Economic History Association
The Microeconomics of Depression Unemployment
Author(s): Robert A. Margo
Source: The Journal of Economic History, Vol. 51, No. 2 (Jun., 1991), pp. 333-341
Published by: Cambridge University Press on behalf of the Economic History Association
Stable URL: http://www.jstor.org/stable/2122578
Accessed: 05/04/2010 15:09
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=cup.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
Cambridge University Press and Economic History Association are collaborating with JSTOR to digitize,
preserve and extend access to The Journal of Economic History.
http://www.jstor.org
The Microeconomics of Depression
Unemployment
ROBERT A. MARGO
Microeconomicevidence reveals that the incidence and durationof unemployment in the 1930s varied significantlywithin the labor force. Long-termunemployment, which was especially high by historical standards, may have been
exacerbatedby federal relief policies.
I. INTRODUCTION
Between 1931 and 1940 the aggregateunemploymentrate, including
persons holding so-called work relief jobs, never dipped below 14
percent. Why were unemploymentrates so high for so long? How can
we reconcile persistently high unemploymentwith the behavior of real
wages, which rose throughoutmuch of the decade? Answers to these
questions have been pursuedfrom a variety of macroeconomicperspectives, but as yet there is little consensus on the relative merits of the
differenttheories.'
This article is a preliminaryreport on a new project dealing with
macroeconomic aspects of labor market behavior during the Great
Depression. The project is motivated by a belief that further insights
into the 1930s economy can be gleaned from disaggregateddata, which
have been relativelyneglected by modernscholars.2Here I examine the
characteristicsof the unemployedand the durationof unemploymentin
the 1930s. The unemployed were drawn disproportionatelyfrom the
low-wage portion of the labor force. There was an unusually large
fraction of persons unemployed for over a year, which kept aggregate
unemployment rates high. Federal relief policies designed to aid the
long-termunemployed may have exacerbated the problem of reducing
their ranks. The implicationsof these findingsfor understandingmacroeconomic events are briefly addressed here.
The Journal of Economic History, Vol. 51, No. 2 (June 1991). ? The Economic History
Association. All rightsreserved. ISSN 0022-0507.
The authoris Associate Professorof Economics, VanderbiltUniversity, Nashville, TN 37235,
and ResearchAssociate, NationalBureauof EconomicResearch.
I am gratefulto CharlesCalomiris,Susan Carter,Stanley Engerman,Thomas Ferguson, Lou
Galambos,Helen Hunter,and MaryMacKinnonfor their helpfulcomments.
' Martin Baily, "The Labor Market in the 1930s," in James Tobin, ed., Macroeconomics,
Prices and Quantities: Essays in Honor of Arthur Okun (Washington, DC, 1983).
2 Exceptionsare Ben Bernanke,"Employment,Hours, and Earningsin the GreatDepression:
An Analysisof EightManufacturing
Industries,"AmericanEconomicReview, 76 (Mar. 1986),pp.
82-109; and John Joseph Wallis, "Employmentin the GreatDepression:New Data and Hypoth-
eses," Explorations in Economic History, 26 (Jan. 1989), pp. 45-72.
