The Scourge of Long

The Scourge of Long-Term
Unemployment
(La piaga della disoccupazione di lunga durata)
Alan B. Krueger
Bocconi-Boroli Lecture
May 29, 2014
Definition
scourge
skərj/
noun: scourge
1.1.
historical
a whip used as an instrument of punishment.
2.2.
a person or thing that causes great trouble or suffering.
“the scourge of mass unemployment”
Synonyms: affliction, bane, curse, plague, menace, evil, misfortune, burden, cross to
bear, blight, cancer
“inflation was the scourge of the mid-1970s”
Alan B. Krueger (Princeton University)
May 29, 2014
1
Motivation
•
•
•
•
Motivation: Short-term unemployment rate predicts price inflation, wage
growth and vacancies much better than long-term unemployment rate in U.S.
Hypothesis: Long-term unemployed are on the margins of the labor market,
exerting little pressure on wages and prices, because of both supply-side (e.g.,
discouragement, skill obsolescence) and demand-side (e.g., statistical
discrimination) factors.
Evidence:
 Long-term unemployment is devastating to individuals
 Employers discriminate against the long-term unemployed
 Long-term unemployed rarely return to consistent, full-time employment,
even in strong economy
 Reservation wages don’t fall over unemployment spell
 Calibrated model with labor force withdrawal returning to historical
average fits data well – i.e., the long-term unemployed are increasingly
likely to exit the labor force
 Even in states with relatively low unemployment rate, the long-term
unemployed are struggling
Policies should be tailored to the long-term unemployed.
Alan B. Krueger (Princeton University)
May 29, 2014
2
Outline
• Trends in U.S. Unemployment
• Three Puzzles: Price Inflation, Wage Inflation,
Vacancies
• Duration Dependence
• Characteristics of the Long-Term Unemployed in
U.S. and Italy
• Transition Rates: Short-term and Long-Term
• Regional U.S. Evidence
• Calibrated Model to Explore Role of Labor Force
Exits
• Conclusions
Alan B. Krueger (Princeton University)
May 29, 2014
3
A Record Share of the Unemployed are Long-Term in the
United States
Unemployed 27 Weeks & Over as Share of Total
Percent of Total Unemployed (Seasonally Adjusted)
50
Nov-13
45
40
35
30
25
20
15
10
5
0
Jan-48
Jan-58
Jan-68
Jan-78
Jan-88
Jan-98
Jan-08
Note: Shading denotes recession.
Source: Bureau of Labor Statistics (Current Population Survey); National Bureau of Economic Research.
Alan B. Krueger (Princeton University)
May 29, 2014
5
Long-Term Unemployment Jumped in Recession and is
Still Well Above Normal
Alan B. Krueger (Princeton University)
May 29, 2014
6
Price Phillips Curve in the United States Using Overall
and Short-Term Unemployment Rate
Alan B. Krueger (Princeton University)
May 29, 2014
7
Real Wage Phillips Curve in the United States Using
Overall and Short-Term Unemployment Rate
Alan B. Krueger (Princeton University)
May 29, 2014
8
Beveridge Curve in the United States Using Overall and
Short-Term Unemployment Rate
Alan B. Krueger (Princeton University)
May 29, 2014
9
Long-Term Unemployment Takes a Toll
• Mental Health and Self-Esteem (e.g., Clark and Oswald,
Krueger and Mueller, 2011)
• Also, unlike other stressful events, little psychological
adaptation to unemployment (e.g., Clark and Oswald).
• Physical Health and Mortality (e.g., Sullivan and von
Wachter, 2009)
• Social Isolation (e.g., Krueger and Mueller, 2012)
• Family Dissolution and Stress (e.g., Lindner and Peters,
2013)
• 15-30% Lower Re-employment Earnings (e.g., Farber)
• Scarring effect probably bigger in Europe, and bigger for
youth.
 Erosion of Human Capital and Discouragement
Alan B. Krueger (Princeton University)
May 29, 2014
10
Long-Term Unemployment Takes a Toll
Source: Pew Research Center, “The Impact of Long-Term Unemployment: Lost Income, Lost Friends—and Loss of Self-Respect”
(July 22, 2010).
Alan B. Krueger (Princeton University)
May 29, 2014
11
Reservation Wages Barely Fall Over the Spell of
Unemployment, Exerting Little Downward Wage Pressure
Alan B. Krueger (Princeton University)
May 29, 2014
12
Job Search Hours per Week
Job Search Intensity Declines as the Duration of
Unemployment Rises
Source: Wanberg, Connie R., Jing Zhu, Ruth Kanfer, and Zhen Zhang. 2013. “After the Pink Slip: Applying Dynamic Motivation Frameworks to
the Job Search Experience.” Organizational Psychology Review 3 no. 1: 3–40.
