Slouching Towards Utopia?: The Economic History of the Twentieth Century

-XIV. The Great Slump-

J. Bradford DeLong

University of California at Berkeley and NBER

August 1996

In 1929 many economists saw business cycles as unfortunate but unavoidable elements of economic development. Cycles in GNP and employment were principally investment cycles. The value of investment today depends heavily on future interest rates and growth rates. With an uncertain future, there will inevitably be times when new information shows that future growth has been overestimated, and that too much capital-or the wrong kind of capital-has been put into place. The economy must then shift resources from investment for the future to present consumption. This shifting of resources out of the investment sector entails frictions that temporarily reduce production and employment.

In this monetary overinvestment view, demand stimulation is counterproductive. In a recession it keeps resources from shifting out of the investment sector where their marginal product is less than their resource cost, increases the overhang of excess capital, and thus deepens and prolongs the subsequent depression; in a boom it increases the chance that in the future new information will reveal that recent investment was unproductive and so trigger a recession. Proper policy is "neutral" so as not to disturb the market's balancing of savers' required real rate of return and investors' expectations of the value of new capital. Proper policy can moderate the business cycle, especially that part of the business cycle that arises because the money supply is too elastic in response to transitory demand shocks; but proper policy cannot be expected to come anywhere near to smoothing out the cycle.

The "liquidationist" monetary-overninvestment view of business cycles collapsed in the Great Depression. While the monetary overinvestment view provided a framework for economists to analyze the busts of the nineteenth and early twentieth century, its interpretation of the Great Depression was found unconvincing. Periods of high unemployment lasted not for months or years but for decades. They seemed too persistent to be attributed to the frictions that resulted as the market reallocated productive resources away from what were now seen as low value goods. Although workers seemed willing to labor hard for low implicit wages-George Orwell, for example, saw "...several hundred men risk their lives and several hundred women scrabble in the mud for hours... searching eagerly for tiny chips of coal... more important almost than food.," while "all around... are the slagheaps and hoisting gear of the [idle] collieries" at which they used to work.

In response to the high persistence of unemployment in the interwar years, economists abandoned the idea that business cycles were the economy's best feasible response to inevitable shocks to present circumstances and expectations about the future, and that the Great Depression had been generated by the largest such shock ever seen. Instead, they turned to alternative--Keynesian--approaches to explain the persistence of high unemployment, even though these alternative approaches were not so much theories of business cycles as policy recommendations accompanied by promises that supporting theories would be constructed later.

Some nations-Scandinavian countries that abandoned the gold standard early-among the European, North American, and Australasian set analyzed in Eichengreen and Hatton experienced the Great Depression as little more than an ordinary recession, albeit in some cases beginning from a position of relatively high unemployment in 1929. The collapse of international trade in the 1930's idled resources in specialized export industries, but for countries that had abandoned the gold standard early domestic manufacturing took up the slack and returned GNP and employment to relatively high levels by the middle of the decade. These fortunate nations experienced the Great Depression as more-or-less another episode of "normal" cyclical unemployment in response to a large shock, in this case the world market's signal that export sectors were too large.

Other countries-largely nations like the United States and France that remained on the gold standard beyond 1930-31-were not so fortunate. Their unemployment rose to and remained at levels that seemed too high to square with the normal mechanisms of standard business cycles. There experience seems to have been an important key factor leading economists away from "monetary overinvestment" theories and toward "underemployment semi-equilibrium" theories.

Were economists right to judge that unemployment was too high and too persistent to be interpreted as the best feasible redeployment of resources in response to a large shock? Aggregate data-fragile as they are-suggest that they were. Speeds of adjustment in the Great Depression do seem much slower than in other recessions. For example, consider the time-series properties of employment in the United States and the United Kingdom. An error-correction model of the gap between actual and usual United States manufacturing employment estimated over a pre-Great Depression 1900-1929 period finds an error correction coefficient of about 0.65 per year; the economy closes 97% of the gap back to usual employment in three years. But the same regression run over a 1924-1939 period that includes the Depression shows a different picture: the error-correction coefficient is about 0.20; the economy closes only half of the gap in three years. For the United Kingdom, the analogous error correction coefficients are about 0.55 and 0.20, respectively. These correspond to an economy that closes nine and fifty percent of the gap to usual employment in three years, respectively.

