The Jobless Recovery From the 2001 Recession: A Comparison to Earlier Recoveries and Possible Explanations.

CRS Report for Congress
The “Jobless Recovery” From the 2001
Recession: A Comparison to Earlier Recoveries
and Possible Explanations
Updated August 12, 2004
Marc Labonte
Analyst in Macroeconomics
Government and Finance Division
Linda Levine
Specialist in Labor Economics
Domestic Social Policy Division

Congressional Research Service ˜ The Library of Congress

The “Jobless Recovery” From the 2001 Recession:
A Comparison to Earlier Recoveries
and Possible Explanations
The National Bureau of Economic Research determined that the tenth recession
in the post-World War II era ended in November 2001 based upon its assessment that
most of the relevant economic variables had since shown sustained improvement.
Labor market data were the aberration. This report examines the period from
November 2001 to August 2003, when private nonfarm sector employment declined
by 1.3 million. Beginning in September 2003, employment began a sustained
increase. The labor market was also slow to rebound after the recession of the early
1990s, and the term “jobless recovery” was coined. Since this has occurred twice in
a row and is in contrast with the historic pattern following most other recessions,
there is concern that the two recoveries might indicate a trend rather than an anomaly.
The labor market indeed rebounded more slowly than usual coming out of the
2001 recession. For example, during just two of the previous nine recoveries, the
unemployment rate worsened for more than a year (14 and 16 months). At 19
months through June 2003, the rise in the unemployment rate persisted for longer
than any other post-war recovery. The weakness of employment is even more
unusual. Employment declined by 1.2% after the recession ended, declining for the
first 21 months of the recovery. The second largest decline in the post-war period
was a 0.6% decline that lasted 11 months into the recovery from the 1990-1991
recession. Typically, employment rebounds within three months after a recession has
The most compelling explanation for the jobless recovery is the weakness of
aggregate spending during the recovery. Growth in the first seven quarters of a
recovery typically exceeds 5%; in the jobless recovery it averaged 2.6%, below the
sustainable growth rate. The nation has undergone a number of shocks since 2001
that could have depressed spending, including oil shocks, the stock market crash, the
corporate scandals, September 11, and the Iraqi War. Some claim that strong
productivity growth is reducing employment; but productivity growth would not
depress demand as long as the recipients of the income generated by productivity
growth, a firm’s workers or investors, quickly spend it on consumption or capital
investment. Another explanation of the jobless recovery that has been offered is the
argument that sectoral shifts in employment have recently been unusually high, and
this has temporarily depressed overall employment; this argument is difficult to
It is also possible that the rise in the unemployment rate is caused by a rise in
the nation’s long-run “natural rate” of unemployment, determined by the nation’s
labor market characteristics and policies. The natural rate of unemployment is
unemployment that is unrelated to temporary changes in the business cycle. It is
unusual for the natural rate to significantly change in a couple of years. If the natural
rate is playing a role today, it is more likely because unemployment was further
below the natural rate than previously suspected in the late 1990s, and is now
returning to the natural rate. This report will not be updated.

A Look at the Labor Market at the Trough of the Jobless Recovery...........2
Unemployment ................................................2
Discouraged Workers...........................................3
Employment ..................................................5
Underemployment and Hours Worked.............................6
Comparison of the Current Recovery to Past Recoveries...................6
Explanations for the Jobless Recovery................................10
Inadequate Aggregate Demand..................................10
What Role Does Productivity Play?...............................13
A Change in the Natural Rate of Unemployment....................14
Structural Reallocation.........................................15
List of Tables
Table 1. Selected Labor Market Statistics During the 2001 Recession and
Jobless Recovery..............................................4
Table 2. The Unemployment Rate Following Recessions..................7
Table 3. Long-Term Unemployment during Recoveries...................8
Table 4. Decline in Employment in the Non-Farm Private Sector During
the Post-War Recessions and Recoveries...........................9
Table 5. Economic Growth in Recession and First Seven Quarters After
the Recession Ended..........................................11

The “Jobless Recovery” From the 2001
Recession: A Comparison to Earlier
Recoveries and Possible Explanations
The Business Cycle Dating Committee of the National Bureau of Economic
Research (NBER) determined that the tenth recession in the post-World War II era
ended in November 2001, a scant eight months from its inception in March. The
more than 9 million people who remained unemployed and the more than 4 million
people who wanted full-time jobs but remained in part-time positions more than one-
and-a-half years since the recession’s end were unlikely to agree with its assessment.
Nonetheless, the committee concluded that — aside from data on labor market
conditions — the economic variables it takes into account to determine the turning
points in the business cycle (e.g., real personal income) had shown sufficient,
sustained improvement to declare November 2001 the beginning of a recovery
After the not-quite-as-brief and not-quite-as-mild recession of the early 1990s,
the labor market was similarly slow to rebound. The 1990-1991 recession became
widely known as having been followed by a jobless recovery, the same appellation2
being applied to the current situation. The fact that this has occurred twice in a row
and seemingly contrasts with the historic pattern following other recessions has
prompted concern that these two recoveries might be a trend rather than an anomaly.
It has also prompted concern that the jobless recovery could presage another period
of economic contraction in the near future.
This report begins by examining the labor market situation at the trough of the
jobless recovery compared to the 2001 peak and trough in the business cycle. It then
compares the behavior of selected labor market indicators during the recovery with
their behavior during earlier post-World War II recoveries. The report concludes by
exploring explanations that have been offered for the differences.