Alan B. Krueger (Princeton University)
May 29, 2014
13
The Probability of Receiving a Callback Declines as the
Duration of Unemployment Rises
Source: Kroft, Lange, and Notowidigdo, “Duration Dependence and Labor Market
Conditions: Evidence From a Field Experiment,” Quarterly Journal of Economics, 128(3):
1123-1167, August 2013.
Alan B. Krueger (Princeton University)
Source: Ghayad, “The Jobless Trap” (2013).
May 29, 2014
14
Re-Weight Unemployment Rate
(1) Use Search Intensity
(2) Use Callback Rate
Alan B. Krueger (Princeton University)
May 29, 2014
15
Price Phillips Curve in the United States After Adjusting
for Long-Term Unemployment
Alan B. Krueger (Princeton University)
May 29, 2014
16
Real Wage Phillips Curve in the United States After
Adjusting for Long-Term Unemployment
Alan B. Krueger (Princeton University)
May 29, 2014
17
Recent Changes in Prices and Wages Fall Within the
Confidence Intervals of Our Estimates
Alan B. Krueger (Princeton University)
May 29, 2014
18
Others Have Reached Similar Conclusion
Long-term Unemployed
Matter Less:
• Ricardo Llaudes (2005)
• James Stock (2011)
• Robert Gordon (2013)
• Mark Watson (2014)
• 2014 Economic Report of
the President
• Rand Ghayad and
William Dickens (2012)
• Pat Higgins (Atlanta Fed)
Alan B. Krueger (Princeton University)
Alternatives:
• Ben Bernanke: inflation
expectations anchored
• Laurence Ball and
Sandeep Mazumder:
distribution of price
changes
• Fed minutes: Nominal
wage rigidities
May 29, 2014
19
The Severity of Long-Term Unemployment Differs
Across Countries and Jumped in U.S.
Alan B. Krueger (Princeton University)
May 29, 2014
20
Characteristics of Employed and Short- and
Long-Term Unemployed
•
•
•
•
•
•
•
•
By Age
By Gender and Marital Status
By Race and Ethnicity
By Education
By Industry
By Occupation
By Region
By Urban/Rural
Alan B. Krueger (Princeton University)
May 29, 2014
21
Age Distribution of the Employed and Unemployed, U.S.A.
Percent
70
Pct. of Employed
Pct. of Short-Term Unemployed
60
Pct. of Long-Term Unemployed
57%
50
40
30
40%
34%
33%
33%
31%
29%
20
23%
19%
10
0
16-34
Employed
35-49
Short-Term Unemp. (0-14 Wks)
50+
Long-Term Unemp. (27+ Wks)
Note: Bars represent the percent of the total population of employed or unemployed that is in each age
group. Source: 2012 Current Population Survey.
Alan B. Krueger (Princeton University)
May 29, 2014
22
Age Distribution of the Employed and Unemployed, Italy
Italian Long-Term Unemployed are Younger
Percent
70
Pct. of Short-Term Unemployed
60
Pct. of Long-Term Unemployed
50
52%
49%
47%
40
Pct. of Employed
35%
36%
30
29%
20
24%
13%
10
15%
0
15-34
35-49
50+
Source: 2012 ISTA RCFL (Labor Force Survey) courtesy of Tito Boeri.
Alan B. Krueger (Princeton University)
May 29, 2014
23
Youth UR vs. Adult UR, 2006
Ratio of 2.6 to 1
70%
60%
Youth UR
50%
40%
30%
Greece
Italy
20%
Spain
U.S.
10%
Turkey
0%
0%
5%
10%
15%
20%
25%
Adult UR
Source: OECD data.
Alan B. Krueger (Princeton University)
May 29, 2014
24
Youth UR vs. Adult UR, 2006 and 2012
Ratio of 2.6 to 1
70%
60%
Greece
Youth UR
50%
Spain
40%
Italy
30%
2012
Greece
2006
Italy
20%
US
US
10%
Turkey
Turkey
Spain
0%
0%
5%
10%
15%
20%
25%
Adult UR
Source: OECD data.
Alan B. Krueger (Princeton University)
May 29, 2014
25
Gender and Marital Status of the Employed and
Unemployed in the United States
Distribution of Each Employment Status Category
by Gender
Percent
Distribution of Each Employment Status Category
by Marital Status
Percent
60
Employed
Short-Term Unemp. (0-14 Wks)
60
Long-Term Unemp. (27+ Wks)
50
50
40
40
30
30
53%
54%
Employed
Short-Term Unemp. (0-14 Wks)
Long-Term Unemp. (27+ Wks)
56%
52%
55%
47%
46%
20
44%
20
45%
33%
10
37%
29%
15%
10
15%
19%
0
0
Married
Male
Female
Note: Bars represent the share of each employment status category that is of each gender.