These regressions are useful as descriptions of the data only. But as descriptions they are powerful. The self-regulating mechanisms of the market appear to work less well in the Great Depression; speeds of adjustment seem to slow down. Alternative propagation mechanisms that give increased persistence to downturns appear to be at work. The search for explanations of the Depression outside the compass of pre-1929 business cycle theory thus appears well-motivated.





Gap between Lagged

Sample Actual and Full Present Employment

Period Employment In_ation Growth R2 SEE


1900-29 .700 .681 .60 .053

(.124) (.155)

1900-29 .677 .709 -.091 .59 .054

(.148) (.167) (.144)

1924-39 .196 1.516 .71 .053

(.082) (.272)

1924-39 .162 1.674 -.157 .71 .053

(.090) (.321) (.166)


1900-14 .563 .071 .17 .008

(.267) (.126)

1900-29 .512 .105 .31 .014

(.148) (.041)

1900-29 .594 .102 -.167 .30 .014

(.178) (.043) (.193)

1924-39 .158 .069 .41 .006

(.082) (.033)

1924-39 .203 .131 -.688 .59 .006

(.079) (.036) (.266)

Each paper in Eichengreen and Hatton presents its own distinctive point of view on the question of why Depression unemployment in the country it studies was or was not extraordinarily high and persistent, and as a result the book is sprawling-as is inevitable in conference volumes. Authors stress that France appears very different from Britain. They stress that Australia's labor market adjustment process bears little resemblance to that of the United States. Readers must extract the common themes and contrasts of the interwar unemployment experience by themselves.

It is helpful to group the explanations for why Depression-era unemployment was so high and lasted so long along two axes: government-generated unemployment, and a shadowy and somewhat residual category of market failure-generated unemployment. There are two candidates to take the blame for the persistence of unemployment during the Depression: the government, and the market.

Government-generated unemployment was widespread. In Britain some unemployment (although a small share during the peak unemployment years of the early 1930's) was surely generated by the government's unemployment insurance system. Thomas cites Eichengreen's earlier work, which presented a best estimate that some two or three percentage points of unemployment in 1929-32 could be attributed to the operation of the relief system. Thomas attributes some unemployment among secondary workers and unskilled young men with large families to the "OXO" system in which firms would systematically rotate two platoons of workers between time at work and time receiving unemployment benefit, thus turning unemployment insurance into a highly-subsidized work sharing scheme. Men receiving the standard unemployment benefit in February 1931 had on average experienced 8.6 different spells of employment during the past year, working an average of 151 days.

Given such rapid turnover it is not at all implausible to argue that the availability of unemployment benefit, even with relatively low replacement rates, allowed workers to remain in labor market positions in which they were employed only half the time instead of migrating to some other industry. Thus Thomas' essay tends to strengthen belief in Benjamin and Kochin's (1978) argument that the underlying layer of excess British unemployment-not the unemployment that prevailed in the deep slumps of 1922 and 1932 but the unemployment that prevailed in the relative booms of 1928 and 1937-may well have been maintained by the government's social policy.

Robert Margo's essay on the United States looks at the situation as described in the 1940 census at the very end of the Depression. At that time some 11.1% of U.S. heads of household were counted as unemployed, of whom almost half-4.9% of all heads of household-held relief jobs. Darby (1974) was the first to argue that the government had managed to create a situation in which those on relief found themselves with little incentive to register their labor supply on the private-sector job market and yet were doing little socially productive work. Margo largely supports Darby, thinking that relief jobs were attractive to many in spite of their low levels of relief wages relative to average private sector wages. Relief jobs were secure and required little skill. The risk-averse or the lesser-skilled might well have found that their best option was to stay on relief jobs, and be counted as unemployed, rather than take even an immediately available private sector job.

Robert Salais' essay on France argues that the development of bureaucratic mechanisms of relief for dealing with unemployment in large part called forth the phenomena they were to address. The existence of programs that would not offer relief except to the wholly unemployed put pressure on the economy to shift its pattern of labor relations so that it would adjust to lessened demand and slack not by reducing hours or returning workers to agriculture but by creating the "unemployed."

