C H A P T E R 3

ACHIEVEMENTS AND
CHALLENGES IN THE
U.S. LABOR MARKET

A fundamental metric for judging an economy’s performance is its suc‑

cess in providing abundant job opportunities that pay good wages
and provide an opportunity to get ahead. Five-and-one-half years ago—in
the wake of the worst financial crisis since the Great Depression—the
U.S. economy faced a massive challenge, as GDP shrank and the number
of jobless workers rose to more than 15 million. Since then, a successful
multifaceted policy response, including actions by the President, Congress,
and the Federal Reserve, combined with the determination of the American
people, has enabled the U.S. economy to dig out of that deep hole, putting
more people back to work, reducing the unemployment rate, and creating
a virtuous cycle in which higher consumer purchasing power supports
greater economic activity and job creation. After four years of recovery in
employment, in 2014, the unemployment rate declined at its most rapid rate
in nearly three decades. By the end of the year, it had fallen to 5.6 percent,
close to its pre-recession average of 5.3 percent.1 But the United States labor
market still has more work to do to achieve the full health that comes with
not just low levels of unemployment, but also a labor market that encourages
labor force participation, supports quality jobs, and facilitates productive
matching of workers and positions—all of which are essential to creating
well-paying jobs and supporting robust family incomes.
This chapter begins by discussing the substantial progress that has
been made in healing the labor market since 2009, and the acceleration in
progress seen throughout 2014. By October 2014, the unemployment rate
had fallen more rapidly over the preceding 12 months than in any 12-month

1 Bureau of Labor Statistics, Current Population Survey; CEA calculations. Throughout this
chapter, unless otherwise specified, data and statistics are from the Bureau of Labor Statistics
Current Population Survey or CEA calculations from these data.

103
period since 1984. The sharp drop in unemployment came amid a stabiliza‑
tion in the labor force participation rate and, for the year as a whole busi‑
nesses added 3.0 million jobs—the most in any year since 1997. Moreover,
nominal wage growth for production and nonsupervisory workers—a group
that represents about 80 percent of workers who have lower earnings on
average—continued to rise slightly faster than inflation, a reversal from
what had been seen earlier in the recovery. Real wage growth was aided
by low levels of inflation, including declining prices in the fourth quarter
of 2014. Moreover, workers’ take-home pay was helped by the fact that a
typical worker’s contribution to employer-sponsored family health insur‑
ance coverage rose at roughly one-half the rate seen on average prior to the
recession, continuing a recent trend of subdued health cost growth. Finally,
2014 continued to see the economy shift away from part-time work toward
full-time work, as all of the employment growth was in full-time jobs. Over
the course of the recovery, the share of the labor market in full-time jobs
has increased and by the end of 2014, the number of Americans holding
full-time jobs had increased more from January 2010 than it had added total
jobs over the same period.
Despite these positive developments, more work remains to both
complete the cyclical recovery and address underlying structural issues that
predate the recession, some of which have been present for decades. As
described in Chapter 1, three key factors shape the economic situation of
the middle class: productivity growth, the distribution of income, and labor
force participation. As Chapter 1 also notes, due to a combination of long-
term economic challenges and the Great Recession, the middle class has seen
little improvement in real incomes since 1997 despite productivity growth,
signaling at least one area where much work remains to be done in the labor
market and overall economy.
After reviewing the notable progress in the labor market over 2014, this
chapter steps back to consider a set of five long-run issues the labor market
must address. These are: i) a long-standing decline in the participation rate
that has been compounded by the recession and the retirement boom; ii) a
rapidly recovering long-term unemployment rate that nonetheless remains
elevated; iii) a similar pattern of rapid decline but continued elevation in the
rate of people working part time but who are seeking full-time employment;
iv) cyclical improvements in labor market fluidity that are set against a back‑
drop of a long-term decline in a variety of metrics of labor market fluidity,
or labor market “churn”; and v) real wage growth that is beginning to pick
up but is still insufficient. These phenomena have, to varying degrees, been
building up in the years or decades before the Great Recession and, in many
cases, are following patterns similar to those in other recent recessions,

104 | Chapter 3
particularly those from 1980 on. This suggests that these issues are linked –
for example, when a shock hits the economy, less labor market fluidity can
result in more long-term unemployment and part-time employment and
a lower participation rate than would occur if the labor market were more
dynamic. In many cases, the increasingly rapid recovery in the labor market
will help to address these challenges. In some cases, these trends may reflect
a natural progression that would be undesirable to reverse, such as rising
retirements among aging workers. However, additional policy steps are
needed to counteract the continued effects of the Great Recession as well as
longer-term trends that predated it. Consequently, this chapter concludes by
laying out key elements of the President’s middle-class economics agenda,
which includes policies aimed at growing and supporting middle-class
families, strengthening the labor market and expanding economic oppor‑
tunity. As the past several years suggest, economic policies that focus on
strengthening the middle class create a stronger foundation for shared and
sustainable growth in the years to come.

The State of the U.S. Labor Market in 2014

Since the end of the Great Recession in 2009, the economy has
made enormous strides toward recovery, in terms of output, labor market
indicators, consumer confidence, and numerous other measures. Perhaps
no recent economic development has been more surprising than the rapid
fall in the unemployment rate and commensurate pickup in the rate of job
growth in 2014, which far outperformed forecast expectations. From its
2001-07 average of 5.3 percent, the unemployment rate hit 10.0 percent
in October 2009; but as of December 2014, the rate stands at 5.6 percent,
having recovered 93 percent of the way back to its pre-recession average.2
Notably, 2014 marked the strongest year of job growth since 1999 and the
strongest year of private-sector job growth since 1997. December’s 5.6-per‑
cent unemployment rate was achieved roughly five years ahead of consensus
forecasts made as recently as 2013, as shown in Figure 3-1.
In part due to a vigorous policy response to the economic crisis, the
United States is in a sustained economic recovery. The Administration’s
early actions, including the American Recovery and Reinvestment Act of
2009 and middle-class tax cuts, helped catalyze this recovery: the Council of
Economic Advisers (CEA) estimates that between early 2009 and the end of

2 Throughout this chapter, the phrase “pre-recession average” refers to the average between
December 2001 and December 2007, the most recent expansionary period before the Great
Recession.

Achievements and Challenges in the U.S. Labor Market | 105

 Figure 3-1
Actual and Consensus Forecast Unemployment Rate, 2008–2014
Percent of Labor Force
11

2010 Forecast
10

Actual
9 2011 Forecast

2012 Forecast
8
2013 Forecast
7
2014 Forecast
6

4
2008 2010 2012 2014 2016
Note: Annual forecasts are current as of March of the stated year. Dashed line represents December 2014
value (5.6 percent). Shading denotes recession.
Source: Bureau of Labor Statistics, Current Population Survey; Blue Chip Economic Indicators.

2012, the Recovery Act added a total of more than 6.0 million job years to
the economy (CEA 2014b).
In 2014, the rate of decline in the unemployment rate picked up to an
average of 0.1 percentage point per month, higher than the rate of decline
from 2010 to 2013, with much of the decline reflected in lower long-term
unemployment. Although the long-term unemployed account for only about
one-third of all unemployed, these reductions in long-term unemployment
accounted for about two-thirds of the total unemployment decline in 2014.
Falling long-term unemployment combined with a stable participation rate
in 2014 suggests that the long-term unemployed are going back to work at
higher rates (Cajner and Ratner 2014).
The improvement in the health of the labor market is also apparent
in a range of labor market indicators, as shown in Figure 3-2. The headline
unemployment rate accounts for jobless individuals who are actively seeking
employment. Broader measures of labor market underutilization include
individuals who are not looking for work because they believe no jobs are
available (discouraged workers); others available for work but who have not
looked for work in the past month (other marginally attached); and those
who are working part-time but would like full-time work (part-time for
economic reasons). All of these indicators tell a broadly consistent story:
the U.S. economic recovery has made considerable progress, but it is not

106 | Chapter 3
 Figure 3-2
Elevation and Recovery of Broader Measures of Unemployment
Remaining Elevation as of December 2014 Percent Increase to Great Recession Peak Percent Recovered

Overall Unemployment Rate (UR) 6 90 93

U-4 (Unemployed + Discouraged) 8 92 91

U-5 (U-4 + Other Marginally Attached) 11 84 87

U-6 (U-5 + Part Time for Economic Reasons) 23 87 74

Short-Term (26 Weeks or Fewer) UR -10 64 116

Long-Term (27 Weeks or More) UR 73 326 78

-50 0 50 100 150 200 250 300 350

Percent Change in Indicator Relative to 2001-07 Average
Note: All rates are expressed as a percent of the labor force and are seasonally adjusted.
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

yet complete. Important differences remain in the progress of the recovery

across measures, however, including the continued elevation of long-term
unemployment.
Relative to many other advanced economies, the United States
experienced a large increase in unemployment during the crisis and yet
has also had the strongest recovery since the peak of the crisis, as shown in
Figure 3-3. Between the first quarter of 2008 and the final quarter of 2009,
U.S. unemployment rose from 5.0 percent to 9.9 percent. Over the same
period, unemployment in the Organisation for Economic Cooperation and
Development (OECD) countries (excluding the United States) increased
from an average of 5.1 percent to 7.8 percent.3 Unemployment in the euro
area over this period rose from 6.1 percent to 9.3 percent.
The most significant differences have emerged since early 2010. U.S.
unemployment steadily declined and was down to 5.7 percent by the third
quarter of 2014, over 40 percent below its recession maximum. In contrast,
average unemployment in both the non-U.S. OECD and euro area has made
3 CEA weighted OECD and euro area countries by GDP (in millions of USD), so that countries
with larger economies received more weight than smaller countries. The United States is
excluded from the OECD weighted average. Accordingly, these figures differ from OECD’s
published unweighted average unemployment rate across OECD countries.

Achievements and Challenges in the U.S. Labor Market | 107

 Box 3-1: Unemployment Across Gender, Race, and
Ethnicity Groups: The Situation for Men of Color
Men of color have much higher rates of unemployment than do
White men. For example, in December 2014, adult African-American
men had an unemployment rate of 11.0 percent—6.6 percentage points
higher than that of adult White men. Among adult Hispanic men, the
unemployment rate was 5.3 percent in December 2014, 0.9 percentage
point higher than that of adult White men. Racial gaps in unemploy‑
ment have narrowed over time, but less progress has been made among
African-American men, for whom the gap in the unemployment rate
relative to Whites has fallen the least.

Figure 3-i
Labor Force Participation Rate Gap Between African-American
and White Youth by Gender: Ages 16 to 24, 1973–2014
Gap in Percentage Points
18
Young Men
16
Dec-2014
14

12

10

8
Young Women
6

0
1970 1975 1980 1985 1990 1995 2000 2005 2010
Note: Data are differences between the 12-month moving averages of the non-seasonally adjusted labor
force participation rates of African-American and White youth ages 16-24.
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

In addition to higher unemployment rates, there are also differ‑

ences in labor force participation, which mean that men of color often
have even lower rates of employment than the unemployment rates
alone would suggest. The gap in participation is especially problematic
among young men, since early-life labor market experiences have
significant impacts on later-life labor market success (Edelman, Holzer
and Offner 2006; Raaum and Roed 2006).1 The labor force participation
rates of young White and African-American women have begun to
converge since the 1990s, while convergence among young men largely

1 The literature finds persistent and significant impacts of post-graduation labor market
conditions and opportunities on later-in-life wages and employment.