Source: Estimates from the 2012 Current Population Survey.
Alan B. Krueger (Princeton University)
Widowed/Divorced/
Separated
Never Married
Note: Bars represent the share of each employment status category that is of each marital
status.
Source: Estimates from the 2012 Current Population Survey.
May 29, 2014
26
Race/Ethnicity of the Employed and Unemployed in the
United States (2012)
Distribution of Each Employment Status Category
by Race/Ethnicity
Percent
80
70
60
67%
Employed
Short-Term Unemp. (0-14 Wks)
Long-Term Unemp. (27+ Wks)
55%
51%
50
40
30
22%
20
16%
22%
19%
15%
10%
10
6%
4%
5%
2%
3%
3%
0
White
African
American
Hispanic
Asian/Pacific
Islander
Other
Note: Bars represent the share of each employment status category that is of each rate or ethnicity. The Hispanic category includes Hispanic
individuals of all racial groups. All other categories only include non-Hispanics.
Source: Estimates from the 2012 Current Population Survey.
Alan B. Krueger (Princeton University)
May 29, 2014
27
Educational Attainment of the Employed and
Unemployed in the United States (2013)
Distribution of Each Employment Status Category by
Educational Attainment (US)
Employed
Short-Term Unemp. (0-14 Wks)
40%
36%
34%
33%
35%
30%
27%
25%
22%
19%
18%
20%
21%
19%
15%
10%
Long-Term Unemp. (27+ Wks)
17%
11%
9%
8%
19%
9%
5%
0%
Less than High
School
High School
Some College
Associate Degree
Bachelor Degree or
Higher
Note: Bars represent the share of each employment status category that has attained precisely each level of education.
Source: Estimates from the 2012 Current Population Survey.
Alan B. Krueger (Princeton University)
May 29, 2014
28
Education Matters in Italy too, but Italian Lower
Secondary Looks Like U.S. High School Graduates
Alan B. Krueger (Princeton University)
May 29, 2014
29
Employment and Unemployment are Mostly Evenly
Spread Throughout the United States
Distribution of Each Employment Status Category
by Region
Percent
25
Employed
Short-Term Unemp. (0-14 Wks)
Long-Term Unemp. (27+ Wks)
20
22%
21%
19%
19%
18%
16%
15% 15%16%15%
15
13%
13%
12%
11%
10
8%
7%
6%
5%
5
0
4% 5%
New
England
6% 6%
7%
8%
6%
5%
4%
Middle East North
Atlantic Central
West
North
Central
South
Atlantic
East
South
Central
West
South
Central
Mountain
Pacific
Note: Bars represent the share of each employment status category that lives in each region of the United States.
Source: Estimates from the 2012 Current Population Survey.
Alan B. Krueger (Princeton University)
May 29, 2014
30
Italian Workers are in the North and Unemployed are in
the South and Island
Distribution of Each Employment Status Category by Area
Employed
Short-Term Unemp. (< 3 Months)
Long-Term Unemp. (> 6 Months)
60%
52%
50%
50%
42%
40%
37%
33%
30%
27%
21%
21%
20%
17%
10%
0%
North
Alan B. Krueger (Princeton University)
Center
May 29, 2014
South and Islands
31
Industry Distribution of the Employed and Unemployed
in the United States
Distribution of Each Employment Status Category by Industry
Employed
Short-Term Unemp. (0-14 Wks)
Long-Term Unemp. (27+ Wks)
25%
23%
19%
20%
16%
14% 14%
15%
12%
10%
10%
10%
12%
11%
16% 16%
12%
9%
9%
7%
6%
4%
5%
0%
14% 14%
16%
15% 15%
Construction
Manufacturing
Wholesale and
Retail Trade
5%
Finance and Real Prof. and Business
Estate
Services
Education and
Health Care
Leisure and
Hospitality
All Other
Note: Bars represent the share of each employment status category that either is or was most recently employed in each industry.
Source: Estimates from the 2012 Current Population Survey.