In all of these cases there is no clear alternative way of organizing the unemployment insurance system that would have been a clearly better policy. A good society should offer support to those blocked from earning their wages in the market. And a well-functioning economy should create incentives for the unemployed to strongly register their excess supply of labor in the market. These two goals are inevitably in tension. The inescapable problem was that relief payments were too high for the short-term and too low for the long-term unemployed, and that there was no good way to structure relief programs to tell these two groups apart ex ante. William Beveridge was among the first to lay out the policy dilemma: the long-term unemployed "need... more money rather than less than those who have had short periods of unemployment. Yet they can hardly be given more money without... [creating an incentive] to settle down into permanent unemployment." Moreover, few of the long-term unemployed can "escape physical and psychological deterioration through long idleness."

In view of the general Keynesian orientation of the editors Eichengreen and Hatton, the fact that their contributors argue that government-generated unemployment played a significant role is interesting. My reading of Eichengreen and Hatton led me to place more weight on Benjamin and Kochin's and Darby's accounts of why unemployment was persistent in the interwar period. Nevertheless, a large part of the puzzle remains: Thomas, Margo, and the others note that slightly less than half of mid-1930's unemployment was concentrated among long-term unemployed who did not take advantage of subsidized relief-work schemes.

This form of unemployment, principally long-term and somewhat of a residual category is, in the eyes of Eichengreen and Hatton and their contributors, the key to the persistence of the Depression. Long-term unemployment was strongly present in those countries that suffered worst from the Depression, including non-European nations like Australia (compared to the U.S. by R.G. Gregory, V. Ho, L. McDermott, and J. Hagan), Canada (analyzed by Alan Green and Mary MacKinnon), and the United States and European nations like Britain, Germany (discussed by Dan Silverman), Italy (discussed by Gianni Toniolo and Francesco Piva), and the gold block nations of France and Belgium (analyzed by Martine Goosens, Stefaan Peeters, and Guido Pepermans). Of these only Germany achieved a strong recovery from the Depression in the 1930's.

Long-term unemployment means that the burden of economic dislocation is unequally borne. Since the prices workers must pay often fall faster than wages, the welfare of those who remain employed frequently rises in a depression. Those who become and stay unemployed bear far more than their share of the burden of a depression, as Bernard Harris' discussion of unemployment, insurance, and health in Britain shows. Moreover the reintegration of the unemployed into even a smoothly-functioning market economy may prove difficult, for what employer would not prefer a fresh entrant into the labor force to someone out of work for years? The simple fact that an economy has recently undergone a period of mass unemployment may make it difficult to attain levels of employment and boom that a luckier economy attains as a matter of course.

Indeed, long-term unemployment seemed stubbornly resistent to activist policies. Stimulative policies may well have helped cushion some countries' slide into the Great Depression, but once an economy had fallen deeply into the Great Depression, devalued exchange rates, prudent and moderate government budget deficits (as opposed to the deficits involved in fighting major wars), and the passage of time all appeared equally ineffective ways of dealing with long-term unemployment. Highly centralized and unionized labor markets like Australia's and decentralized and laissez-faire labor markets like that of the United States did equally poorly in dealing with long-term unemployment. Fascist "solutions" were equally unsuccessful, as the case of Italy shows, unless accompanied by rapid rearmament as in Germany.

Eichengreen and Hatton's contributors have no clean answers to the question of why the private sector could not find ways to employ its long-term unemployed. The very extent of persistent unemployment in spite of different labor market structures and national institutions suggests that theories that find one key market failure responsible should be taken with a grain of salt.

Perhaps the source of our puzzlement is our assumption that the economy's laws of motion are more or less linear-that twice as many excess unemployed should induce additional hiring and redeployment of labor so that the number of excess unemployed drops twice as fast. But should we be surprised that the long-term unemployed do not register their labor supply proportionately strongly? They might accurately suspect that they will be at the end of every selection queue. In the end it was the coming of World War II and its associated demand for military goods that made private sector employers wish to hire the long-term unemployed at wages they would accept.

ß ß ß ß ß

The Credit Anstalt:

ß ß ß ß ß

The Fall of the Pound Sterling:

ß ß ß ß ß

Debt and De_ation in the United States:

ß ß ß ß ß

Scandinavia, Latin America, Japan:

ß ß ß ß ß

Nazi Germany: