶

Samuelson’s work for the WPB shows a set of interests that contrasts strongly with the conventional image of a mathematical economist and Keynesian theorist. The concept of the multiplier and the analysis of aggre­gate demand were central to the WPB’s forecasts, but Samuelson’s inter­ests lay as much in the technical, statistical problems and the institutional details as in theoretical factors. Moreover, his command of these issues was such that his advice was respected even by someone who had spent his whole career on problems related to national income accounting. Goldsmith was also pushing him to write in a style accessible to a lay audience, a skill that he was beginning to develop through forays into journalism.

e. E-transactions were ones made out of current income and flowed back into the income stream. S-transactions (so named because saving was the prime example) were made out of current income and did not flow back into the flow of income. I-transactions (named after investment) were not made out of current income but entered the income flow, while N-transactions were completely neutral.