108 | Chapter 3
 stalled until the late 2000s. In December 2014, young African-American
women’s participation was 5 percentage points lower than young White
women’s, while young African-American men’s participation was 9
percentage points lower than young White men’s.
To speed U.S. progress in closing the racial disparities in labor
market outcomes, the Administration has made tackling unemployment
among minority men a priority under the My Brother’s Keeper Initiative.
The initiative supports the education and employment of African-
American, Hispanic and Native American men, all of whom experience
elevated unemployment and lower participation relative to men in other
racial groups.

little progress. Unemployment across the OECD, excluding the United

States, is, on average, roughly unchanged from its peak. As of the fourth
quarter of 2013, the average unemployment rate across non-U.S. OECD
countries was 7.6 percent. Unemployment across euro zone countries fared
worse, with a decline in unemployment in 2010, followed by a sharp increase
between 2011 and mid-2013. These international averages naturally abstract
from varied experiences among OECD countries: Germany’s unemploy‑
ment rate fell between the first quarter of 2008 and the first quarter of 2010,
while Spain’s unemployment rate more than doubled. Nonetheless, the

Figure 3-3
Unemployment in Non-U.S. OECD, Euro Area,
and United States, 2000–2013
Percent
12
United States
10 Euro Area
2013:Q4
(Weighted)

8
Non-U.S. OECD
6 (Weighted)

0
2000 2002 2004 2006 2008 2010 2012
Note: OECD (excluding the United States) and euro area averages are weighted by member
countries' GDP.
Source: Organisation for Economic Co-Operation and Development, Harmonized
Unemployment Rate and GDP series; CEA calculations.

Achievements and Challenges in the U.S. Labor Market | 109

recovery in the U.S. unemployment rate compares favorably against the
general experience of other advanced economies.
Behind the improvement in U.S. unemployment is a historic record
of steady job growth, albeit one that follows historic job losses. As described
in Chapter 2, total employment increased by 3.1 million jobs in 2014—the
strongest year of the recovery—and average monthly job growth was
260,000, as shown in Figure 3-4. The private sector has added jobs for 58
consecutive months through December 2014, the longest period of con‑
tinual job growth on record.
In 2014, private-sector employment growth was particularly strong
in industries that traditionally provide good, middle-class jobs, such as con‑
struction and professional and business services. Since February 2010, more
than 850,000 manufacturing jobs have been added, an increase of 7 percent.
The average workweek for production and non-supervisory workers in
manufacturing has also increased to near its highest level since World War
II. Over the same period, 2.9 million jobs have been added in professional
and business services, an 18 percent increase.
The labor market recovery has been generally shared across the full
spectrum of American workers. Table 3-1 shows that looking across the

Figure 3-4
Average Monthly Job Growth by Year, 2007–2014
Average Monthly Job Gain/Loss, Thousands
400

300 260
188 199
200 173

95 89
100

-100

-200

-300
-298
-400
-424
-500
2007 2008 2009 2010 2011 2012 2013 2014
Source: Bureau of Labor Statistics, Current Employment Statistics. 1900

110 | Chapter 3
 Table 3-1
Tracking the Recovery Across Race, Gender, Age, and
Level of Educational Attainment
Remaining
Percent
Elevation
Pre- Increase
as of Percent
Recession to Great
December Recovered
Average Recessi-
2014
on Peak
(Percent)
Overall Unemployment Rate (UR) 5.3 90 6 93

Male UR 5.4 106 8 93

Female UR 5.2 74 3 96

White UR 4.6 99 4 96
African-American UR 9.8 72 6 91
Hispanic UR 6.5 99 0 100
Asian UR 4.5 72 9 87

Less than High School UR 7.9 100 9 91

High School Graduates UR 4.8 127 9 93
Some College UR 4.1 117 20 83
College Graduates UR 2.5 99 15 85

Age 16-24 UR 11.4 71 8 88

Age 25-54 UR 4.3 108 8 92
Note: Asian unemployment rate is a 12-month moving average of not seasonally adjusted data.
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

population by racial, gender, and educational differences, most groups are

at least 90 percent recovered, and those that have not reached that point are
close to it.
The 1.2 percentage-point fall in the annual unemployment rate in
2014 was the largest such drop since 1984, and some groups experienced
even larger declines in unemployment. Both the Hispanic and African-
American annual unemployment rates fell by 1.7 percentage point in 2014,
one of the largest declines in series history. As of December 2014, the
African-American unemployment rate had recovered 91 percent of the way
back to its pre-recession average, compared to 100 percent for Hispanics, 87
percent for Asians, and 96 percent for White workers.
The labor market gained strength in 2014, and numerous indicators
illustrate that the recovery is robust. Now that much of the direct challenges
of the recession are behind us, the United States must turn its attention to
ensuring that the benefits of the recovery are widespread, benefiting more

Achievements and Challenges in the U.S. Labor Market | 111

middle-class families. This requires addressing five longer-run challenges in
the labor market. The following sections discuss each of these challenges in
greater detail.

Labor Force Participation

The decline in the unemployment rate in the economic recovery has
been driven by the increased pace of job creation. In addition to the decline
in the traditional unemployment rate, a broader measure that also includes
discouraged workers and people who would like to work if a job were avail‑
able (U-5) has come down from a high of 11.4 percent in October 2009 to 6.9
percent in December 2014, or 87 percent of the way back to its pre-recession
average.
At the same time, the economy has continued to go through a sub‑
stantial change in labor force participation. Since peaking in the first quarter
of 2000 at 67.3 percent, the labor force participation rate declined to 62.8 in
the fourth quarter of 2014. A large portion of this decline is explained by the
lower participation rates of an aging labor force and, in spite of continued
demographic pressures in this direction, the participation rate has held
steady since October 2013. This suggests that a stronger labor market is
bringing people back into the labor force, partially off-setting the increasing
size of the retirement-age population. Nevertheless, the participation rate is
unlikely to return to its peak rate in the near future. This section examines
the role of the aging baby boomers in driving declining participation, as well
as the lesser but important roles of a decades-long downward trend in male
labor force participation and a more recent slight trend decline in female
labor force participation discussed in Chapter 1.

A Longer-Term Perspective on Labor Force Participation

The labor force participation rate, defined as the share of the
population ages 16 and older who are working, or who are actively seeking
employment, is an important measure of labor market potential and health.
Labor force nonparticipation is not always a source of concern—many non-
participators are seniors enjoying their retirements, young people investing
in education, or parents caring for their children. However, low labor force
participation—particularly among people of prime working age (ages 25
through 54) — is evidence that we can do more to create job prospects and
support workers. Moreover, low labor force participation may mean that,
even when good economic times return, mobilizing the pool of available
workers will take more time.

112 | Chapter 3
 Box 3-2: Changes in Labor Force Participation
for Different Subpopulations
Overall, the most important factor affecting the aggregate partici‑
pation rate in the recession and recovery has been the aging of the popu‑
lation. But there are a number of important trends and developments
relevant for understanding the changes in participation of different
subgroups of the population:
• Increased participation by older Americans, which may be
attributable to an increase in skills among this population and also to
changes in Social Security retirement benefits;
• Reduced participation by younger Americans as they stay in
school longer;
• Continuation of an at least 65-year long trend of declining male
labor force participation, which is especially stark for young minority
men; and
• Tapering of the long-term trend of increasing female labor force
participation, which dates back to before World War II.
All told, these different trends and factors roughly offset each
other, but they are important for understanding these groups and for
informing policy choices.

Table 3-i
Labor Force Participation Rate by Selected Groups
Average Change Per Year (Percentage Points)
2014:Q4 1948*-1990 1990-2007 2007-2014
All 62.8 0.2 0.0 -0.5
Men 69.1 -0.2 -0.2 -0.6
Women 56.9 0.6 0.1 -0.3
Age 16-24 55.5 0.2 -0.5 -0.6
Age 25-54 80.8 0.4 0.0 -0.3
Age 55+ 40.0 -0.3 0.5 0.2
White* 62.9 0.2 0.0 -0.5
Black* 61.4 0.2 0.0 -0.4
Hispanic* 66.2 0.4 0.1 -0.4
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations. Not all groups have
information starting in 1948, for those groups (marked with a star), the 1948-1990 change is from the
first year for which data is available.

Taking a longer view, as in Figure 3-5, the labor force participation

rate increased from 60.8 percent to 66.6 percent between 1973 and 1995. As
described in Chapter 1, this increase during the “Age of Participation” can
be entirely accounted for by increased participation among women: over this

Achievements and Challenges in the U.S. Labor Market | 113

 Figure 3-5
Labor Force Participation by Gender, 1950–2014
Percent of Civilian Noninstitutional Population Age 16+
90

80
Men Dec-2014

70

60
Overall

50

40
Women

30
1950 1960 1970 1980 1990 2000 2010
Source: Bureau of Labor Statistics, Current Population Survey.

period, the female participation rate increased from 44.7 percent to 59.0 per‑
cent while the male participation rate fell from 78.8 percent to 75.0 percent.
Since 1995, however, the participation rate has fallen from 66.6 per‑
cent to 62.8 percent in the fourth quarter of 2014, with 3.2 percentage points
of this decrease occurring since the fourth quarter of 2007. While some of
this time period coincides with the Great Recession, it also coincides with
the period when the eldest baby boomers entered their peak retirement
years; the first baby boomers turned 62 in 2008, becoming eligible for Social
Security. This demographic shift had been predicted to lower the participa‑
tion rate well in advance of the Great Recession (Aaronson et al. 2006).
Although population aging explains much of the decline in labor force
participation seen in recent years, longer-run trends, cyclical responses, and
other factors also affect participation. CEA evaluated these various factors
in its comprehensive report, The Labor Force Participation Rate Since 2007:
Causes and Policy Implications, summarized in this chapter. This analysis
finds that a combination of demographic changes and typical business-cycle
effects can explain most, but not all, of the decrease since 2007.

Decomposing the Decline in Participation Since 2007

The decline in labor force participation between the fourth quarter of
2007 and the fourth quarter of 2014 can be decomposed into three parts: an

114 | Chapter 3
 Figure 3-6
Labor Force Participation Decomposition, 2009–2014
Percent of Civilian Noninstitutional Population Age 16+
66.0

65.5

Aging Trends
65.0

64.5
Actual Cyclical Effects
64.0

63.5 Residual

63.0

2014:Q4
62.5
2009 2010 2011 2012 2013 2014
Note: Year axis denotes first quarter of year noted. See text for methodological details.
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

aging population, the economic downturn, and a residual that is attribut‑

able to other factors. Figure 3-6 shows the decomposition of this decline
over time based on CEA modeling. By the close of 2014, the participation
rate was down 3.2 percentage points since the end of 2007. Of this, CEA
analysis attributes 1.7 points to long-run aging trends, and 0.5 point to
poor business-cycle conditions. The remaining 0.9 point is not due to either
standard business cycle or aging trends.4 This residual component emerged
in 2012 and grew over the subsequent few years.
CEA’s finding that aging trends explain more than one-half of the
decline in labor force participation over the course of the recession and
recovery is consistent with a wide range of studies that have used a variety
of methodological approaches to better understand the impact of various
factors on the participation rate. These studies, summarized in Table 3-2,
show that research finds that long-term trends such as aging account for
between 25 and 82 percent of the participation decline over the recession.
These findings are not directly comparable, as the time periods they study
differ. Consequently, CEA’s model is estimated over the same time period
as each of these studies, with the results presented in the final two columns
of Table 3-2. CEA’s model finds an aging effect that is between 39 and 55
percent of the decline depending on the time period being analyzed. CEA’s

4 The three components do not sum to the whole due to rounding.

Achievements and Challenges in the U.S. Labor Market | 115

 Table 3-2
Comparison of Participation Rate Estimates
CEA
Estimated
Shares of the
Shares Over
Time Period Total Decline
Same Time
Period
Trend Cycle Trend Cycle
CEA (2014c) 2007:Q4 – 2014:Q4 55% 17%

Beginning in 2007
CBO (2014) 2007:Q4 – 2013:Q4 50% 33% 48% 25%
S. Aaronson et al. (2014) 2007:Q4 – 2014:Q2 82% 11% 51% 21%
D. Aaronson et al. (2014) 2007: Q4 – 2014:Q3 74% 13% 54% 19%
Erceg and Levin (2013) 2007-2012 17% 55% 55% 42%
Fallick and Pingle (2013) 2007:Q4 – 2013:Q2 75% 16% 53% 35%
Kudlyak (2013) 2007-2012 80% 20% 55% 42%
Shierholz (2012) 2007-2011 31% -- 49% 59%
Van Zandweghe (2012) 2007-2011 42% 58% 49% 59%
Aaronson et al. (2006) 2007-2013 82% -- 48% 25%

Other time periods

CEA (2014) 2011:Q1 – 2014:Q4 77% -39%
Fujita (2014) 2000:Q1 – 2013:Q4 65% 30% 39% 19%
Aaronson, Davis and Hu (2012) 2000-2011 40% -- 43% 43%
Source: Cited studies; CEA calculations.

estimate of the aging effects accounting for slightly more than one-half of
the decline between 2007 and the end of 2014 is therefore roughly in the
mid-range of the literature.
The variation across estimates of the cyclical component in the final
column shows that different magnitudes of this component in the literature
are largely driven by the time period of analysis, not variation in analytical
methods. Comparing estimates from the literature to those from the CEA
model in the same time period, the CEA estimate of the cyclical effect is
roughly in the middle of the estimates. The roles of each factor in explaining
the overall change in participation are addressed below.
Aging Population
Lower participation among baby boomers as they aged had been
depressing the participation rate well before 2008, since participation begins
to fall when workers reach their mid-50s. Both men and women decrease
their participation by around 40 percentage points between ages 55 and 65
and participate at even lower rates thereafter. CEA concludes that the aging
population is the single most important factor depressing the participation
rate, accounting for 1.7 of the 3.2 percentage point decline, or more than

116 | Chapter 3
 Figure 3-7
Detrended Participation Rate and
(Inverted) Unemployment Gap, 1960–2014
Percentage Points Percentage Points
1.2 -6
2014:Q4
0.9
Participation -4
Rate
0.6 (left axis)
-2
0.3

0.0 0

-0.3 Unemployment
2
Gap
-0.6 (right axis)
4
-0.9

-1.2 6
1960 1970 1980 1990 2000 2010
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

one-half of the decline, since the end of 2007. This finding is robust to dif‑
ferent methods of modeling the effect of aging on participation, as described
in more detail in The Labor Force Participation Rate Since 2007: Causes and
Policy Implications (CEA 2014c). The effect of aging has also been growing
in magnitude in recent years. The youngest baby boomers will not turn 65
until 2029, so aging will continue to depress labor force participation in
coming years.
Business-Cycle Effects
Economic contractions historically result in both greater unemploy‑
ment and lower labor force participation (Elsby, Hobijn, and Sahin 2010).
Therefore, while movements in the participation rate over decades are
driven largely by the long-term trends, in the short- and medium term,
cyclical factors play a role.
Figure 3-7 shows the cyclicality of the participation rate by comparing
the detrended participation rate and the (inverted) detrended unemploy‑
ment gap, defined as the difference between the unemployment rate and
CBO’s estimate of the natural rate of unemployment.5 For example, the
detrended participation rate declined in the 1990s expansion and rose
during the Great Recession. Visual inspection further suggests that move‑
5 Detrending was performed using the methods described in CEA (2014c). A trend component
of each series was estimated using a semiparametric procedure. The trend components are then
subtracted from the original data series to produce the series shown in Figure 3-7.

Achievements and Challenges in the U.S. Labor Market | 117

 Box 3-3: Post-Recession Participation in the
United States and United Kingdom
In late 2014, the U.S. and U.K. economies exhibited some striking
similarities. The two countries’ year-end unemployment rates were
nearly identical, at 5.6 percent in the United States in December versus
5.8 percent in the United Kingdom as of the three months ending in
November. Moreover, the International Monetary Fund predicted in
October that the United Kingdom and the United States would see the
fastest year-ahead GDP growth among G-7 economies, although output
in the United States currently exceeds its pre-crisis peak by a substan‑
tially wider margin than in the United Kingdom.
However, some elements of the labor market have followed very
different paths in the two economies. The United Kingdom has seen
overall labor force participation hold roughly steady since 2007, despite
the fact that the demographically adjusted participation series for the
United Kingdom show a downtrend similar to that for the United States
(Carney 2014). Yet more than a quarter of the increase in employment in
the United Kingdom has been in part-time work, whereas all of the jobs
added back in the United States have been full-time. And while average
wages in the United States have been roughly keeping pace with inflation,
U.K. workers have seen large declines in real earnings (Figure 3-iii). The
average weekly inflation-adjusted paycheck for British private-sector
workers is now more than 8 percent below its 2007 average. In short, the

118 | Chapter 3
United Kingdom experienced stable labor force participation at the same
time that many jobs offered fewer work hours and lower pay.
To explain this set of circumstances, Bank of England Governor
Mark Carney (2014) has argued that the United Kingdom experienced a
labor supply surge in the wake of the crisis, with about 1.5 million people
joining the U.K. labor force. Carney suggested this was likely fueled by a
number of factors, including the need for households to rebuild savings
or pay down debt in the wake of the financial crisis, as well as policy
changes that have raised the retirement age for public-sector workers
and introduced more stringent job-search requirements for some welfare
recipients. The U.K. government has also undertaken efforts to improve
job search assistance for unemployed workers, potentially facilitating
faster matches of workers and positions.
Ultimately, the differences between the United States and United
Kingdom on key labor market variables are a puzzle that is not yet fully
understood. To an extent, some of the factors that have affected the
United States and United Kingdom are similar—for instance a high
number of indebted households. It is clear that both the United States
and the United Kingdom face the challenge of facilitating transitions
for workers currently employed in lower-wage and -hour jobs to jobs
offering higher wages and more full-time work. Nevertheless, these dif‑
ferent experiences are also a reminder of the many possible paths from
recession to recovery.

Achievements and Challenges in the U.S. Labor Market | 119

ments in the participation rate lag movements in the unemployment rate by
perhaps a year or so. CEA estimates that business-cycle effects explain 0.5
percentage point (about one-sixth) of the total decline in labor force partici‑
pation between the end of 2007 and the end of 2014.6
As the labor market continues to recover, business cycle effects should
wane. For example, cyclical factors depressed the participation rate by 1.1
percentage point in 2011 when the unemployment rate was about 9 percent,
but by the fourth quarter of 2014, the unemployment rate had fallen to 5.7
percent and cyclical factors had shrunk to 0.5 percentage point.
Other Factors
While most of the decline in the participation rate since the end of
2007 is due to the combination of the aging population and standard cycli‑
cal effects, 0.9 percentage point, or a little over one-quarter, of the decline
is not fully understood. CEA’s analysis finds that this portion of the decline
is not explained by either the aging of the population or the normal cycli‑
cal impact of the current recession. Between 2007 and 2012 the decline in
participation is fully (and at some points more than fully) explained by the
aging of the population and standard business-cycle effects. Beginning in
2012, however, the labor force participation rate decline began to exceed
what was predicted from aging and cyclical factors. Since late 2013, the labor
force participation rate has stabilized and the portion of the decline that was
unexplained shrank, albeit slowly, between the second and fourth quarters
of 2014 (Figure 3-6).
One driver of this unexplained component may be long-term trends
within age groups. There was a general downward trend in participation
rates prior to 2007, even after conditioning on age. In the case of prime-age
men, the decline dates back to at least 1950; as noted in Chapter 1, prime-
age male participation declined 0.1 percentage point a year between 1948
and 1973 and then 0.2 percentage point a year since 1973. More recently,
prime-age female participation has declined at 0.1 percentage point a year
on average since 1995. Because of these general trends toward lower partici‑
pation, pre-recession models predicted a decline in participation over this
period—greater than what would be predicted based on aging alone—even
with the assumption of no major recession (Aaronson et al. 2006).
A second set of explanations is that the unexplained portion reflects
the very severe nature of the Great Recession, which led to a greater-than-
normal cyclical relationship between unemployment and participation.

6 CEA uses the unemployment gap as a measure of the state of the business cycle. CEA
regresses the quarter-on-quarter difference in the detrended labor force participation rate on
the contemporaneous year-over-year difference in the detrended unemployment gap, along
with a one-year lag and a two-year lag.

120 | Chapter 3
CEA’s model assumes that the relationship between the unemployment
rate and the labor force participation rate remained the same as in earlier,
shallower recessions. However, the particularly long average duration of
unemployment in the last recession might discourage participation even
more. Adding unemployment duration to the model explains a part of the
previously unexplained portion. Thus, the model suggests that a recession
that leads to greater long-term unemployment leads to greater declines in
labor force participation, conditional on the unemployment rate.
CEA’s analysis finds no unusual rise in disability insurance in
response to the recession—in fact, disability insurance rose less than would
be predicted based on the severity of the recession—so this does not account
for the unexplained decline in participation. The rise in schooling also does
not account for the unexplained portion. Overall, it is likely that a combina‑
tion of factors, including both non-aging trends and factors unique to the
Great Recession, played a role in the participation-rate decline.

Outlook for the Participation Rate

While the evolution of the participation rate is subject to uncertainty,
it is unlikely that the trend of decreasing labor force participation will
reverse in the medium-term without policy changes. As of the fourth quarter
of 2014, the cyclical effect depressed the labor force participation rate by 0.5
percentage point. In the short-run, as the economy fully recovers from the
Great Recession, the cyclical component should dissipate, adding this 0.5
percentage point to the participation rate. At the same time, however, as
more baby boomers retire, the aging population will depress the participa‑
tion rate by roughly an additional 0.25 percentage point each year. The size
of this aging effect is projected to grow gradually from 0.24 percentage point
in 2015 to 0.27 percentage point in 2022, at which point the magnitude of the
effect is expected to start receding. That older workers are able to retire is in
many ways a positive development. But it also creates challenges, especially
for overall fiscal policy and, in particular, for programs like Social Security
and Medicare.
The unexplained component of the participation decline is subject
to greater uncertainty. To the extent that the decline represents trends that
pre-date the Great Recession, it could persist. However, if the unexplained
portion primarily reflects temporary factors related to the Great Recession,
as the economy recovers, the participation rate may increase more than what
cyclical factors alone predict. However, under a range of feasible scenarios,
it is likely the labor force participation rate will continue to decline in the
medium-term.

Achievements and Challenges in the U.S. Labor Market | 121

 Long-Term Unemployment
In 2014, not only did the annual unemployment rate fall by more
than any year since 1984, but also most of the decline came from a decrease
in long-term unemployment. The long-term unemployed, defined as those
unemployed for 27 weeks or longer, accounted for 37 percent of the unem‑
ployed population as of December 2013. Nearly two-thirds of the 2014
decrease in unemployment resulted from a decrease in long-term unem‑
ployment, and by December 2014 they were 32 percent of the unemployed
(Figure 3-8).
Broader measures of unemployment also fell slightly faster than the
overall unemployment rate in 2014, while labor force participation was
largely stable, suggesting that this reduction in long-term unemployment
reflects workers finding employment rather than leaving the workforce
or becoming discouraged. While this constitutes important progress, the
long-term unemployment rate remains elevated relative to its pre-recession
average.

Figure 3-8
Share of Recovery in Overall Unemployment Rate Due to
Declines in Short- and Long-Term Unemployment
Percent
Short-Term Unemployment (26 Weeks or Fewer)
90 Long-Term Unemployment (27 Weeks or More)
80

70 64 64
60

50

40 36 36

30

20

10

0
From Oct-2009 to Dec-2013 From Dec-2013 to Dec-2014
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

122 | Chapter 3
 Figure 3-9
Unemployment Rate by Duration, 2000–2014
Percent of Labor Force
8

6
Unemployed 26 Dec-2014
5 Weeks or Fewer

3
Unemployed 27
2 Weeks or More

0
2000 2002 2004 2006 2008 2010 2012 2014
Note: Shading denotes recession. Dashed lines represent pre-Great Recession (December 2001-
December 2007) averages.
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

Trends in Long-Term Unemployment

In the previous expansion, the short-term unemployment rate (work‑
ers unemployed for less than 27 weeks) averaged 4.2 percent of the labor
force while the long-term unemployment rate averaged 1.0 percent. Both
types of unemployment increased in the recession, with a markedly larger
surge in long-term unemployment, as shown in Figure 3-9. Both have since
substantially recovered, and Figure 3-9 shows that as of December 2014
the short-term unemployment rate was below its pre-recession average,
although the long-term unemployment rate remained elevated. However, as
discussed earlier, the long-term unemployment rate recovered more relative
to the short-term unemployment rate in 2014.
The Great Recession saw a larger than typical increase in both the
number and the share of the long-term unemployed. The number of long-
term unemployed rose from 1.3 million at the end of 2007 to 6.8 million in
April 2010, or 46 percent of all unemployed workers. By December 2014,
however, this number had fallen to 2.8 million workers, or 32 percent of
unemployed workers. By comparison, between 1948 and 2001, workers
unemployed for at least 27 weeks accounted for about 12 percent of unem‑
ployed workers on average with a previous peak share of 26 percent in June
1983. The share of the unemployed who are long-term unemployed of
longer durations also rose sharply in the recession, as shown in Figure 3-10.

Achievements and Challenges in the U.S. Labor Market | 123

 Figure 3-10
Share of Unemployed Workers by Duration of
Unemployment, 2002–2014
Percent of Unemployed
50
27 Weeks or More
45
Dec-2014
40

35
52 Weeks or More
30
25

20
99 Weeks or More
15
10

5
0
2002 2004 2006 2008 2010 2012 2014
Note: Calculations are a 12-month moving average of the share of unemployed by duration as a
share of the overall unemployed population.
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

That figure also shows that even among the long-term unemployed, there
have been greater improvements for those more recently unemployed.
This rise in the prevalence and severity of long-term unemployment
in the Great Recession may in part be a continuation of longer-term trends
in the cyclical pattern of long-term unemployment. Compared to recessions
in earlier decades, the past several recessions have seen sharper increases in
the share of the unemployed who are long-term unemployed as the unem‑
ployment rate climbs, as shown in Figure 3-11.
Moreover, aside from changes during business cycles, there appears to
have been a secular increase in the long-term share of the unemployed for
decades before the crisis occurred.7 Figure 3-12 shows a gradual increase
in the share long-term unemployed since 1948, when the data are first
available.8 The estimates suggest that, between 1948 and 2007, the share of
the unemployed out of work for 27 weeks or more increased by about 0.2
percentage point a year on average.
If the share of unemployment that is long term returns to trend
at the end of 2016, it would be about 20 percent, well above its October
2006 trough of 16 percent. However, recent cycles suggest that the long-
term upward trend may be increasing even during expansionary periods.

7 Also reported in Aaronson, Mazumder and Schechter (2010).

8 The linear time trend is not adjusted for business cycles.

124 | Chapter 3
 Figure 3-11
Increase in Long-Term Unemployment as a Percent of Increase in
Overall Unemployment Rate
Percent
60 56

49
50 46

39
40
32
30 26

19
20 16
11 12
10

0
1948 1953 1957 1960 1969 1973 1980 1990 2001 2007
Recession Start Year
Note: Increases are measured from the first month of the recession to the peak in the overall
unemployment rate. The 1980s recessions are consolidated into a single cycle.
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

Figure 3-12
Long-term Unemployed as Share of Total Unemployed, 1960–2014
Percent
50
Dec-2014
45

40

35

30

25 Actual
20

15
Linear Time
10 Trend

0
1960 1970 1980 1990 2000 2010
Note: Time trend projection is based on data from 1948 through 2007.
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

Achievements and Challenges in the U.S. Labor Market | 125

Moreover, during the Great Recession, long-term unemployment increased
even more than would have been expected from the historical relationship
(Aaronson, Mazumder, and Schechter 2010), suggesting that while long-run
trends have contributed to higher rates of long-term unemployment, other
factors may contribute to a more persistent increase.

Factors behind Elevated Rates of Long-Term Unemployment

The likelihood of finding new employment falls as an unemployment
spell extends, as shown in Figure 3-13. During the Great Recession, the
long-term unemployed were 20 to 40 percent less likely than the short-term
unemployed to obtain employment within two years (Krueger, Cramer, and
Cho 2014). In addition, audit studies show that callback rates from prospec‑
tive employers decline with the length of unemployment (Kroft, Lange, and
Notowidigdo 2013; Ghayad 2013).
The literature offers potential explanations for why the long-term
unemployed are less likely to find employment than the short-term unem‑
ployed. One explanation, “worker heterogeneity,” argues that the long-term
unemployed are different from the short-term unemployed in ways that
make them less attractive to employers, which extends how long they must
search to land a new job (Pries 2008). However, this is less likely to be true
following a deep recession. Moreover, research by Krueger, Cramer, and
Cho (2014) and Mitchell (2013) find that the long-term unemployed resem‑
ble the short-term unemployed on many dimensions. Kroft et al. (2014)
and Aaronson, Mazumder, and Schechter (2010) reach similar conclusions,
and show that rates of long-term unemployment increased for nearly all
demographic, occupation, industry, and regional groups during the Great
Recession.
This research suggests that another explanation for why the long-term
unemployed are less likely to be hired is more relevant to our recovery:
that becoming long-term unemployed itself makes it harder to escape from
unemployment. Employers may interpret a spell of long-term unemploy‑
ment as a negative signal of a worker’s ability because of stigma (Blanchard
and Diamond 1994; Kroft, Lange, and Notowidigdo 2013). Additionally,
employers’ hiring processes may lead to discrimination against the long-
term unemployed by, for example, screening out all workers with a long
spell of unemployment regardless of their other qualifications (Ghayad
2013). Research has shown that the long-term unemployed conditional on
all other characteristics remaining the same are less likely to get called for
interviews (Kroft, Lange, and Notowidigdo 2013). Another explanation is
that as people remain out of work for extended periods of time, their general
and job-specific skills or connections to industry may deteriorate (Edin and

126 | Chapter 3
 Figure 3-13
Monthly Job Finding Rates by Duration of
Unemployment in Previous Month, December 2014
Probability of Reemployment, Percent
35 33

30

25
22

20
17
15
15

10 8

0
Less than 5 5-14 Weeks 15-26 Weeks 27-52 Weeks 53+ Weeks
Weeks
Note: Seasonally-adjusted data as of December 2014. Data refer to the probability of reemployment in
December 2014 based on duration of unemployment in November 2014.
Source: Bureau of Labor Statistics, Current Population Survey.

Gustavsson 2005; Autor et al. 2015). These explanations are not mutually
exclusive, and both could affect the likelihood of transitioning from unem‑
ployment to employment (Jackman and Layard 1991).
Pre-recession patterns in the rate of transition from long-term unem‑
ployment to employment, controlling for duration of unemployment, do a
good job predicting these transitions during this recovery (Kroft et al. 2014).
This implies that, despite the much larger, more diverse pool of long-term
unemployed as compared with past recessions or even non-recessionary
periods, transitions from long-term unemployment back to employment
are not any faster. Unemployment duration appears to be more important
than worker characteristics in determining the transition back to employ‑
ment. However, the long-term unemployed were more likely during the
recession and recovery to stay in the labor market than past transition rates
from long-term unemployed to out of the labor force would have predicted.9
Some research suggests that the extensions of unemployment insurance
encouraged the long-term unemployed to continue looking for work and
reduced the likelihood that they exited the labor force (Krueger, Cramer,

9 Specifically, Kroft et al. (2014) show that the transition probability from unemployment to
non-employment fell markedly over the recession and began to recover around 2010. Their
transition probabilities are constructed from a series in which monthly flows are harmonized
to stocks for the employed, unemployed, and non-participants.

Achievements and Challenges in the U.S. Labor Market | 127

and Cho 2014; Aaronson, Mazumder, and Schechter 2010; Kroft et al. 2014;
Rothstein 2011).

Why Long-term Unemployment Matters

Higher levels of long-term unemployment are concerning because
they place greater strain on household resources and sometimes necessitate
drastic changes in household behavior, such as selling a home or postpon‑
ing medical care, which can have disruptive impacts on family members,
the wider community, and the economy. Long-term earnings loss after
resuming work also appears to increase with the duration of unemploy‑
ment (Schmieder, von Wachter, and Bender 2013; Addison and Portugal
1989). Moreover, it does not appear that these earnings losses are unique
to experiencing unemployment during an economic expansion or recovery,
nor are they concentrated in the manufacturing or service sector (Couch
and Placzek 2010). Former Federal Reserve Chairman Ben Bernanke has
said that long-term unemployment “imposes economic costs on everyone,
not just the unemployed themselves,” as their loss of skills and lower rates
of employment reduce the economy’s overall productive capacity (Bernanke
2012).

Part-Time Work for Economic Reasons

Part-time employment tends to grow in recessions as some businesses
hold on to workers by cutting their hours, and those businesses continuing to
hire may need only part-time hours from new workers. Between December
2007 and December 2009, the share of the labor force usually working part-
time rose from 16.1 percent to 17.9 percent. This increase was driven by
a large rise in people working part-time for economic reasons, defined as
employees who would prefer to have full-time work but either cannot find
a full-time job or have a job that does not provide full-time hours (even if it
once did). As the economy has recovered, the share of the labor force that
is part-time has begun to recede as all the growth in employment has been
driven by growth in full-time employment, as Figure 3-14 shows. Five years
into the recovery, more than 9 million more people are working full-time,
while the number of people employed in part-time jobs has been largely
unchanged. Moreover, part-time jobs have been increasingly held by those
who say they do not want to work full-time.
Rates of part-time employment for economic reasons doubled dur‑
ing the recession from 3 percent to 6 percent, exceeding the previous peak

128 | Chapter 3
 Figure 3-14
Net Change in Employment Since January 2010,
Household Survey Estimates
Millions of Workers
10 Dec-2014
9
8
7
Full-Time
6
5
4
3
2
1 Part-Time
0
-1
-2
2010 2011 2012 2013 2014
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

reached in 1982, as shown in Figure 3-15.10 The share of the labor force
working part-time for economic reasons has since fallen, and the pace of the
decline in this share picked up during 2014, declining 0.7 percentage point
over the 12 months ending in December 2014. The rate is 4.3 percent as of
December 2014, 54 percent of the way back to its pre-recession average, with
over one-third of this overall progress occurring in 2014.

Patterns in Part-Time For Economic Reasons

As a general rule, the share of workers who are part-time but would
prefer full-time work rises in a downturn and then trends slowly back down
during the recovery and boom. As Figure 3-15 shows, in a typical business
cycle rates of part-time employment rise and these jobs go disproportion‑
ately to those who would prefer full-time work, with rates of part-time work

10 Care must be taken when comparing the share of workers who are part-time for economic
reasons before and after the 1994 redesign of the Current Population Survey. CEA used the
multiplicative adjustment factors reported by Polivka and Miller (1998) in order to place the
pre-1994 estimates of the part-time for economic reasons rate on a comparable basis with post-
redesign estimates. For the part-time series for which Polikva and Miller do not report suitable
adjustment factors, the pre- and post-redesign series were spliced by multiplying the pre-1994
estimates by the ratio of the January 1994 rate to the December 1993 rate. This procedure
generates similar results to the Polikva and Miller factors for series for which multiplicative
factors are available.

Achievements and Challenges in the U.S. Labor Market | 129

 Figure 3-15
Rates of Part-Time Work, 1960–2014
Percent of Labor Force
20
Dec-2014
18 Total Part-Time
16

14

12 Part-Time for Non-

Economic Reasons
10

6 Part-Time for
Economic Reasons
4

0
1960 1970 1980 1990 2000 2010
Note: Shading denotes recession. See footnote 10 for details on comparability over time.
Source: Bureau of Labor Statistics, Current Population Survey; Polivka and Miller (1998); CEA
calculations.

for other reasons declining. This shift likely reflects several factors: firms
finding it easier to hire highly qualified workers for part-time jobs since
fewer full-time jobs are available, and therefore hiring more people for part-
time work who would prefer full-time work; firms cutting hours of full-time
employees who are unable to find full-time work elsewhere; and workers in
part-time jobs increasing their preferences for full-time work as household
income falls (Bednarzik 1975; Bednarzik 1983; Maloney 1987).
Figure 3-15 also shows that, following some recessions, the rate did
not fully recover to its prerecession low before rising again. This is partially a
result of the fact that the relationship between unemployment and part-time
for economic reasons has varied across recessions and may also be due partly
due to differences in the length of the recovery period. Figure 3-16 reports
the change in the share of the labor force working part-time for economic
reasons relative to the change in the unemployment rate during contractions
and expansions over the last five decades. Like the current cycle, both the
1980s recessions and the 2001 recession saw above-average increases in part-
time employment for economic reasons for a given percentage point rise in
the unemployment rate, but did not see commensurately rapid declines as
the unemployment rate declined in the ensuing expansion.
Figure 3-17 uses the relationship between part-time employment for
economic reasons and unemployment from prior recessions and the path
of unemployment during the current business cycle to predict the path of

130 | Chapter 3
 Figure 3-16
Change in Share Part-Time for Economic Reasons Per
Percentage-Point Change in the Unemployment Rate, 1957–2014
Ratio
0.7 0.65
Contraction Expansion
0.61
0.6 0.57

0.5 0.47 0.47

0.45
0.41 0.40
0.4 0.36
0.35 0.36
0.32
0.29
0.3
0.21
0.2 0.17
0.13
0.1

0.0
1957 1960 1969 1973 1980 1990 2001 2007
Business Cycle
Note: The 1980s recessions are consolidated into a single cycle. The expansion period runs through 22
quarters or until the next peak, whichever is earlier.
Source: Bureau of Labor Statistics, Current Population Survey; Polivka and Miller (1998); CEA
calculations.

part-time employment for economic reasons. Consistent with the patterns

described in the last paragraph, predictions based on the 1980s recessions
and the 2001 recession generate a path similar to that observed during
the current business cycle: a relatively sharp initial increase, followed by a
recovery that, while steady, does not match the pace of the initial increase
and, thus, leaves part-time employment for economic reasons elevated.
Modeling the path in this recession using relationships from other post-1957
recessions generates a much smaller initial increase but a broadly similar
pace of recovery.
Figures 3-16 and 3-17 imply that the mystery of part-time employ‑
ment for economic reasons in the Great Recession (as well as of recessions
in the 1980s and 2001) is the sharper increase of such work during the con‑
traction, not a lack of full-time job creation during the recovery. Similarities
across the 1980s and 2001 recessions suggest that the behavior of part-time
employment for economic reasons in the 2007 recession may not be due to
factors unique to the Great Recession, like its depth or duration. Instead, it
may reflect longer-term changes in the cyclical sensitivity of this measure,
suggesting that this challenge may return in future recessions.

Achievements and Challenges in the U.S. Labor Market | 131

 Figure 3-17
Share Part-Time for Economic Reasons,
Actual and Predicted, 2005–2014
Percent of Labor Force
7

6
Actual 2014:Q4
5 Predicted Based
on 2001 Cycle
4 Predicted Based
on 1980s Cycles
3 Predicted Based
on Other Post-
1957 Cycles
2

0
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Note: The 1980s recessions are consolidated into a single cycle.
Source: Bureau of Labor Statistics, Current Population Survey; Polivka and Miller (1998); CEA
calculations.

The Outlook for the Rate of Part-Time for Economic Reasons

The question arises of whether the share of employees who work
part-time for economic reasons will remain elevated over the long term. The
answer depends in large part on the reasons behind this elevation.
One possibility is that this type of part-time employment remains
elevated because it recovers later, even after the headline unemployment rate
has fully recovered. The view suggests that part-time workers who prefer
full-time work will accept more hours or a full-time job if it becomes avail‑
able, and therefore they represent a pool of available workers to businesses
wishing to expand employment. In this situation, a higher share of workers
who are part-time for economic reasons indicates that there is more slack
in the labor market than is suggested by a given unemployment rate. If this
interpretation describes our current labor market, and the robust labor mar‑
ket momentum seen over 2014 continues, then the rate of part-time work for
economic reasons should continue to decline in the years ahead, ultimately
returning to pre-recession levels assuming the economy remains strong for
long enough. Some evidence consistent with this scenario comes from the
rapid decline in this rate in recent months, even measured relative to the
increased pace of progress in reducing unemployment. Over 2014, the rate
of part-time work for economic reasons has declined by 0.5 percentage point
for each percentage-point reduction in the unemployment rate, whereas

132 | Chapter 3
it declined, on average, by 0.3 percentage point for each percentage-point
reduction in the unemployment rate since the start of 2010. Furthermore,
experience from the late 1990s and mid-to-late 1960s provides historical
precedent: part-time employment for economic reasons rapidly decreased
relative to overall unemployment during these strong labor market periods.
On the other hand, another possibility is that recent recessions have
accelerated ongoing structural changes that cause employers to demand
more part-time workers relative to full-time workers. In this scenario, the
part-time for economic reasons rate may remain elevated even once the
unemployment rate has fully recovered, depending on the supply of part-
time workers. The more rapid recovery in the goods sector relative to the
service sector may provide some evidence that employer demand for part-
time workers in the service sector has shifted. To the extent that the overall
rate remains elevated mainly due to the incomplete recovery of the labor
market, that incomplete recovery might be expected to affect both sectors
similarly (Figure 3-18).
The timing of the shifts in part-time work also suggest that the
Affordable Care Act’s employer responsibility provision, which requires
large employers to offer coverage to employees working 30 or more hours
per week or pay a penalty, is not playing a meaningful role in recent trends in
part-time work. First, both the share of the labor force working part-time and
the share in part-time jobs who would prefer to be in full-time jobs declined
more sharply in 2014 than in the earlier years of the recovery. In contrast,
if the Affordable Care Act’s employer responsibility provision was driving
a substantial structural increase in the demand for part-time workers, one
would, all else equal, have expected the opposite—that progress in reducing
part-time employment would have slowed over the months leading up to the
provision’s implementation in 2015. Second, the most striking way in which
the behavior of part-time employment, particularly among those who would
prefer full-time, in the most recent recession and recovery differs from prior
recessions is that it rose unusually sharply during the contraction, not that
it has fallen unusually slowly during the recovery, as discussed above. This
unusually sharp increase occurred essentially entirely before the Affordable
Care Act became law in March 2010 and many years before employer
responsibility took effect, so it cannot have been caused by the Affordable
Care Act. Finally, as noted earlier, other recent recessions—most notably the
2001 recession and, to a lesser extent, the 1980s recession—also experienced
sharp rises in the rate of involuntarily part-time workers that were not fully
reversed by this point in the ensuing recovery, so the phenomenon may tell
us more about a structural shift in the economy in the last several decades.

Achievements and Challenges in the U.S. Labor Market | 133

 Figure 3-18
Share of Employees Working Part-Time for
Economic Reasons, by Industry, 1995–2014
Percent of Industry Employment
9

7
Dec-2014
6

5 Goods
Services
Industries Industries
4

0
1995 2000 2005 2010
Note: Data are 12-month moving averages of non-seasonally adjusted data. Shading denotes recession.
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

Labor Market Fluidity

Labor market fluidity (used interchangeably in this chapter with
“dynamism” or “churn”) refers broadly to the frequency of changes in who
is working for whom in the labor market. From the worker’s perspective,
this is measured by hires and separations; from the firm’s perspective, it
is measured by new positions (job creation) and eliminated positions (job
destruction). Although separations, hires, creation, destruction, and other
measures capture different concepts of fluidity, increases in these measures
generally indicate more fluidity.
A range of measures suggest that fluidity has risen in the labor market
recovery, as shown in Figure 3-19.11 The number of new workers hired has
steadily increased: there were 5.0 million workers hired into new positions in
November 2014, compared to 4.6 million in November of the previous year.
The hires rate was 3.6 percent in November, a number that has nearly fully
recovered to its rate of 3.7 percent in the month prior to the recession’s start.
11 The Longitudinal Employer-Household Dynamics (or LEHD) data are a restricted-access
data source compiled and maintained by the Census Bureau. The LEHD data are the result
of matching data across many sources—in particular, by matching household information
from the Census and American Community Surveys to state administrative Unemployment
Insurance system wage records and to employer data from economic censuses. For detail, see
Abowd et al. (2005). The job-to-job (or J2J) data are newly available data constructed from
the LEHD and published by Census. The J2J data provide information on the flows of workers
joining, leaving or changing employers under various circumstances (Hyatt et al. 2014).

134 | Chapter 3
 Figure 3-19
Hires, Separations, and Job-to-Job Flow Rates, 2000–2013
Percent of Employment
16

14
J2J Hires
12 J2J Separations 2013:Q3

10 JOLTS Hires JOLTS

Separations
8

4
J2J Job-to-Job
2 Hires

0
2000 2002 2004 2006 2008 2010 2012
Note: J2J job-to-job hires are generally equal to J2J job-to-job separations (not shown). Shading
denotes recession.
Source: Bureau of Labor Statistics, Job Openings and Labor Turnover Survey; Census Bureau, Job-to-
Job Flows.

Direct transitions of workers from one job to another also show recovery.
Worker flows out of jobs (separations), including voluntary quits, have also
slowly risen during the recovery. Naturally, involuntary separations spiked
during the recession, but recovery in voluntary separations indicates that
workers are feeling comfortable in changing employers, which reflects the
increasing strength of the labor market.
Consistent with the strong employment growth over the last 58
months, the rate of new job openings as a share of total positions is now
above its pre-recession average after falling by more than 40 percent dur‑
ing the recession (Figure 3-20). This increase in job openings offers further
opportunities for workers to change their employment status or situation if
desired. Taken together, these data indicate that greater fluidity has accom‑
panied the labor market strengthening.
While the short-term trend shows increased labor market dynamism,
a growing body of evidence finds that there are long-run downward trends
in fluidity that likely date back several decades. The recent gains in fluidity
measures reflect the strength of the recovery and should therefore generally
be viewed as positive. It is less clear, however, how the long-run decline
should be viewed given that it has the potential for both positive aspects in
terms of job stability and better matches, and negative aspects in terms of
potentially less effective reallocation of labor to its highest productivity uses.

Achievements and Challenges in the U.S. Labor Market | 135

 Figure 3-20
Job Opening Rates, 2000–2014
Vacancies as Percent of Total Positions
4.0

Nov-2014
3.5

3.0

2.5

2.0

1.5
2000 2002 2004 2006 2008 2010 2012 2014
Notes: Shading denotes recession. Dashed line represents 2001-2007 average.
Source: Bureau of Labor Statistics, Job Openings and Labor Turover Survey; CEA calculations.

This section examines these longer-run trends and their potential impact on
the economy.

Trends in Labor Market Fluidity

Recent research has identified long-run declines in a variety of mea‑
sures of worker mobility. Research has shown that workers are less likely to
leave a job, are less likely to move to a new job, and are less likely to physi‑
cally move for a job (Kaplan and Schulhofer-Wohl 2012; Molloy, Smith, and
Wozniak 2014; Hyatt and Spletzer 2013). Research has also identified long-
run declines in dynamism in firm-side measures, including job creation, job
destruction, and the entry and exit of establishments from the marketplace
(Decker et al. 2014; Davis and Haltiwanger 2014). Taken together, this body
of work indicates a U.S. labor market characterized by considerably lower
levels of fluidity of all kinds than was the case two to three decades ago.
Lower Hires and Separations Rates
Worker flows have declined since at least the late 1990s, including
the entire period for which the best direct data on worker flows are avail‑
able from the Job Openings and Labor Turnover Survey (JOLTS, available
since 2001). Hyatt and Spletzer (2013) document declines of 10 percent
(using Current Population Survey data) to 38 percent (using Longitudinal
Employer-Household Dynamics data) in hires and separations since 2001,

136 | Chapter 3
as shown in Figure 3-21.12 Davis and Haltiwanger (2014) have a longer series
on hires and separations that extends back to 1990, which shows a decline in
worker flows over this longer period.
Other studies examine fluidity indirectly by looking at outcomes for
which worker or job flows are likely important, such as flows between labor
market statuses, long-distance migration, and transitions between industries
and occupations. Some of these indirect measures can be calculated over
longer historical periods and also point to long-term declines in fluid‑
ity. Hyatt and Spletzer (2013) find that job-to-job transitions declined by
roughly 50 percent from 1998 to 2010. Davis et al. (2010) show that flows
into and out of unemployment fell by nearly one-half over the two decades
prior to the early 2000s. Long-distance migration in the United States, which
typically involves a change of employer or labor force status, has been in a
decades-long decline, falling by as much as 50 percent since the late 1970s
(Molloy, Smith, and Wozniak 2014; Kaplan and Schulhofer-Wohl 2012).
Industry, occupation, and employer transitions have also fallen markedly
over a similar period, with declines in those measures accelerating since the
1990s, as shown in Figure 3-22.13
Lower Job Creation and Job Destruction Rates
More is known about job flows (job creation and destruction) than
worker flows (hires and separations) since series data are available back
to the 1980s. Literature based on these data concludes that job flows have
markedly declined over the last 20 to 30 years. For example, Decker et al.
(2014) and Davis and Haltiwanger (2014) document that job creation and
job destruction fell from the late 1980s to just prior to the 2007 recession.
Hyatt and Spletzer (2013) find larger declines, of roughly one-quarter to
one-third, for both job creation and destruction between the late 1990s
and 2010. To the degree that this reflects structural improvements in the
economy that lead to more stable jobs, this would be an encouraging trend.
But a potential concern is that it could reflect less reallocation of resources
toward their most productive uses and thus fewer high-paying jobs.
Factors in Decreasing in Labor Market Fluidity

12 Differences in the duration of jobs and types of establishments captured by the three series
explain the level differences. The smaller decline in the Current Population Survey may be
related to the fact that it misses more short-term jobs than does the Longitudinal Employer-
Household Dynamics data (Abraham et al. 2013), and Hyatt and Speltzer (2013) show that the
declining share of short-term jobs can explain some of the decline in hires and separations.
13 A caveat is that some studies using CPS data find less clear trends in transitions for the 1980s
to the 1990s, but again, for the late 1990s onward, the trend is clearly downward. Kambourov
and Manovskii (2009) tabulate occupation mobility from the CPS and find an increasing trend.
Moscarini and Thomsson (2007) characterize the trend in occupational mobility as weakly
increasing in the 1980s. In addition, Stewart (2007) finds no trend in job-to-job flows from the
1980s to the 1990s using the annual retrospective question CPS question.

Achievements and Challenges in the U.S. Labor Market | 137

 Figure 3-21
Trends in Hires and Separations, 1995–2012
Percent of Total Employment
35

30 LEHD Hires

25
CPS Hires
20 2012:Q3

15 JOLTS Hires

10

0
1995 1997 1999 2001 2003 2005 2007 2009 2011
Source: Hyatt and Spletzer (2013); Bureau of Labor Statistics, Current Population Survey; Bureau of
Labor Statistics, Job Openings and Labor Turnover Survey; Census Bureau, Longitudinal Employer-
Household Dynamics.

Figure 3-22
Employer, Occupation, and Industry Transitions, 1983–2013
Percent of Total Population Age 16+
14

12 Employer
Change

10 2013

8 Occupation
Change
6
Industry Change

0
1980 1985 1990 1995 2000 2005 2010
Source: Molloy, Smith, and Wozniak (2014)

138 | Chapter 3
 The empirical literature has only recently begun to examine why
job and worker transitions have fallen. Two basic hypotheses have been
explored: that firms or that workers have changed over time in ways that
lower fluidity. Evidence shows that the first of these can explain a portion
of declining fluidity. The average age and number of associated establish‑
ments per firm have both risen in recent decades (Davis and Haltiwanger
2014; CEA calculations). Older, larger firms are associated with lower job
flows, as these firms are less likely to contract or expand rapidly. Consistent
with this change in firm composition, rates of firm entry and exit have also
declined over the last three decades (Figure 3-23). Because the change in
the composition of firms has shifted in a way that, all else equal, would
suggest fewer worker hires and separations, researchers have tested to see
how much of the shift in worker flows can be explained by changes in firm
composition. Hyatt and Spletzer (2013) and Davis and Haltiwanger (2014)
decompose changes in worker flows into those due to job flows and those
due to worker movements between existing jobs. They find that changes in
job flows account for between one-third to one-half of the decline in worker
flows. Because job flows are determined in part by firm size and age, chang‑
ing firm characteristics contribute to the decline in worker flows (Hyatt and
Spletzer 2013). In contrast, changes in characteristics of the average worker,
like age and education, have been found to contribute little to declines in
fluidity (Molloy, Smith, and Wozniak 2014; Davis and Haltiwanger 2014).

Potential Consequences of Reduced Fluidity

Some explanations for reduced fluidity may be benign. For example,
employers may be increasing efforts to reduce turnover for a variety of
reasons: increased cost of switching workers as job training requirements
increase or better worker-firm matching at the point of hire, to name a few.14
A reduced level of labor market transitions may also have benefits for work‑
ers, like more stable jobs with less disruption that allow them to invest more
in skills that their employer values.
Reduced flows could be cause for concern, however, because they
may undermine workers’ abilities to improve their employment situations.
In particular, reduced fluidity may preclude employees from realizing the
wage gains of switching jobs or make it difficult for part-time workers to find
full-time work or result in fewer high-paying jobs in productive industries.

14 Cairo (2013) finds that job-training requirements have risen over time, which supports a
theory that on-the-job experience has also become more important. Both would likely lead
firms to want to lower turnover. No direct evidence exists on trends in the quality of worker-
firm matches, but a substantial literature outlines the importance of this matching for wages
(Nagypál 2007; Crane 2014; Jovanovic 1979).

Achievements and Challenges in the U.S. Labor Market | 139

 Figure 3-23
Firm and Establishment Entry Rates, 1978–2012
Percent of Total Firms/Establishments
15

14

13

12
Establishments
11

10
2012
Firms
9

7
1975 1980 1985 1990 1995 2000 2005 2010
Source: Census Bureau, Business Dynamics Statistics; CEA calculations.

A growing body of evidence finds that wages and earnings increase substan‑
tially when a worker changes jobs, as summarized in Table 3-3. In general,
workers gain at rates considerably above inflation.
Even when workers ultimately stay with their employer, the potential
for them to land better employment can generate wage growth as incumbent
employers raise wages to retain these workers (Beaudry and DiNardo 1991).
Lower fluidity may reduce workers’ abilities to raise their wages by changing
jobs, and consequently also their bargaining power with their incumbent
employer. In this way, reduced fluidity may contribute to slower wage
growth. Alternatively, lower fluidity may result from limited opportunities
for wage growth through employer transitions. Regardless, Table 3-3 shows
that the gains from switching jobs have varied over time. The largest wage
gains from switching jobs were seen in the late 1990s, while wage gains from
switching jobs in the 2000s were much lower.15
Other consequences of lower fluidity are perhaps more speculative but
warrant careful observation nonetheless. Greater fluidity—or more precisely
the conditions and institutions that enable greater fluidity—may prevent
the share of long-term unemployed from rising, and may thereby reduce
the negative consequences of long-term unemployment. More fluid labor

15 Molloy, Smith, and Wozniak (2014) note that point estimates in both the PSID and NLSY
are similar when the recession years are excluded.

140 | Chapter 3
 Table 3-3
Wage and Earnings Gains Associated with Job Switching
Data Age Gain to Switching
Source Group Time Period Jobs
1957:Q1 -
Topel and Ward (1992) LEED 18 to 34 9%
1972:Q4
1983-1994 4%
PSID 22 to 29 1995-2001 10%
Molloy, Smith, and 2003-2011 2%
Wozniak (2014) 1966-1981 7%
NLSY 22 to 29 1979-1994 3%
2002-2011 4%
1995:Q2 8%
Fallick, Haltiwanger, and
LEHD 25 to 55 1999:Q2 14%
McEntarfer (2012)
2001:Q2 6%
Note: Topel and Ward (1992) and Molloy, Smith, and Wozniak (2014) are wage regression models,
while Fallick, Haltiwanger, and McEntarfer (2012) use sample earnings medians from job switchers.
All regression estimates are statistically significant, except for the Molloy, Smith, and Wozniak (2014)
estimates from the 2000s.

markets may also be more resistant to cyclical shocks, or, at minimum, may
experience faster recoveries after a recession (Blanchard and Wolfers 2000).
If this is the case, the slower recoveries in the shares of part-time for eco‑
nomic reasons and in long-term unemployment in recent recessions could
in fact be related to the long-run decline in fluidity.

Wage Growth and Job Quality

In 2014, average real wages for production and nonsupervisory work‑
ers increased 0.8 percent after increasing 0.7 percent in 2013. Although
not sufficient, these increases are a marked improvement from the 2000s,
including the pre-Great Recession years of 2001 to 2007, when real wage
growth averaged 0.5 percent a year, as shown in Figure 1-4 of Chapter 1.
While these recent wage gains are further evidence of a strengthening labor
market, there is more work to be done to ensure that middle-class families
fully share in the benefits of the recovery.
The evidence presented below shows that 2014 was a strong year for
growth across almost all sectors, but it was particularly strong in several that
have traditionally provided good, middle-class jobs. A longer-run perspec‑
tive, however, shows that over the past several decades the composition of
jobs has shifted toward both high- and low-skilled sectors while employment
in the middle of the skill distribution has declined.

Achievements and Challenges in the U.S. Labor Market | 141

Job Growth in 2014
Not only was 2014 the strongest year for job growth since the 1990s,
but the pickup in growth between 2013 and 2014 occurred more strongly in
industries with higher average wages, as shown in Figure 3-24. For instance,
average weekly earnings for manufacturing workers are about $170 higher
than the average for all private-sector workers, and manufacturing job
growth almost doubled from 10,000 a month in 2013 to 19,000 a month in
2014. Similarly, employment in the construction sector, which has average
weekly earnings about $200 above the private-sector average, rose by an
average of 28,000 a month in 2014, up from 18,000 a month in 2013.16 It
is important to note, however, that this—like any estimate of job growth
by industry or occupation—does not necessarily tell the full story, which
depends not just on job growth across sectors, but also on what is happening
to job growth within sectors as well.

Patterns in Wage Growth since the 1980s

As discussed in Chapter 1 and shown in Table 3-4, for most workers,
earnings gains have not kept pace with productivity gains over the last sev‑
eral decades.17 The official estimate of labor productivity grew an average of
2.0 percent a year between 1980 and 2014. To make it comparable to the real
wage and compensation data used below, CEA adjusted labor productivity
using an index of consumer prices, the CPI-U-RS, yielding an estimate of
1.3 percent annual growth in productivity.18 Over this period, hourly com‑
pensation for the average worker rose 0.9 percent annually, indicating that
compensation did not keep up with productivity growth and that the share
of gross domestic income going to capital was rising. Average hourly wages
(calculated from wage and salary earnings in the CPS microdata) fell even
16 Bureau of Labor Statistics, Current Employment Statistics; CEA calculations.
17 All of the consumer price deflation in Table 3-4, and in this section, is done using the
CPI-U-RS, as is common in the labor literature. The CPI-U-RS is the Consumer Price
Index adjusted backwards to make a methodologically consistent historical series. Footnotes
in this subsection indicate results using an alternative index, the price index for Personal
Consumption Expenditure (PCE) from the National Income and Product Accounts. The PCE
price index has the property relative to the CPI of not covering the same consumer basket as
the one consumers purchase through their wages—for example, it includes Medicare costs
for the government and the costs facing nonprofits. However, the PCE deflator also has the
properties associated with using a chain-weighted index. As a result, PCE-adjustment implies
real wage increases over time that are about 0.3 percentage point per year higher than CPI-
based adjustment.
18 The difference between the two estimates of productivity growth reflects slower growth in
prices of investment goods and the terms of trade, relative to consumption good prices. As a
result, the implicit price deflator used to deflate productivity rises more slowly than consumer
prices over this period. If the labor share was constant, productivity adjusted for consumer
prices should keep pace with wages adjusted for consumer prices.

142 | Chapter 3
 Figure 3-24
Change in Job Growth vs. Average Earnings
by Industry, 2013–2014
Change in Percent Annual Employment Growth, 2014 vs. 2013 (Percentage Points)
3.0 Average Weekly Mining and
Earnings for All Logging
2.5 Private-Sector
Workers: $852.20
Construction
2.0
Transportation and
1.5 Warehousing Utilities
Education and
Temporary Health Services
1.0 Help
Manufacturing
Professional and
Services Business Services*
0.5 Wholesale
Other Services Financial Activities
Trade
0.0
Leisure and Information
-0.5 Hospitality

-1.0 Retail Trade

$200 $400 $600 $800 $1,000 $1,200 $1,400 $1,600
Average Weekly Earnings, December 2014
Note: *Excludes Temporary Help Services (shown separately). Average earnings for Temporary Help
Services are not seasonally adjusted.
Source: Bureau of Labor Statistics, Current Employment Statistics.

further short of productivity growth, rising only 0.6 percent a year, because
they do not include the faster-growing components of compensation like
employer-paid health insurance. Finally, median hourly wages grew only
0.3 percent per year—slower than average wages because the increase in
wage inequality meant larger increases in wages for workers near the top,
raising the average much more than the median. In total, the disconnect
between the 2.0 percent annual productivity growth and the 0.3 percent
annual growth in the median wage reflects the combination of these factors:
a methodological issue involving different price indices, the rapid rise of
benefit costs, and the increase in inequality.19
The slowdown in wage growth has been felt most in the middle and
bottom of the wage distribution. Aside from the late 1990s—a period that
saw rapid wage growth across the distribution—over most of the last three
decades, wages have been stagnant or deteriorating for all except the highest
earners. Figure 3-25 shows that these patterns have led to a widening in wage
inequality since the late 1970s (Juhn, Murphy, and Pierce 1993; Lemieux
2006; Autor, Katz, and Kearney 2008). Between 1979 and 2014, real wages
for the highest earners (the 90th percentile of the wage distribution) have
grown by around 35 percent. At the same time, median wages rose by 8
19 If Table 3-4 were produced using the PCE index, the average annual percent increase would
be 1.6 for labor productivity; 1.2 for compensation; 0.9 for mean wages; and 0.6 for the median
wage.

Achievements and Challenges in the U.S. Labor Market | 143

 Table 3-4
Average Annual Percent Change in Real Productivity,
Compensation, and Wages, 1980–2014
Real Labor Productivity 2.0
Labor Productivity* 1.3
Labor Compensation* 0.9
Mean Hourly Wage (CPS)* 0.6
Median Hourly Wage (CPS)* 0.3
Note: Series marked with (*) are adjusted for inflation using the CPI-U-RS. Wages are calculated using the
same method as Figure 3-25.
Source: Bureau of Labor Statistics, Productivity and Costs; Bureau of Labor Statistics, Current Population
Survey (Merged Outgoing Rotation Groups); Bureau of Labor Statistics, Consumer Price Index; CEA
calculations.

percent while wages at the 10th percentile declined slightly.20 As a result, the
ratio between wages at the 90th and 10th percentiles widened by 37 percent
since 1979. The 90th-to-50th percentile ratio grew by 26 percent, and the
ratio between the 50th and 10th percentiles increased only slightly. As the
figure shows, inequality at the bottom of the wage distribution—that is,
between the 50th and 10th percentiles—grew rapidly during the 1980s and
has been relatively constant since, whereas inequality between the highest
earners and the rest of the distribution has grown since the late 1970s.
Figure 3-25 shows that the lack of wage growth in the lower half of
the wage distribution has been a continuing challenge for more than three
decades. Lee (1999) documents that an important factor explaining this
decline is the erosion of the real value of the minimum wage. Increasing
the value of the minimum wage in 2014 to its real average value in 1979
would have directly increased wages for the lowest 8 percent of wage earn‑
ers.21 Economists have found that the minimum wage can also “spill over”
to increase wages for those with wages above the new minimum, since
employers may adjust their compensation schedules to preserve relative pay
among their workers (Autor, Manning, and Smith 2014). Autor, Manning,
and Smith (2014) find that the effect of the minimum wage on inequality in
the lower part of the wage distribution can be quite substantial: an approxi‑
mately 10 percent increase in the minimum wage, relative to the median
wage, reduces the 50-10 ratio by about 1.5 percent.
20 Using the PCE deflator, 90th percentile wages would have grown by 50 percent, median
wages by 20 percent, and 10th percentile wages by 10 percent. While the levels would be
increased with this deflator, the evolution of inequality—the differences between the levels—is
unaffected by the deflator.
21 Bureau of Labor Statistics, Current Population Survey (Merged Outgoing Rotation Groups);
CEA calculations. Inflation-adjusted using the CPI-U-RS. This is the percentage of workers
making below the 1979 inflation-adjusted value of the minimum wage.

144 | Chapter 3
 Figure 3-25
Wage Inequality, 1979–2014
Real Hourly Wage Index, 1979=100
150
2014
140 90th Percentile

130
120 50th Percentile

110
100
10th Percentile
90
80
70
60
1979 1984 1989 1994 1999 2004 2009 2014
Note: The figure depicts real hourly wage quantiles for workers age 18 to 64, excluding individuals
who are self-employed, who have real wages below $0.50 or greater than $100 (in 1989 dollars), or
whose wages are imputed. Top-coded earnings adjusted following Lemieux (2006). Inflation adjusted
using the CPI-U-RS.
Source: Bureau of Labor Statistics, Current Population Survey (Merged Outgoing Rotation Groups);
CEA calculations.

The Rise of the Skill Premium and Employment Growth in High-

and Low-Skill Occupations
The rise in inequality shown in Figure 3-25 is also seen in earnings
differentials for workers with different levels of education. Since the 1980s,
the college income premium—the ratio of income among workers with at
least a college education to workers with only a high school diploma—has
increased. In 1963, men and women with college educations earned incomes
33 and 76 percent higher, respectively, than men and women with only
high school diplomas. Since about 1980, however, these income gaps have
widened so that by 2013, college-educated workers’ incomes were more than
twice the incomes of high school graduates.
Economists Claudia Goldin and Lawrence Katz (2010) explain this
phenomenon as a “race” between technological advancements that increase
the demand for highly-skilled workers and the supply of such workers.
In particular, they document a slowdown in the growth of the college-
educated workforce around 1980. This slowdown has meant that growth in
the demand for skills (technology) outpaced growth in the supply of skills
(educational attainment of workers), leading the college earnings premium
to increase.
In spite of the long-term rise in demand for skill, employers appear
to be offering less training than in the 1990s (Figure 3-27). To some extent,

Achievements and Challenges in the U.S. Labor Market | 145

 Figure 3-26
College Income Premium by Gender, 1963–2013
Ratio of College Income to High School Income
2.50
2013
Women
2.25

2.00

1.75

Men
1.50

1.25

1.00
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
Note: Income premia calculated using median annual income of persons 25 and older. Prior to 1991, "high
school graduates" refers to respondents with 4 years of high school, and "college graduates" refers to
respondents with at least 4 years of college.
Source: Bureau of Labor Statistics, Current Population Survey (Annual Social and Economic Supplement);
CEA calculations.

these changes may reflect shifts in industry structure: historically, jobs with
high vocational requirements are most likely to offer on-site training and
financial assistance (Altonji and Spletzer 1991). Nevertheless, it appears that
fewer workers are able to acquire new skills either on the job or with the
support of their employer than in the past. Less access to training may con‑
tribute to inequality, since when employers invest in their workers’ human
capital by paying for training or offering training on the job site, workers
also benefit in the form of future wage increases (Bartel 1992; Lynch 1991).
At the same time that wages and employment have been growing
among high-skill workers, employment in middle-skill jobs has declined,
especially relative to higher- and lower-skill jobs. Economists use the term
polarization to describe this pattern: employment loss in the middle of the
wage or job skill distribution combined with relative job growth at the bot‑
tom and at the top. The concept of polarization has its roots in the large
literature on skill-biased technological change that developed to try to
understand changes in wage inequality since the 1970s (Bound and Johnson
1995; Katz and Murphy 1992; Juhn, Murphy, and Pierce 1993). In the
past decade, economists have refined the skill-biased technological change
model, arguing that technology is a substitute for some, but not all, types of
labor. For example, Autor, Levy, and Murnane (2003) and Acemoglu and
Autor (2011) develop a model in which technology can replace labor in
tasks that are easily automated, such as manual labor, and in which highly

146 | Chapter 3
 Figure 3-27
Percent of Workers Receiving Employer-Sponsored
or On-the-Job Training, 1996–2008
Percent
20 19.4
Employer Paid for Training
On-the-Job Training
16.7

15
13.1
12.4
11.7
11.2

10 8.6 8.4

0
1996 2001 2004 2008
Note: Fraction of workers ages 18-65 receiving training of any duration in the last year.
Source: Census Bureau, Survey of Income and Program Participation (Employment and Training Topical
Module); CEA calculations.

skilled managerial professions are complementary to labor. The tasks that

are most easily automated tend to be in the middle of the skills distribution,
so that over time employment moves to both the lower and higher ends of
the occupational ranking, as shown in Figure 3-28, where occupations are
ranked by average wage.
Figure 3-29 uses smoothed data from employment by occupations
harmonized over a longer time period to show this pattern more clearly:
since the late 1970s, employment growth has been greatest in the highest
and lowest earning occupations. The middle of the distribution has actually
experienced employment losses, with fewer workers employed in middle-
wage occupations in 2012 than in 1979.
As demand falls for manual tasks, wages and employment in these
positions also fall relative to highly-skilled workers, leading to greater
inequality. The results from this research show that, in theory, automation
can lead to both job and wage polarization (Acemoglu and Autor 2011;
Goos, Manning, and Salomon 2007) and some have demonstrated a link
between changing tasks and other forms of wage inequality (Black and Spitz-
Oener 2010).
This stylized model, however, has not always matched the data.
Some economists argue that the automation hypothesis cannot explain the
timing of the trends in wage inequality and employment growth by real
wage level (Card and DiNardo 2002; Mishel, Shierholz, and Schmitt 2013).

Achievements and Challenges in the U.S. Labor Market | 147

 Figure 3-28
Change in Employment by Detailed Occupation, 1989–2014
Change in Total Employment, Thousands
7,000 Service Occupations,
Except Protective Executive,
6,000 and Household Technicians Administrative,
and Related and Managerial
5,000 Support Occupations
Teachers, Except Occupations
4,000 Postsecondary
Management
Related
3,000 Occupations
2,000
1,000
0
-1,000
Secretaries,
-2,000 Stenographers,
and Typists
-3,000 Machine Operators
$10 $14 $18 $22 $26 $30
Average Hourly Wage, 1989 (in 2014 Dollars)
Note: Excludes five occupational categories with outlying wages and relatively small changes in
employment (Farm Occupations, Except Managerial; Private Household Occupations; Engineers,
Architects, and Surveyors; Lawyers and Judges; and Health Diagnosing Occupations). Wages are calculated
using the method of Figure 3-25 and are adjusted for inflation using the CPI-U-RS.
Source: Bureau of Labor Statistics, Current Population Survey; CEA calculations.