1 Jon E. Hilsenrath, “Despite Job Losses, the Recession Is Finally Declared Officially
Over,” The Wall Street Journal, July 18, 2003. Note: The article reports that the anomalous
behavior of employment was the major point of contention within the committee on
choosing a date.
2 Although the 1991 recovery is frequently referred to as the first jobless recovery, there was
one earlier recession that followed a similar pattern: after the 1969-1970 recession,
unemployment rose for 14 months.

A Look at the Labor Market at the Trough of the
Jobless Recovery
To gauge the extent of the jobless recovery, it is useful to compare its trough
(August 2003) with the trough (November 2001) and the preceding peak (March
2001) in the business cycle. By almost any indicator in Table 1, workers on average
were worse at the trough of the jobless recovery than at the end of the recession.
As shown in Table 1, there were about 1 million more persons unemployed at
the trough of the jobless recovery than at the bottom of the business cycle (9.0
million and 8.0 million, respectively). The unemployment rate was higher (6.1% in
August 2003, a peak of 6.3% in June 2003) than it was at the end of the recession
(5.6%) and was about 2 percentage points above its level at the pre-recession peak
(4.3%). At 21 months into the current recovery, the unemployment rate was above
the level it reached at a comparable point during five of the eight preceding
recoveries.3 (The recovery from the 1980 recession is excluded from the total
number of postwar recessions in this instance because another recession began a year
after its end.)
The jobless rate fluctuated in a narrow range (5.6%-6.0%) for more than a year
after November 2001. Not until some 900,000 people searching for work joined the
labor force between the first and second quarters of 2003 did the unemployment rate
rise above 6.0%. The growth of the labor force (i.e., the number of people employed
and the unemployed) had been so small up until then that it exerted limited upward4
pressure on the unemployment rate despite substantial job losses. In fact, the rise
in the unemployment rate would have been larger if it were not for the decline in the
proportion of the population in the labor force, a decline that actually began before
the recession. Other recessions, in contrast, saw little or no decline in the labor force5
participation rate.
Another indicator of conditions in the labor market is how quickly people are
able to find jobs. In August 2003, as seen in Table 1, the average time a worker
spent unemployed was 19.2 weeks. In contrast, jobseekers were able to become
employed 4.8 weeks faster, on average, at the trough of the business cycle. The
average duration of unemployment thus lengthened more during the jobless recovery

3 Although the unemployment rate was always falling 21 months into the current recovery,
its level varied greatly at comparable periods in earlier recoveries. The unemployment rate
was above 7% at comparable points in the recoveries of the 1970s and 1980s, and below 5%
in the recoveries of the 1940s, 1950s, and 1960s.
4 The unemployment rate is defined as the number of people in the civilian noninstitutional
population age 16 or older who do not have jobs and who are actively seeking employment,
divided by the number of people age 16 or older in the civilian labor force.
5 Mark Schweitzer, “Another Jobless Recovery?,” Economic Commentary, Federal Reserve
Bank of Cleveland, March 2003.

(4.8 weeks) than it did during the recession (1.5 weeks).6 Similarly, the share of
unemployed workers who had not held a job in at least 27 weeks (22.4% in August
2003) exceeded levels recorded in March 2001 (11.2%) and November 2001
Discouraged Workers
It often has been noted that the unemployment rate does not fully reflect the
state of the labor market because it measures only the extent of joblessness among
those in the labor force. In light of the marked decrease in the fraction of the
population that either has a job or has actively sought a job in the four weeks before
the monthly administration of the Current Population Survey (CPS), this is a very
salient point at the present time: between November 2001 and August 2003, the
proportion of the civilian noninstitutional population age 16 or older that was in the
labor force (i.e., the labor force participation rate) fell by 0.5 percentage point to
66.2%.7 As previously mentioned, it was not until people surged into the labor force
in second quarter 2003 that the unemployment rate rose above 6.0%.
People who are able and willing to work may not recently have sought a job for
a variety of reasons, in which case they would not be officially classified as
unemployed. Some may not have searched for reasons unrelated to economic
conditions (e.g., child care or transportation problems). For others — who perhaps
have themselves experienced a lengthy spell of unemployment or who have
repeatedly heard of major corporations announcing layoffs involving thousands of
workers8 — the reason could be discouragement over their employment prospects
(e.g., because they think no work is available or because they lack education or
In August 2003, about 500,000 people were not seeking jobs because they
believed looking was futile.9 This marked the highest level of worker
discouragement for the month of August since the CPS underwent substantial
revisions in 1994, the first year for which comparable data are available. There were
significantly more discouraged workers in August 2003 (503,000) than in August
2002 (378,000), and more in August 2002 than mid-way through the 2001 recession
(337,000). 10

6 For a discussion of the relationship between the average duration of unemployment and
the unemployment rate following recessions see Daniel Sullivan, “Unemployment Duration
and Labor Market Tightness,” Chicago Fed Letter, Federal Reserve Bank of Chicago, n.

103, March 1996.