333
334
Margo
TABLE 1
CHARACTERISTICS OF THE UNEMPLOYED: SELECTED SURVEYS, 1930-1938
(A) Unemployment Rates by Age (%)
Pennsylvania, 1934
United States, 1937
15-19
20-24
25-34
35-44
45-54
55-64
60.0
26.0
36.0
18.0
22.0
13.0
19.0
13.0
22.0
16.0
27.0
19.0
(B) Unemployment Rates by Occupation and Schooling
Michigan, 1935
(Percentage
Unemployed,
Both Sexes)
6.1
10.3
17.0
14.8
26.2
Professional/Managerial
Clerical
Skilled
Semiskilled
Unskilled
Serviceb
Government
Philadelphia, 1936
(Percentage
Unemployed,
Males)
Difference in
Schoolinga
12.1
21.3
1.4
0.9
29.2
2
43.9
29.0
5.1
0.3
1.6
0.2
0.2
(C) U.S. Industry Unemployment Rates (%)
Industry
Oct. 1930
Mar. 1933
May 1938
Agriculture
Manufacturing
Construction
Transportation
Public Utilities
Trade/Finance
Services
8.2
17.7
13.1
9.1
4.3
4.7
3.7
14.5
40.3
73.3
39.7
29.0
22.9
16.2
10.2
27.7
55.0
37.1
27.2
18.5
1.8
(D) Duration of Unemployment Among Currently Unemployed (%)
Buffalo
Massachusetts
Weeks
1929
1930
Months
1934
<10
10-19
20-29
30-39
40-51
-52
68.4
12.3
6.2
3.1
0.7
9.3
53.2
17.9
14.3
7.9
5.6
21.1
0-3
3-6
6-12
?12
11.3
12.1
14.0
62.6
This is the difference in median years of schooling between employed and unemployed males.
Service includes domestic and personal service.
Sources: New York Department of Labor Special Bulletin No. 167, "Unemployment in Buffalo,
November 1930" (Albany, 1930); Palmer, "Employment and Unemployment"; and U.S. Federal
Works Agency, "Report on Public Assistance," typescript (New York, 1939).
a
b
II. DATA
Table 1 contains selected evidence on the characteristics of the
unemployed and on the duration of unemployment during the Great
Depression. Unemploymentfollowed a U-shaped pattern with respect
to age. Those out of work were also more likely to be blue-collar
Microeconomics of Depression Unemployment
335
workers or the unskilled. Within occupations, the unemployed had
fewer years of schooling. Workers in construction and durable goods
manufacturingexperienced higher unemploymentthan did workers in
the service sector.
Althoughinitiallyquite low, long-termunemployment(of over a year)
became common as the downturn progressed. Data from 1930 for
Buffaloreveal that fully 21 percent of the city's unemployedlabor force
had been out of work for over a year, comparedwith 9 percent in 1929.
According to the state census of 1934, 63 percent of unemployed
persons in Massachusettshad been out of work a year or longer. Similar
levels of long-termunemploymentprevailedin Philadelphiain 1936and
1937.3
Detailed evidence on unemploymentat the end of the Great Depression is available in the public use sample of the 1940census.4 The data
I examine here refer to nonfarm males (wage and salary workers)
between the ages of 14 and 64. Informationis given on employment
status in the survey week (March24 to 30, 1940), the numberof weeks
of unemployment for those unemployed in that week, and weeks
worked in 1939. Duration of unemployment refers to the number of
weeks since the person last held a private-sectoror nonrelief governmentjob-or, if he never held such a job, the numberof weeks since he
began looking for work. Persons "employed" by the Works Progress
Administration(WPA) or related federal or state relief agencies were
separatelyidentified. Weeks worked in 1939were reportedfor persons
on work relief, even thoughthey were officiallycountedas unemployed.5
Evidence on personal characteristicsby employmentstatus is shown
in section A of Table 2. Approximately 15 percent of the sample was
officially counted as unemployed. Slightly more than a third of the
unemployed held work relief jobs. As is shown in Table 1, unemployment varied according to age and other personal characteristics. The
unemployedwere older, single, foreign-born,urban,concentratedin the
Northeast, and had completed, on the average, 1.5 fewer years of
schooling than the employed. They were far more likely to be unskilled
laborers working in constructionthan were the employed.
Persons with work relief jobs differedfrom other unemployed workers. They were younger;more likely to be married,nonwhite, and rural;
Gladys Palmer, Employment and Unemployment in Philadelphia in 1936 and 1937 (Philadel-
phia, 1938).
4 U.S. Bureauof the Census, Census of the Population, 1940: Public Use Microdata Sample
(Washington,DC, 1983). The 1940 census sample is arrangedinto 20 subfiles, each a random
sample of the population.My analysis is based on the first subfile.
5 Michael Darby, "Three and a Half Million U.S. Employees Have Been Mislaid: Or, an
Explanationof Unemployment,1934-1941,"Journal of Political Economy, 84 (Feb. 1976), pp.
1-16); see also J. Kesselmanand N. E. Savin, "Three and a Half MillionWorkersNever Were
Lost," EconomicInquiry, 16 (Apr. 1978),pp. 176-91.
336
Margo
TABLE
2
CHARACTERISTICS OF UNEMPLOYED NONFARM MALES, AGES 14-64 (MAR. 1940)
(A) By Employment Status
Unemployment
Rate (%)a
Employed
Unemployed
Work Relief
Sample
Darby
BLS
9.4
14.7
Age (%)
14-24
25-34
35-44
45-54
55-64
Married (%)
Nonwhite (%)
Foreign (%)
Years of schooling
16.0
31.3
25.0
18.1
9.6
67.9
7.7
13.7
9.5
13.2
22.7
25.6
22.4
16.1
56.0
7.4
19.5
8.1
33.6
29.8
28.1
17.3
11.2
73.0
16.2
11.2
7.6
8.0
7.0
9.6
11.4
14.7
7.8
8.6
13.1
12.9
12.1
15.5
16.4
20.5
13.6
19.2
17.3
Location (%)
Northeast
Midwest
South
West
Urban
32.3
28.1
25.8
13.8
76.2
44.6
26.1
16.4
12.9
79.9
26.5
33.7
23.0
16.8
67.3
12.7
8.7
6.2
8.8
9.9
17.0
15.0
11.1
15.2
14.6
Occupation (%)
Professional/Technical
Managerial
Clerical/Sales
Skilled
Semiskilled
Service
Unskilled
6.4
8.0
17.3
19.1
25.3
12.8
11.1
1.7
2.6
10.9
23.0
25.9
10.6
25.3
4.1
2.6
7.1
15.3
15.8
5.1
50.0
2.9
7.0
16.9
22.0
11.7
2.9
25.6
17.0
18.8
7.5
2.0
59.7
5.1
8.1
5.1
13.7
6.5
5.3
4.6
9.4
13.2
7.8
4.6
1.7
0.9
4.1
3.6
4.6
3.1
4.6
3,146
348
196
Industry
Mining
Construction
Nondurable manufacturing
Durable manufacturing
Transportation/Communications/
Public utilities
Trade
Finance/Real estate
Personal services
Professional services
Government
N
and less likely to be foreign-bornor living in the Northeast.6 On the
average, they had even less schooling than other unemployedworkers.
6 The underrepresentation
of reliefjobs in the Northeastappearsto undercutthe WPA's belief
that the distributionof reliefjobs matchedthe distributionof population;see RobertA. Margo,
"InterwarUnemploymentin the United States: Evidence from the 1940Census Sample," in B.
Eichengreen and T. Hatton, Interwar Unemployment in International Perspective (Dordrecht, The
Netherlands, 1988),p. 350.
337
Microeconomics of Depression Unemployment
TABLE
2-continued
(B) Distributionof Weeks UnemployedAmong the CurrentlyUnemployed(%)
Weeks
Not on WorkRelief
On WorkRelief
Total
28.1
22.7
9.2
8.3
31.6
9.5
16.3
10.2
7.7
56.3
20.8
20.2
9.6
8.0
41.5
0 x<13
13 X < 26
x < 39
27
39 x < 52
52 C x
a
Darby counts persons on work relief as employed. BLS as unemployed;see Michael Darby,
"Threeand a Half MillionU.S. EmployeesHave Been Mislaid:Or, an Explanationof Unemployment, 1934-1941,"Journalof Political Economy,84 (Feb. 1976),pp. 2-5, for a discussionof both
procedures.Occupationsand industriesof persons on work relief refer to their usual occupation
and industry,as calculatedfrom "sample line" individuals;see the text.