Figure 3-29
Changes in Employment by Occupational
Wage Percentile, 1979–2012
Change in Employment Share, Percentage Points
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
-0.05
-0.10
-0.15
-0.20
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Percentile (Ranked by Occupational Mean Wage)
Source: Census Bureau, 1980 Census; Census Bureau, 2012 American Community Survey;
calculations by David Autor and Brendan Price.

148 | Chapter 3
In particular, Mishel, Shierholz, and Schmitt (2013) find that changes in
employment across occupations explains little of the rise in inequality in the
overall wage distribution in contrast to what would be expected if occupa‑
tions accurately reflect differences in tasks for which technology may have
shifted demand.22

Broader Measures of Job Quality

Broader measures of compensation take into account the value of
nonwage features of jobs. Sometimes these are benefits, like employer-
provided retirement plans, paid vacation days, and employer-sponsored
health insurance, but these can also be features like family-friendly schedul‑
ing practices and possibilities for advancement. Research has found that
trends in the combination of employer-provided benefits plus wages and
salary (called total compensation) broadly mirror those in wage compensa‑
tion — both have become substantially more unequal since the early 1980s,
though compensation inequality has generally grown more rapidly than
wage inequality (Pierce 2001, 2010).
Coverage of major employer benefits—specifically health insurance
and retirement plans—are tracked for long periods of time in surveys such
as the National Health Interview Survey and the Current Population Survey.
Changes in access to employer-sponsored health insurance and retirement
plans are shown separately in Figures 3-30 and 3-31. Access to these benefits
generally declined between 2000 and 2010, particularly for lower-skilled
workers. Recently, these trends have stabilized or begun to reverse: in 2013,
the share of employees with access to retirement plans increased, while
access to employer-sponsored health insurance held relatively steady from
2012.
Other important aspects of job quality are the number of hours a
worker is required to work, whether they are paid by salary, and whether
they are eligible for overtime pay for hours they work over 40 hours a week.
Figure 3-32 shows that since the mid-1990s, more full-time workers have
been earning salaries. Prior to the recession, the share of full-time workers
earning a salary was at or near its 1982 high. That share fell in the Great
Recession, as it did in the 1991 and 2001 recessions, but has recently started
to rise again. However, concern remains about the long hours of some sala‑
ried workers and whether they are properly compensated for those hours.
The value of the threshold at which salaried workers qualify for overtime pay
has eroded since it was last raised in 2004, and over this period the share of

22 Mishel, Shierholz and Schmitt (2013) show that occupations explain a small and decreasing
portion of the variance in wages.

Achievements and Challenges in the U.S. Labor Market | 149

 Figure 3-30
Share of Workers With an Offer of Employer-Sponsored
Insurance Coverage, by Education, 1997–2013
Percent
90 2013
College
Graduate
80

Some College
70

60 High School
Less than High Graduate
School Diploma
50

40

30
1997 1999 2001 2003 2005 2007 2009 2011 2013
Source: National Health Interview Survey; CEA calculations.

Figure 3-31
Share of Workers Included in Employer-Provided
Retirement Plan, by Education, 1997–2013
Percent
70

College 2013
60 Graduate

50

Some College
40

High School
30 Graduate
Less than High
School Diploma
20

10
1997 1999 2001 2003 2005 2007 2009 2011 2013
Source: Bureau of Labor Statistics, Current Population Survey (Annual Social and Economic
Supplement); CEA calculations.

150 | Chapter 3
 Figure 3-32
Share of Full-Time Workers Paid a Salary, 1979–2013
Percent of Full-Time Wage and Salary Workers
45

44

43

42 2013

41

40

39

38

37

36

35
1975 1980 1985 1990 1995 2000 2005 2010
Source: Bureau of Labor Statistics, Current Population Survey (Characteristics of Minimum Wage
Workers, 2013).

salaried workers afforded overtime protection has fallen from 45 percent to

39 percent.

The Agenda for a Stronger Labor Market

This chapter has documented strong progress in the labor market over
the past year. The headline unemployment rate is now 93 percent returned
to its 2001-07 average, and broader measures of labor underutilization
show a similar pattern. Despite this progress, however, the labor market
continues to face five related challenges. These challenges pre-date the Great
Recession, although a recovery may lessen these challenges going forward.
Nevertheless, policy is also needed to overcome the many obstacles to a
better functioning labor market. The challenges described in this chapter—
decreased labor force participation; more long-term unemployed workers;
more part-time workers, particularly among those who would like full-
time hours; lower labor market fluidity; and insufficient real wage growth
amidst a more polarized job market—are potentially all inter-connected.
For example, decreased labor force participation; longer unemployment
durations; and more people working, at least temporarily, in part-time jobs
when they want full-time jobs might all be related to decreased labor market
fluidity. If transitions among jobs, employers, and firms are less common,
it can take longer for people to find work, leading to longer unemployment

Achievements and Challenges in the U.S. Labor Market | 151

durations. In addition, some of those workers may accept part-time work,
at least temporarily, and some workers may stay out of the labor market
because they are less likely to be aware of potential opportunities or find the
longer searches needed too discouraging.
One key element of a successful strategy to address these challenges
is providing workers with skills that help raise job security, earnings, and
job quality—and a highly-trained workforce can also contribute to further
long-term growth. The President’s plans to improve access to education and
training from birth through college are at the forefront of this strategy. The
President’s Fiscal Year 2016 Budget shows this commitment through a range
of proposals, from funding for early learning initiatives, including ensuring
that all 4-year-olds have access to pre-school, to proposing that two years
of high-quality community college be free for hard-working students. In
addition, he has proposed expanding apprenticeships and improving our
workforce training systems by expanding career counseling and training in
high-growth fields.
To further help workers access jobs that match their skills and quali‑
fications, the President has also proposed working with states to spread best
practices for occupational licensing systems and to reduce unnecessary
training or high fees that keep people from doing jobs that best utilize their
talents. This builds on the leadership that First Lady Michelle Obama and
Dr. Jill Biden have undertaken to reduce licensing barriers for military
spouses, through which 48 states have eased licensing requirements for cur‑
rent military spouses and veterans.
A second key aspect of the President’s proposals to support and help
build the middle class are policies that help working families stay in the
labor force, by supporting flexible workplace practices, access to paid leave
and paid sick days, and greater access to high quality child care. In addition
to the work-family policies discussed in Chapter 4, the Administration’s
proposal for a new secondary earner credit recognizes the additional costs
that families with two earners face and therefore would help dual-earning
couples make ends meet.
Moreover, these policies are intricately linked to the President’s early
childhood education proposals since ensuring that children are well-cared
for also supports their parents while they are at work. To this end, the
Administration has proposed a continuum of early learning opportunities
that could support working parents while benefiting children’s cognitive
and socio-emotional development. These initiatives include tripling the
existing child care tax credit for families with very young children and
expanding access to high-quality early education, including child care and
preschool. These steps can help parents enter the labor market knowing

152 | Chapter 3
that their children are cared for in a safe and nurturing environment, while
also improving children’s academic performance and future outcomes in
adulthood.
Better skills and better employment supports are two key ingredients
for higher wages and higher incomes, but they are not sufficient. That is why
the President supports raising the minimum wage, a step that would help
tens of millions of workers and help ensure that no full-time worker raises
a family in poverty. Other institutional steps, like strengthening collective
bargaining, would further help ensure that everyone shares in the benefits
of growth.
Finally, the Administration continues to prioritize reducing long-term
unemployment. The President’s FY 2016 Budget proposes $16 billion for
High-Growth Sector training grants, disbursed across states based on their
unemployment rates, to double the number of dislocated workers who can
receive the training necessary to transition to high-quality jobs. By making
more funds available during economic downturns to provide training for
those who face difficulties finding work in weak labor markets, this proposal
could also reduce increases in long-term unemployment during future
downturns. The President has also engaged businesses in hiring and recruit‑
ing the long-term unemployed.
The President’s FY 2016 Budget also proposes a package of reforms
to modernize the Unemployment Insurance (UI) program, which provides
critical income support to those who are unemployed. These reforms will
improve the solvency of state programs, strengthen the program’s connec‑
tion to work, and reach more workers who lose a job through no fault of
their own. In addition, the proposal would make the UI program more tar‑
geted and responsive to economic downturns by implementing an Extended
Benefits program that provides added benefits as soon as a state experiences
a sharp rise in unemployment, even if a national increase in unemployment
has not yet occurred.
Taking steps to foster more growth and high-quality jobs, better pre‑
pare workers for these jobs, and ensure that all workers share in the benefits
of these jobs are the central tenets of the President’s approach to middle
class economics. All of these actions will help capitalize on the strengths of
the U.S. economy while moving to address the long-standing challenges in
the labor market.

Achievements and Challenges in the U.S. Labor Market | 153

 Box 3-4: Immigration Reform and Labor Markets
A large body of academic research finds that, on balance, immigra‑
tion has strong benefits for both the U.S. economy in general and U.S.
labor markets in particular. Immigrants increase the productivity of the
American workforce, both directly through increases in innovation and
indirectly through spillovers to U.S. workers. For example, not only do
high-skilled immigrants patent at a higher rate than their nonimmigrant
peers, but their innovation also has spillover effects on patenting by
native-born workers (Hunt and Gauthier-Loiselle 2010). At the same
time, lower-skilled immigration can have positive effects on worker
productivity by allowing for greater task specialization. While there is
ongoing discussion in the academic literature about the direct wage and
employment effects of immigration on native workers, it is important to
note that researchers have found positive effects of immigration on these
outcomes (for example, Peri, Shih, and Sparber 2014) as well as negative
(for example, Borjas et al. 1997). Nevertheless, a number of recent studies
suggest that complementarities between immigrant and nonimmigrant
workers—interactions that indirectly raise the productivity, and thus
wages, of both groups—may be substantial (e.g. Peri and Sparber 2009).
In addition to these benefits, immigration has the potential to raise the
overall labor force participation rate in the United States because immi‑
grants participate in the workforce at higher rates than the native-born
population (CBO 2015a). Researchers have shown that immigration is
associated with a range of characteristics that may be related to greater
labor force participation (Chiquiar and Hanson 2005; Butcher and Piehl
2007).
Despite these potential gains to the economy — and to American
workers — from immigration, the U.S. immigration system remains
badly broken. In November 2014, President Obama announced a series
of executive actions to begin moving our immigration system into
the 21st century. These provisions included actions designed to better
attract high-skilled immigrants and foreign entrepreneurs and to allow
advanced-degree holders in science, technology, engineering, and math‑
ematics (STEM) fields to extend on-the-job training. The actions will
also provide deferred action from removal to millions of undocumented
workers who have substantial ties to the United States, pass a criminal
background check, and pay payroll and income taxes. Drawing on a large
body of research examining the economic effects of previous immigra‑
tion reforms, the Council of Economic Advisers (2014a) estimated
that the actions announced by the President would raise U.S. GDP by
between 0.4 and 0.9 percent within ten years, equivalent to $90 to $210
billion in additional real GDP (in 2014 dollars) in 2024.

154 | 
 While these gains are substantial, they are small when com‑
pared with the potential economic effects of Congressional action on
commonsense immigration reform. The Congressional Budget Office
(2013) found that the Border Security, Economic Opportunity, and
Immigration Modernization Act (S. 744) – the bipartisan immigration
reform bill passed by the Senate in 2013 – would increase real GDP
by 3.3 percent, or roughly $700 billion, over ten years and would raise
average wages for all workers by 0.5 percent in twenty years. In addition,
CEA estimates that the Senate’s commonsense immigration reform bill
would raise the overall labor force participation rate by approximately
0.3 percentage point in ten years.

  | 155