7 Data available at U.S. Bureau of Labor Statistics’ website []
8 For information of mass layoff activity see CRS Report RL30799, Unemployment through
Layoffs: What Are the Reasons? by Linda Levine.
9 BLS data refer to individuals who looked for a job during the prior 12 months and were
available to take a job during the reference week of the CPS, i.e., those who show some
attachment to the labor force.
10 Unlike the other variables included in this report, monthly data on discouraged workers

Table 1. Selected Labor Market Statistics During the 2001
Recession and Jobless Recovery
VariableEmploymentBusiness Cycle Peak and Trough
August 2003March 2001November 2001
Number of unemployed8,966,0006,136,0008,020,000
Unemployment rate6.14.35.6
Average weeks of19.212.914.4
Persons unemployed at
least 27 weeks as a share of22.411.214.3
total unemployment
Employed persons as a62.264.363.0
share of the population
Persons employed part-
time (1-34 hours a week)4,498,0003,279,0004,337,000
who would prefer full-time
Employees on nonfarm
payrolls in the privatea108,209,000111,564,000109,535,000
Average weekly hours of
workers on private nonfarmb33.734.133.8
Average weekly overtime
hours of production4.14.13.7
workers in manufacturing
Source: U.S. Bureau of Labor Statistics (BLS). All but the last three variables in the table are derived
from the Current Population Survey, a survey of households. The last three variables come from BLS’
establishment survey, which queries employers rather than the population.
a This reflects the number of jobs at establishments in the private nonfarm sector rather than the
number of people employed throughout the economy (e.g., a person may have a job at more than one
firm and some individuals are self-employed).
b Hours relate to production and nonsupervisory workers, who represent the majority of all jobholders.

10 (...continued)
are not adjusted for seasonal factors. Therefore, comparisons across different months
cannot be made.

Switching to the employment side of the equation does not reveal a more
sanguine picture. (See Table 1.) Fewer members of the population had jobs (62.2%)
in August 2003 compared to the fraction either at the outset of the recession (64.3%)
or at its end (63.0%). (As mentioned above, the unemployment rate is affected by
changes in the size of the labor force, which does not affect the employment-to-
population ratio.)
Employment, as measured by the number of employees on nonfarm payrolls in
the private (non-government) sector, continued to contract for the first 21 months of
the recovery. At that point, there were 1.3 million fewer jobs than there were at the
recession’s end. The economy would need to add 3.3 million jobs to August 2003’s11
total to equal the number that existed at the outset of the recession.
Manufacturers had been laying off employees since July 2000, well before the
recession began; and the jobless recovery did not reverse the job cutbacks.
Manufacturers shed 1.4 million jobs during the jobless recovery, according to BLS
data. The continuing difficulties of one segment of the “information technology (IT)
industry” can be seen at firms that manufacture computer and electronic products,
where nearly 250,000 jobs have disappeared since the recession’s end. On the
software side of the IT industry, another 100,000 jobs were lost at firms in the
service-producing sector that provide computer system design and related services.
The continuing impact of the burst dot-com or internet bubble can be seen in the
sector’s information services component, which employs many workers with IT
skills: during the jobless recovery, telecommunications providers have let go about
150,000 workers; internet service providers, search portals, and data processing firms
have terminated 35,000 employees; and internet publishers and broadcasters have
separated almost 5,000 workers from their payrolls — for a total of 175,000 fewer
jobs. These losses and those in other segments of the service-producing sector (e.g.,
retail trade and transportation) have offset job growth elsewhere in the sector (e.g.,
health services and temporary help services). As a result, the private service-
producing sector as a whole eked out a 164,000 job gain during the jobless12

11 As measured by the Current Population Survey (CPS), which surveys individuals rather
than firms, employment followed a somewhat different pattern. According to the CPS, the
jobless recovery ended in July 2002 (although employment fell again in late 2002). Most
economists consider the establishment survey a better source of data than the CPS because
it is a much larger survey and its respondents are thought to be more reliable. The CPS also
cannot be used to make comparisons over long periods of time because of periodic
population re-estimates. However, the CPS does include self-employed individuals, which
the establishment survey does not. See Daniel Aaronson, Ellen Rissman, Daniel Sullivan,
“Assessing the Jobless Recovery,” Economic Perspectives, Chicago Federal Reserve, vol.28,
second quarter 2004.
12 For a comparison of employment trends by industry during pre-2001 recessions and
recoveries see William E. Cullison, “The Case of the Reluctant Recovery,” Economic
Review, Federal Reserve Bank of Richmond, July-Aug. 1992.

Underemployment and Hours Worked
As with the unemployment rate, it is thought that changes in employment do not
fully reflect the conditions faced by workers. Some individuals are underemployed,
that is, they accept part-time jobs (i.e., 1-34 hours a week) despite their preference
for full-time positions. In August 2003, as shown in Table 1, some 4.5 million part-
timers wanted full-time jobs, which is about 1.2 million more underemployed13
persons than in March 2001 and 0.2 million more than in November 2001.
Employers have not only continued to cut employees from their payrolls since
the recession’s end, but they also have not generally lengthened employees’14
workweeks. After expanding by 12 minutes from 33.8 hours in November 2001 to
a peak of 34.0 hours in June 2002, the average workweek held at about 33.8 hours
through March 2003. It then contracted, and in August 2003 was 33.7 hours on
average, or 6 minutes shorter than at the outset of the recovery.
The overtime hours of factory workers (which are included in the workweek
total) have followed a similar path, that is, elevated during 2002 but then falling
during the last several months of 2003.15 Nonetheless, factory workers were still
putting in slightly more overtime than they had at the start of the recovery (a total of

4.1 hours per week as compared to 3.7 hours, respectively).