Source: U.S. Bureau of the Census, Census of the Population, 1940: Public Use Microdata Sample
(Washington,DC, 1983),first file.
Instructions to census enumeratorsspecified that the occupation and
industry of persons on work relief pertain to their work relief jobs.
Usual occupation and industryof employment,however, were reported
for so-called sample line people. According to these data, half the
sample line persons on work relief reportedtheir usual occupation to be
unskilled labor; 59 percent claimed that their usual industry was
construction.
Section B of Table 2 shows the distributionof weeks unemployed in
the census sample. Among those not on work relief, the proportion
declined steadily as the number of weeks increased. Among persons
with work reliefjobs at the time of the census, the proportionrose from
0-13 weeks to 14-26 weeks and declined thereafter.Still, 32 percent of
unemployedpersons not on work relief had been out of work for a year
or longer; a staggering56 percent of persons on work relief had been
unemployed for more than a year. A more detailed breakdown (not
shown) reveals that 12 percent of persons with reliefjobs had not held
a private-sectoror nonreliefjob for at least five years.
Section A of Table 3 gives the distributionof weeks worked in 1939
among persons who reported65 or more weeks of unemploymentas of
March 1940, and who held a work relief job at the time of the census.
Fully half of the long-term unemployed on work relief worked 39 or
more weeks in 1939;23 percent worked the full year (52 weeks). Given
census definitions, they could not have been doing anythingother than
workingfull-timefor the WPA. The quantitativesignificanceof permanent WPA employment not only applies to 1940; the proportion of
permanentWPA employees was probably higher in 1937 and 1938 as
well. According to a WPA study, 57 percent holding relief jobs in
September 1937 held them continuously to February 1939. Fully 16
338
Margo
TABLE
3
THE LONG-TERM UNEMPLOYED ON WORK RELIEF
(A) Weeks Worked in 1939: Nonfarm Males on Work Relief in 1940 with 65 or More Weeks of
Unemployment
Weeks Worked
x < 13
Percentage
N = 370
13 c x < 27
28 c x < 39
39 c x < 51
52
21.4
22.2
25.6
23.8
7.0
(B) Characteristics of Long-Term Unemployed
Not on Work Relief a
On Work Reliefb
Mean
0a
Mean
Age (%)
25-34
35-44
45-54
55-64
Married
Nonwhite
Foreign
Years of schooling
0.21
0.29
0.20
0.21
0.54
0.10
0.26
7.59
0.41
0.46
0.40
0.41
0.50
0.30
0.44
3.57
0.19
0.31
0.28
0.13
0.78
0.12
0.07
7.40
0.39
0.46
0.45
0.33
0.41
0.32
0.26
3.41
Location
Midwest
South
West
Urban
0.26
0.07
0.06
0.94
0.44
0.26
0.24
0.24
0.39
0.21
0.13
0.73
0.49
0.41
0.34
0.44
N
196
179
(C) Coefficients of Employment Growth:c Logistic Regressions of the Probability of Long-Term
Unemployment
p8
Not on Work Relief
On Work Relief
-0.85d
-0.20
a
Not on Work Relief includes persons who were unemployed for 65 weeks or longer and who
worked zero weeks in 1939.
b On Work Relief includes persons who were unemployed 65 weeks or longer and who worked 39
weeks or more for the WPA in 1939.
c Employment growth means the percentage of growth in employment in the 1940 state of residence
from 1930 to 1940; see John Joseph Wallis, "Employment in the Great Depression: New Data and
Hypotheses," Explorations in Economic History, 26 (Jan. 1989), pp. 45-72.
d Significant at the 5 percent level.
Sources: For sections A and B, U.S. Bureau of the Census, Census of the Population, 1940: Public
Use Microdata Sample (Washington, DC, 1983), first file; for section C, Robert A. Margo,
"Interwar Unemployment in the United States: Evidence from the 1940 Census Sample," in B.