Comparison of the Current Recovery to Past
As can be seen in Table 2, there has always been some lag time before the
unemployment rate showed sustained improvement coming out of a recession, but
never as long as in the current recovery. In all but two of the previous nine
recoveries, the unemployment rate began to decline within six months of the
recession’s end. The exceptions are the recoveries that followed the 1969-1970 and
1990-1991 recessions, during which the unemployment rate continued to climb for
14 months and 16 months, respectively. At 20 months through July 2003, then, the
current lag in the unemployment rate showing steady progress is unusually long. The
three recessions that were followed by a jobless recovery, based on this criterion, are
three of the four mildest recessions in post-war history. The fourth was the short
recession of 1980. (The relationship between the strength of recessions and

13 For information on the long-run increase in involuntary part-time employment, see CRS
Report 98-695, Part-Time Job Growth and the Labor Effects of Policy Responses: An
Overview, by Linda Levine.
14 For information on the long-run decrease in the length of the workweek, see Katie
Kirkland, “On the Decline in Average Weekly Hours Worked,” Monthly Labor Review, July


15 Overtime has been identified as a leading indicator for changes in employment. If this
relationship holds, it would suggest some decline in manufacturing employment in the
months ahead. See Stephen Stanley, “Manufacturing Employment and Overtime,” Cross
Sections, Federal Reserve Bank of Richmond, vol. 9, no. 2, summer 1992.

recoveries is discussed in Explanations for the Jobless Recovery, which appears later
in the report.)
Table 2. The Unemployment Rate Following Recessions
Trough of Business CycleUnemployment Rate
UnemploymentNumber of
DateUnemploymentRateRate BeganSustainedUnemploymentRateMonthsfrom
Im provem e nt Trough
Oct. 1949a7.9Jan. 19506.53
May 19545.9Oct. 19545.75
Apr. 19587.4Aug. 19587.44
Feb. 19616.9Aug. 19616.66
Nov. 19705.9Jan. 19725.814
Mar. 19758.6June 19758.83
July 19807.8Aug. 19807.71
Nov. 198210.8Jan. 1983 10.42
Mar. 19916.8July 19927.716
Nov. 20015.6July 2003 6.2 20
Source: U.S. Bureau of Labor Statistics data from the CPS; business cycle dates determined by
a Two strikes occurred in October 1949, which inflated the unemployment rate in that month. (Before
1967, strikers were counted as unemployed.) In September 1949, before the strikes began, the
unemployment rate had been 6.6%.
Policymakers are interested not only in changes in the unemployment rate, but
also in the proportion of persons who experience long periods without work. The
federal-state unemployment compensation system provides income for varying
lengths of time to those displaced workers who meet differing state eligibility criteria.
In response to some past recessions and the 2001 recession, Congress provided
additional funds to workers who had been jobless for so long that they exhausted
their regular unemployment benefits.16 It appears that during the latest recovery the
fraction of workers unemployed for at least 27 weeks may have peaked in November

16 CRS Report RL31277, Temporary Programs to Extend Unemployment Compensation, by
Jennifer E. Lake. Note: Congress also has passed legislation to mitigate the labor market
impact of some recessions by directly creating jobs. For more information, see CRS Report
RL31138, Countercyclical Job Creation Programs of the Post-World War II Era, by Linda

2003 at 23.5%, 24 months after the trough of the recession — a later peak than in any
other post-war recession and the second largest share in the post-war period, even
though many of the other recessions were deeper. (See Table 3.)
Table 3. Long-Term Unemployment during Recoveries
Trough of Business CycleRecovery Period
Share ofDate Share ofPeak Share ofNumber of
DateUnemployedWithout Jobs forLong-termUnemploymentLong-termMonthsfrom
at Least 27 WeeksPeakedUnemploymentTrough
Oct. 19497.8April 195013.46
May 19548.5Feb. 195514.79
Apr. 195810.5Sept. 195820.75
Feb. 196113.7July 196119.95
Nov. 19706.7April 197213.617
Mar. 19759.7Feb. 197621.011
July 198010.7Jan. 198115.76
Nov. 198219.5June 1983 26.07
Mar. 199111.1Oct. 199223.119
Nov. 200114.3Nov. 200323.524
Source: U.S. Bureau of Labor Statistics data from the CPS; business cycle dates determined by
Rather than utilizing unemployment data to determine whether a recovery is
jobless, one could analyze how employment has fared in the post-war recessions and
recoveries. Employment (measured in the establishment survey as the number of
jobs) and unemployment (measured in the CPS, which counts individuals) need not
move together since the latter is affected by changes in the size of the labor force.
Looking at employment data, the difference between the last two recoveries and the
other post-war recoveries is even more pronounced. Historically, employment has
been more of a coincident indicator of economic activity, and less of a lagging
indicator, than the unemployment rate. Employment typically began to increase in
the month the recovery began, and at most three months after the recession had
ended. The last two recoveries have been quite a different story. As shown in Table
4, employment did not reach a trough until 11 months after the 1990-1991 recession
had ended. Employment reached its trough 21 months after the end of the 2001
recessi on.17

17 For a similar historical comparison see Stacy Schreft and Aarti Singh, “A Closer Look at

Table 4. Decline in Employment in the Non-Farm Private Sector
During the Post-War Recessions and Recoveries
Recession DatesPercentPercentNumber ofDate
Decline inDecline inMonths ThatEmployment
Employment Employment Employment Surpassed
DuringAfterDeclined AfterPrevious
Recession Recession Recession Peak
Nov. 1948-Oct. 19496.20.00Aug. 1950
July 1953-May 19543.80.53July 1955
Aug. 1957-Apr. 19584.90.42July 1959
Apr. 1960-Feb. 19612.10.00Feb. 1962
Dec. 1969-Nov. 19701.80.00Dec. 1971
Nov. 1973-Mar. 19752.70.41June 1976
Jan. 1980-July 19801.40.00Feb. 1981
July 1981-Nov. 19823.4a1Oct. 1983
July 1990-Mar. 19911.30.611May 1993
Mar. 2001-Nov. 20011.91.221 —
Source: U.S. Bureau of Labor Statistics data from the establishment survey; recessions dated by NBER.
a. less than 0.1%.
Another measure of employment recovery is the date in which post-recession
employment surpasses its previous peak. By this measure, the last two recoveries
stand out as unusually sluggish. Not until May 1993 did employment surpass its
previous peak in March 1990, 26 months after the 1990-1991 recession had ended.
(See Table 4.) As of July 2004, 32 months after the 2001 recession ended,
employment is still more than 1 million below its February 2001 peak. After every
other post-war recession, employment surpassed its previous peak 10-15 months after
the recession had ended.
Another important difference in the past two recoveries is that a significant
portion of the decline in employment occurred after the recession had ended. As
measured by the change in employment during the recession itself, the 2001 recession
ranks as the third mildest recessions in the post-war period, with less than a 2%
decline. Yet, the post-recession decline in employment through August 2003 is by
far the largest of the post-war period (1.2%).

17 (...continued)
Jobless Recoveries,” Economic Review, Federal Reserve Bank of Kansas City, second
quarter, 2003.

Explanations for the Jobless Recovery
Inadequate Aggregate Demand
In the short run, the unemployment rate is very sensitive to changes in economic
growth. Although the recession ended in November 2001, the recovery has been
unusually weak. Growth was below its sustainable rate for four of the first six
quarters of the recovery.18 At the least, one would not expect unemployment to fall
when growth is weak, and it is not surprising to see it rise.
The current recovery has not followed the pattern of most post-war recoveries,
which featured post-recession booms. Part of the reason the recovery has been
unusually mild is that the recession was so mild. In the first 3 quarters of 2001,
output contracted by a cumulative 0.6% of gross domestic product (GDP), making
it the mildest recession of the post-war period. After a deep recession, output can be
rapidly increased because there are many existing unutilized labor and capital
resources that can be brought back into production when aggregate spending
recovers. Because the contraction in output was negligible in 2001, following this
recession, there were far fewer idle resources to be put back into use.19 A look at the
historical record confirms this hypothesis (see Table 5). Unemployment peaked
seven quarters after the recovery began. At a similar stage in the recovery, there were
only two other post-war recoveries besides the current one in which average growth
was 3.1% or below: the recoveries beginning in 1970 and 1991. Altogether, the mild
recoveries followed three of the five mild post-war recessions, in which output20
declined by less than 2%. In every other recovery, the average growth rate
exceeded 5% at a comparable point.

18 The sustainable rate of growth is the rate at which additions to the labor force, capital
stock, and technical efficiency can increase output in the long run when the economy is at
full employment. Most economists today put the sustainable growth rate for the United
States at 3%-3.5%.
19 Capacity utilization data tell a different story. The fall in capacity utilization in this
recession was much larger than the fall in output, and this suggests there are potentially
more idle resources to be brought back into use in the recovery. Annually, capacity
utilization fell from 82.7% in 2000 to 74.6% in the third quarter of 2003. This is larger than
the decline during the 1990-1991 recession and similar to the decline during the 1981-1982
recession. The capacity utilization data cover only the industrial sector, however, which was
harder hit than the service sector in the 2001 recession. If similar data were available for
the service sector, it would presumably show a much smaller decline.
20 The unusual behavior of growth in the 1949 recession and subsequent recovery is
attributable to the demobilization of the wartime economy. The 1980 recession is omitted
from this discussion because the subsequent recovery lasted less than 7 quarters.

Table 5. Economic Growth in Recession and First Seven
Quarters After the Recession Ended
Recession Recovery
Percent DeclineGDP Growth
Periodin GDPPeriodRate
( c um ul at i v e) ( a nnual i zed)
1949:1-1949:4-1.6 1950:1 — 1951:3 10.8
1953:3-1954:1-2.7 1954:2 — 1955:4 5.7
1957:4-1958:1-3.71958:2 — 1959:46.0
1960:2-1960:4-1.61961:1 — 1962:35.9
1969:4-1970:1-0.61970:2 — 1971:42.5
1973:3-1975:1-3.01975:2 — 1976:45.0
1981:4-1982:3-2.91982:4 — 1984:26.8
1990:3-1991:1-1.51991:2 — 1992:43.1
2001:1-2001:3-0.62001:4 — 2003:22.6
Source: U.S. Bureau of Economic Analysis.
Note: For the purposes of this table, the recession is dated as beginning in the first quarter of negative
growth and the recovery is dated as the first quarter of positive economic growth. Data for the 1980-
1981 recovery are omitted because the 1981-1982 recession began less than 7 quarters after the 1980
recession ended.
The correlation between mild recessions and mild recoveries also holds when
measured by employment. There were only four post-war recessions in which
employment decreased by less than 2%: 1969-1970, 1980, 1990-1991, and 2001 (see
Table 4). Unsurprisingly, all three of the recoveries in which the unemployment rate
continued to rise for six months or more are included in this group (see Table 2),
with only the 1980 recovery leading to a quick decline in unemployment (followed
shortly thereafter by another recession).21
What initially held the current recovery back? The economy faced a number of
unusual shocks after the recession ended. Before the recession began, oil prices more
than doubled on an annual basis. They dipped a little in 2002, but increased again

21 These results are consistent with an economic rule of thumb known as Okun’s Law, which
roughly relates changes in economic growth to changes in the unemployment rate. In one
version of Okun’s Law, the unemployment rate will rise if economic growth is not rapid
enough to accommodate growth in productivity and the labor force. If the potential growth
rate of productivity is 2.5% and the labor force grows at 1% a year, then unemployment will
rise when economic growth is below 3.5%. As can be seen in Table 5, Okun’s law correctly
predicts all three of the post-war jobless recoveries (1970, 1991, 2001). See CRS Report
RS21139, Unemployment and Economic Growth, by Brian Cashell.

in the months heading up to the war in Iraq. Oil spikes temporarily reduce aggregate
production by increasing the price of an important input in the production process
and reduce aggregate spending if the shift in income to foreign oil producers is not
quickly spent on American goods and services.22 The recession also coincided with
a large decline in equity prices (e.g., the Standard and Poor’s 500 Index fell 45%
from peak to trough), which began to recover only in early 2003. This reduced the
willingness or ability of firms to borrow to finance new investment spending and
reduced personal consumption through a negative wealth effect. Toward the end of
the recession, September 11 occurred. This may have weakened consumer and
investor demand by creating uncertainty and reducing confidence. It also caused
large sectoral shifts on the production side of the economy, with long lasting effects
on certain industries such as the airlines, tourism, and insurance, and certain regions,
such as New York City.23 The recovery was also buffeted by the corporate
governance and accounting scandals set off by the Enron bankruptcy, which
undoubtedly reduced investment spending and equity prices further. Finally, the
lead-up to military action in Iraq was thought to harm the economy both through the
oil price channel outlined above and by further reducing consumer and investor
It is difficult to quantify many of these effects since they were each unique and
of a qualitative nature. Some of them may have been relatively insignificant in
relation to overall economic activity. Nevertheless, added up, it is impressive how
many negative events harmed the economy in 2001-2002, compared to how few
positive events there were to offset them. To the extent that these events heightened
uncertainty, they may have made employers less willing to quickly expand their
workforces than they otherwise would have been.24 Still, it is somewhat surprising
that the economy has not reacted more quickly to the large stimulus provided by
monetary and fiscal policy over the past two years. During that time, the federal
funds rate was reduced from 6.5% to 1% and the budget moved from a surplus of

1.3% of GDP in 2001 to a deficit of 3.5% of GDP in 2003.

During recessions, firms engage in what economists call “labor hoarding.”
Rather than laying off enough workers so that the remaining workforce is fully
employed in a downturn, firms prefer to keep more workers than output requires
because it is less costly than hiring and training new workers when the economy
recovers.25 It is possible that more labor hoarding occurs in shallow recessions, and
firms are forced to make larger layoffs in deep and prolonged recessions. If so, that

22 See CRS Report RL31608, The Effects of Oil Shocks on the Economy: A Review of the
Empirical Evidence, by Marc Labonte.
23 See CRS Report RL31617, The Economic Effects of 9/11: A Retrospective Assessment,
coordinated by Gail Makinen, and CRS Report RL31250, The Worker Adjustment and
Retraining Notification Act (WARN), by Linda Levine for data on mass layoffs resulting
from the terrorist attacks.
24 Bharat Trehan, “Why Has Employment Grown So Slowly?,” Weekly Letter, Federal
Reserve Bank of San Francisco, no. 93-14, Apr. 9, 1993.
25 The seminal article on labor hoarding is Walter Oi, “Labor as a Quasi-Fixed Factor,”
Journal of Political Economy, vol. 70, no. 6, Dec. 1962, p. 538.

implies that firms will initially need to make fewer hires following a mild recession
since the initial increase in output can be met by the previously underutilized
“hoarded” workers. This offers one reason why shallow recessions followed by
shallow recoveries would initially lead to less hiring. It is difficult to test this theory
empirically, however, since there is no meaningful way to measure labor hoarding.26
What Role Does Productivity Play?
It has been argued that strong productivity growth can explain why firms did not
hire in the jobless recovery.27 According to this argument, strong productivity growth
allows firms to meet increases in final demand from their existing workforce, so they
have no need to hire new workers. Strong productivity growth has been described
as a mixed blessing because, so the argument goes, firms would have to make more
hires if productivity growth were weaker. The productivity-induced weakness in
labor markets is then blamed for the sluggish recovery on the grounds that it has
caused workers to be too uncertain about the employment outlook to spend
While it is true that productivity growth has been strong enough to increase
output without any increase in employment, this argument confuses cause and effect.
Increased productivity does not automatically cause production (supply) to outpace
spending (demand). Higher productivity creates higher income for the firm that
flows either to its workers or its investors (a firm’s owners, creditors, and
shareholders). As long as the recipients then spend their higher income on
consumption or capital investment, increases in supply will be quickly matched by
increases in demand.29
The real issue then is the forces holding back aggregate demand, the subject of
the previous section, despite the increases in productivity that are boosting output.
And the strong productivity gains make demand seem stronger than it is in a casual
comparison to the past. Economic growth is caused jointly by increases in the labor
force, which grows at a fairly steady 1% a year, and increases in productivity (due to

26 One measure that seems to confirm this theory is the ratio of job losers to total
unemployed. If firms are hoarding labor, then there should be a smaller increase in the ratio
of job losers to total unemployed. There was a much larger increase in this ratio in the deep
recessions of 1973-1975 and 1981-1982 than in the mild recessions of 1969-1970, 1990-

1991, and 2001.

27 This argument is made, for example, in “A Jobless Recovery?,” Time Magazine, July 15,

2002, p.Y9.

28 It is interesting to note that in the late 1990s, many economists were arguing that faster
productivity growth was temporarily reducing the unemployment rate. They argued that
workers, who had not anticipated the increase in productivity growth, were raising their
wage demands too slowly. This made labor relatively inexpensive and led firms to hire
more workers. For example, see Laurence Ball and Gregory Mankiw, “The NAIRU in
Theory and Practice,” Journal of Economic Perspectives, vol. 16, no. 4, fall 2002.
29 This assumes that the Federal Reserve Board will allow monetary policy to accommodate
the increase in supply. With the federal funds rate at its lowest point in decades, it is
difficult to argue that such an accommodation did not occurred.

capital investment or efficiency gains). From the mid-1970s to the mid-1990s, when
productivity was growing at about 1.5% a year, the economy could grow at 2.5%
without an increase in the unemployment rate. Now that the productivity growth rate
seems to have increased, a 2.5% economic growth rate is no longer sufficient to keep
the unemployment rate stable, and no longer indicates that demand is growing
quickly enough to keep production at full employment.
A Change in the Natural Rate of Unemployment
Another possible explanation for why the unemployment rate has risen for so
long during this recovery is that the economy’s natural rate of unemployment is
rising, independent of the effects of changes in aggregate spending.
In the long run, the economy always adjusts so that aggregate spending
(demand) matches aggregate production (supply). The unemployment rate that
would prevail in this situation is called the natural rate of unemployment. When
unemployment is at the natural rate, none of the unemployment is caused by weak
economic growth. Rather, workers are unemployed because they are either in the
process of moving from one job to another or their skills are incompatible with the
jobs available in the local area. In other words, the natural rate of unemployment is
caused by the supply side of the labor market. The natural rate can change over time
as the characteristics of the labor market change. For example, since older workers
have a lower unemployment rate than younger workers, as the population ages, the
natural rate of unemployment would automatically decline. The natural rate can also
change if labor market policies alter the characteristics of the labor market. For
example, most economists believe that the natural rate in the United States is about
half the rate in many Western European countries because the U.S. labor market is
more flexible. In the United States, it is easier for a firm to alter the size of its
workforce, there is a less generous social safety net, and there is more regional labor
Because the natural rate is a long-run concept, it is difficult to believe the natural
rate could have changed significantly over the past two-and-a-half years. There has
not been any major change in labor market policy during that time, and demographic
changes are incremental. If the natural rate has changed, it would be part of a longer
trend that will not be identifiable in the near term.
There is another reason why the unemployment rate might have continued to
rise for so long related to the natural rate concept. Even if the natural rate had not
changed over the past two-and-a-half years, it is possible that when unemployment
reached 3.9% in December 2000, it was further below the natural rate than suspected.
Just as the unemployment rate can temporarily rise above its natural rate when
growth is too slow, unemployment can temporarily fall below the natural rate when
growth is unsustainably fast. In these circumstances, one would expect to see a rising
inflation rate as wages are pushed above productivity because too many jobs are
chasing too few workers. Few economists believed the natural rate had reached as

30 See CRS Report RL30765, Causes of Unemployment: A Cross Country Comparison, by
Marc Labonte.

low as 3.9% in 2000, but many assumed that 3.9% was not too far from the natural
rate since there was no significant upward pressure on inflation at that time. In
hindsight, if the natural rate has been higher than suspected in recent years, say 6.0%
versus 5.0%, then the prolonged increase in the unemployment rate could partly be
attributable to the long-term adjustment back toward the natural rate from an
unsustainably low level. In this case, one would expect the unemployment rate to fall
once the recovery becomes more robust, but it will fall less than expected. Those
who argue that the natural rate was underestimated in the late 1990s point to the fact
that the natural rate averaged 6%-6.5% in the 1970s and 1980s. To put that figure
in perspective, consider that the recent peak unemployment rate would have been
considered to be full employment, only attainable near the peak of the business cycle,

20 years ago.

Since the natural rate is a long-run concept, it is too soon to determine which
portion, if any, of the recent increase in the unemployment rate is supply-side driven,
and which portion is demand-driven. But the inflation rate is one piece of evidence
to determine whether inadequate demand or a change in the natural rate is currently
driving the rise in unemployment. If the economy is suffering from insufficient
demand, the inflation rate should be falling; if the unemployment rate is being driven
by changes in the labor market, inflation should be unaffected. The core inflation
rate, which strips out volatile food and energy prices, fell from 2.6% in 2001 to 2.4%
in 2002 to 1.4% in 2003. This indicates that insufficient demand is likely at least part
of the story behind the rise in the unemployment rate.
Structural Reallocation
Declines in aggregate employment are often blamed on restructuring, or the
“structural” (i.e., permanent) reallocation of resources from some industries to
others.31 For example, the decline in employment since 2001 is often attributed to
the collapse of the “dot-com” industry. The argument is made that resources were
overinvested in the dot-com industry in the late 1990s. When this situation was
reversed, workers in that industry were no longer needed, causing overall
employment to decline. Similarly, the large decline in manufacturing employment
in recent years may have ongoing structural rather than cyclical causes.
This argument conflates gross job loss, which is the total number of jobs
eliminated in a period, and net job loss, which is gross job loss less gross job gains.
The reallocation of resources in the economy is the primary reason that gross job loss
occurs. Changes in tastes, technology, and comparative advantage continually cause
labor and capital to be shifted from one industry to another in a market economy.
Sizeable reallocations of labor across industries has been a constant in the United
States, as shown in Figure 1. But in most years, the economy has not had any
problem offsetting gross job loss with a greater number of gross job gains. Thus,
economic restructuring typically is not accompanied by net job loss overall, even
though it may result in net job loss at the local or industry level.

31 An early example in this line of research is David Lillen, “Sectoral Shifts and Cyclical
Unemployment,” Journal of Political Economy, vol. 90, no. 4, Aug. 1982.

Figure 1: Gross and Net Job Loss and Gains, 1992:3-2003:3
1992 1994 1996 1998 2000 2002
Source: BLS
It is possible that if restructuring were unusually large at any given time, perhaps
like the dot-com collapse, the economy could be unable to absorb that many workers
in new jobs fast enough to prevent net job loss. Unfortunately, there is no way to
systematically identify which restructuring events are large enough and separate their
effects on net job loss from other economic phenomena occurring simultaneously.
What recently appeared to be significant structural reallocation could simply be the
result of the cyclical decline in aggregate spending since some industries are
systematically more affected by recessions than others. In the absence of the decline
in aggregate spending, it is possible that the dot-com collapse would not have had
any effect on aggregate employment. Looking at Figure 1, the data seem to refute the
argument that heightened restructuring caused the jobless recovery — gross job flows
were lower from 2001 on than they had been in the late 1990s.
A recent New York Fed article attempts to determine how much of the jobless
recovery is caused by shifts in structural rather than cyclical employment.32 The
authors conclude that structural shifts were a major cause of the jobless recovery, and
were larger than the structural shifts that occurred in the past three decades. They
look at two measures to reach this conclusion. First, they find that fewer layoffs were

32 Erica Groshen and Simon Potter, “Has Structural Change Contributed to a Jobless
Recovery?”, Current Issues in Economics and Finance, Federal Reserve Bank of New York,
vol. 9, no. 8, Aug. 2003. Daniel Aaronson, Ellen Rissman, and Daniel Sullivan (“Can
Sectoral Reallocation Explain the Jobless Recovery?” Economic Perspectives, Federal
Reserve Bank of Chicago, second quarter, 2004) show that the results are sensitive to the
dates used.

temporary in the 2001 recession compared to past recessions. They argue that if the
layoffs were caused by cyclical factors, the layoffs would not be permanent. (The
distinction between temporary and permanent layoff is self-reported by workers at
the time of the layoff. It is not determined by looking after the fact to see if a worker
was actually re-hired.) Second, looking at employment at the industry level, they find
that 79% of industries experienced structural employment change in the current
business cycle, compared to closer to 50% in other business cycles. However, their
findings are problematic due to their definition of structural change. If the change
in employment is in the same direction (whether it rises or falls) in both recession
and recovery, it is classified as structural change. If not, it is classified as cyclical
change. By this definition, it is a foregone conclusion that the authors would find
structural change dominating the current period since aggregate employment moved
in the same direction in both periods. These results confirm that the aggregate
pattern is replicated at the industry level, but do not explain why that pattern
occurred. If all industries had been affected equally in the jobless recovery, by their
definition, 100% of the jobless recovery would be caused by structural change. This
is the opposite of what most economists have in mind by structural change, which is
persistent changes in employment that affect some industries but not others. The
New York Fed article does not shed light on structural change by that definition.
A recent Chicago Fed article studies the role of structural change in the jobless
recovery using the more common concept.33 It splits employment change at the
industry level into cyclical and structural components. It finds that structural
employment reallocation typically rises during recessions, as it did in the 2001
recession. But the increase in structural change during the 2001 recession is smaller
than during any other recession of the past four decades. After the recession ended,
structural change declined. These findings cast doubt on the theory that structural
change is responsible for the jobless recovery

33 Ellen Rissman, “Can Sectoral Labor Reallocation Explain the Jobless Recovery?”,
Chicago Fed Letter, Federal Reserve Bank of Chicago, no. 197, Dec. 2003.