Eichengreen and T. Hatton, Interwar Unemployment in International Perspective (Dordrecht, The
Netherlands, 1988), p. 340.
percent of all persons with relief jobs in February 1936 worked
continuously to February 1939.7
The importance of permanent WPA employment suggests that the
' U.S. FederalWorksAgency, FinalReportof the WPAProgram,1935-1943(Washington,DC,
1946),p. 41.
Microeconomics of Depression Unemployment
339
long-term unemployed may have responded differently to improved
economic conditions, dependingon whetheror not they held work relief
jobs. Some preliminaryevidence of such differences is presented in
section B of Table 3. Aggregatedemandwas measuredby state-specific
indices of employment growth constructed by John Wallis.8 The hypothesis under investigation was that the probability of long-term
unemploymentwould be lower in states that had a more robustrecovery
in terms of employmentgrowth. The results indicatedthat employment
growthhad an insignificant(thoughnegative)effect on the probabilityof
holdinga long-termjob with the WPA; thus, the long-termunemployed
on work relief were not very responsive to improved economic conditions. Long-term unemployed not on work relief, however, were
responsive to improved economic conditions: the incidence of longterm unemployment, among persons not on work relief, was significantly lower in states with higher-than-averagerates of employment
growth.
III. DISCUSSION
In the traditionalKeynesian model, persistence of high unemployment is the outcome of sluggish adjustmentof nominal wages and of
insufficient aggregate demand. Although output prices fell, various
economic, legal, and institutional factors kept nominal wages from
falling in tandem to restore full employment. Firms did not follow "a
policy of aggressive wage cutting" when unemployment was high,
because such a policy hurt worker morale and the firm's reputation.9
New Deal legislation created a climate that made wage cutting difficult.
The upshot was high unemployment,which remainedhighuntil reduced
by the insatiable demands of a wartime economy in the early 1940s.
The notion that nominalwages in the 1930swere insufficientlyflexible
comes largely from aggregate wage series. Such series, however,
pertain to the employed labor force: they are not adjusted for the
characteristicsof the unemployedor for the durationof unemployment.
Because the unemployed were drawn disproportionately from the
low-wage portion of the labor force, and because such workers were
more likely to remainunemployedfor long periods of time (whether or
not they got on work relief), aggregate wage data overstate nominal
inflexibility.
More robust aggregatedemandin the late 1930swould have reduced
the incidence of long-term unemployment. My econometric results
suggest that this reduction would have bypassed some fraction of the
long-term unemployed who held work relief jobs. That many WPA
workers would have tried to keep theirjobs in the face of (counterfac8
Wallis, "Employmentin the GreatDepression," pp. 65-66.
9 Baily, "The LaborMarket,"p. 53.
340
Margo
tual) increases in aggregate demand may seem puzzling.10The WPA
published statistics on rates of assignment to and separationsfrom its
projects, and these suggest a turnoverrate similarto that in the private
sector.11WPAjobs were poorly paidcomparedwith private-sectorjobs.
Standard WPA turnover statistics, however, refer to projects, not
employment.12Many WPA employees were reluctant to leave and
search for a private-sectorjob because they believed their prospects of
findingequally stable work were nonexistent.13 As one WPA workerput
it,
Why do we want to hold onto these jobs? Well, you know, we know all the time
about persons who are on direct relief. . . just managing to scrape along....
My
advice, Buddy, is better not take too muchof a chance. Know a good thingwhen
you got it. 14
To encourageits workersto leave, the WPA promisedthose who took
private-sectorjobs "immediate reassignment" to work relief projects
"if they . . . lost theirjobs throughno fault of their own. ,15 Conscious
that it was incurringhigh political costs because so many of its workers
held relief jobs for so long, the WPA summarilydismissed 783,000 of
them in August 1939-only to re-employ 57 percent of them a year
later.16
By providing an alternativeto the employment search (which many
WPA workers perceived, correctly or not, to be fruitless), work relief
may have lessened downward pressure on nominal wages.17 The
demandfor labormay also have been alteredin subtle ways. In the early
twentieth century unemployment was egalitarian, spread relatively
evenly in its incidence among the workingclass. Unemploymentspells
were generallybrief;except amongthe infirm,long-termunemployment
was uncommon, even in severe downturns.18 As the labor market
evolved, so did the nature of unemployment. Firms increasingly
adopted bureaucraticmethods of hiringand firingworkers, along with
10 It is possible that the long-term unemployed on work relief differed from the nonrelief
long-termunemployedin unobservableways that would have made them less employable had
(marginal)improvementsin aggregatedemandoccurred.In this case, the WPAwouldnot have had
a causal effect on reemploymentprobabilities.I am gratefulto CharlieCalomirisfor this point.
1 Margo, "InterwarUnemployment,"p. 345.
12 U.S. FederalWorks Agency, Final Report, p. 32.
13 Turnoverstatistics in manufacturing
suggestthat the averagedurationof new jobs createdin
the 1930swas very short by historicalstandards;see Baily, "The LaborMarket,"pp. 28-31, 48.
14 Quoted in E. W. Bakke, The UnemployedWorker
(New Haven, 1940),pp. 421-22.
15 U.S. FederalWorksAgency, Final Report, p. 32.
16 Ibid., p. 41.
17 For a similarconclusion, see Baily, "The Labor Market,"p. 53.
18
Alexander Keyssar, Out of Work: The First Century of Unemployment in Massachusetts
(Cambridge,MA, 1986);and RobertA. Margo, "The Incidenceand Durationof Unemployment:
Some Long-TermComparisons,"EconomicsLetters, 32 (Jan. 1990),pp. 217-20.
Microeconomics of Depression Unemployment
341
seniority-basedwage scales.19The trend toward bureaucraticmethods
accelerated during the 1930s. The time was ripe for more careful
screening of job applicants, picking and retainingthe most productive
workers, and weeding out the rest. Those with few occupationalskills,
the elderly (who were expensive to rehire),and the poorly educated had
a difficulttime findingwork and ended up unemployedfor long periods
of time. Responding to opposition by organized labor, however, the
WPA made few efforts to upgrade the skills of the long-term unemployed or to improve their education. Workingfor the WPA may also
have stigmatized individuals, making it even more difficult for the
long-termunemployed to find a stable private-sectorjob.20
Ultimately, the dilemma of long-termunemploymentwas "solved"
by the massive increase in labor demandassociated with WorldWarLI.
What was significant about the war-relateddemand was not just the
magnitude of the increase, but its persistence. Faced with severe
shortages of both skilled and semiskilledlabor, employers were willing
to hire and train the long-term unemployed. Convinced that the increased demand for their services was more than transitory, these
persons were finally willing to give up WPA employment. Unemployment rates declined sharply, never again reaching the aggregatelevels
commonly observed in the 1930s.
IV. CONCLUSION
The unemployed of the 1930s were disproportionately low-wage
workers who, once the depression was under way, tended to be out of
work for a long time. A significant fraction of them ended up on
sustained work relief because, comparedwith the jobs that might have
been available to them in the private sector, WPA jobs seemed more
stable. Higher levels of aggregatedemandin the late 1930swould have
lowered the incidence of long-term unemployment, but only among
those not on work relief. As a result, the economy of the 1930sended up
with a peculiarcombinationof rising real wages for the employed, long
unemployment durations for the unemployed, and high aggregate unemployment rates.
' Sanford M. Jacoby, Employing Bureaucracy: Managers, Unions, and the Transformation of
Work in American Industry, 1900-1945 (New York, 1985).
20
RichardJensen, "The CausesandCuresof Unemploymentin the GreatDepression,"Journal
of Interdisciplinary History, 19 (Spring 1989), pp. 553-83.