Alan B. Krueger (Princeton University)
May 29, 2014
32
Industry Distribution of the Employed and Unemployed
in Italy
Distribution of Each Employment Status Category by Industry
Employed
Short-Term Unemp. (< 3 Months)
Long-Term Unemp. (> 6 Months)
25%
20%
20%
19%
19%
16%
14%
15%
15%
12%
11%
10%
10%
5%
16%
15%15%
8%
7%
5%
3%
9%
3%
5%
6%
11%
10%
10%
8%
6%
4%
2% 2%
4%
3%
1%
6%
1% 1%
2% 2%
0%
Alan B. Krueger (Princeton University)
May 29, 2014
33
Unemployed Workers Typically Return to Their Previous
Occupations and Industries
Occupations of Workers Who Were Employed or Found Work in 2012
(Percent of Total in Each Category of Employed Workers)
Employees Who Had
Employees Who Had
Previously Been
Previously Been
Unemployed
Unemployed
Short-Term But
Long-Term But
Total
Found Work
Found Work
Occupation
Employed
Old Work New Work
Old Work New Work
Blue-Collar Occupations
17.5 %
33.4 %
31.9 %
26.8 %
25.6 %
Sales & Services
32.2
37.1
38.2
37.4
40.1
Administrative
12.4
9.1
9.5
12.9
12.9
Professional & Technical
22.0
14.7
14.8
12.1
12.8
Managerial & Financial
15.9
5.7
5.6
10.8
8.5
• Unemployed workers who found a new job in 2012
typically did not change occupations.
 Among the short-term unemployed, 33 percent were
blue-collar workers before losing their jobs, about the
same fraction as those who found a new job.
 Among the long-term unemployed who found a job, 13
percent worked in administrative occupations, roughly
the same fraction as before losing their jobs.
Note: Short-term unemployed defined as 26 weeks or fewer of unemployment. Long-term unemployed defined
as 27 weeks or more of unemployment.
Source: Estimates from the 2012 Current Population Survey.
Industries of Workers Who Were Employed or Found Work in 2012
(Percent of Total in Each Category of Employed Workers)
Employees Who Had
Employees Who Had
Previously Been
Previously Been
Unemployed
Unemployed
Short-Term But
Long-Term But
Total
Found Work
Found Work
Industry
Employed
Old Work New Work
Old Work New Work
Construction
6.3 %
14.3 %
13.6 %
10.9 %
11.5 %
Manufacturing
10.3
8.1
7.7
9.5
6.5
Wholesale & Retail Trade
14.0
13.2
12.9
15.0
15.7
Financial Activities
6.7
3.1
2.9
5.4
3.7
Professional & Business Services
11.6
13.2
13.4
15.5
16.9
Educational Services
9.1
9.1
8.6
5.2
6.0
Health Care Services
13.6
8.2
8.7
10.6
10.4
Leisure & Hospitality
9.3
14.6
16.2
12.2
12.6
All Other
19.1
16.3
16.1
15.7
16.6
Note: Short-term unemployed defined as 26 weeks or fewer of unemployment. Long-term unemployed defined
as 27 weeks or more of unemployment.
Source: Estimates from the 2012 Current Population Survey.
Alan B. Krueger (Princeton University)
May 29, 2014
• Unemployed workers who found a new job in 2012 also
did not tend to switch industries.
 Among the short-term unemployed, 14.3 percent
worked in construction before losing their jobs, while
13.6 percent returned to the construction industry
upon finding a new job.
 One exception, however, was for the manufacturing
industry. A somewhat smaller fraction of the longterm unemployed who found a new job returned to
manufacturing.
• These patterns provide further support for the view that a
persistent shortfall in aggregate demand—rather than job
mismatch—is the main reason why unemployment
remains elevated. They also indicate the challenge in
helping workers transition to new sectors.
34
Average Monthly Job Finding Rate
Alan B. Krueger (Princeton University)
May 29, 2014
35
Little Effect of Composition Changes on Job Finding
Rate Over Business Cycle
Alan B. Krueger (Princeton University)
May 29, 2014
36
Average Monthly Labor Force Withdrawal Rate in the
United States
Alan B. Krueger (Princeton University)
May 29, 2014
37
Small Effect of Composition Changes on Labor Force
Withdrawal Rate Over Business Cycle
Alan B. Krueger (Princeton University)
May 29, 2014
38
How Do the Long-Term Unemployed Fare a 15 Months
Later?
Alan B. Krueger (Princeton University)
May 29, 2014
39
Even in Good Times, the Long-Term Unemployed are Unlikely to
Hold Steady, Full-Time Jobs a Year Later
Short-Term
Long-Term
Alan B. Krueger (Princeton University)
May 29, 2014
40
The Long-Term Unemployed Generally Have High Labor Force
Exit Rates After 15 Months
Long-Term
Short-Term
Alan B. Krueger (Princeton University)
May 29, 2014
41
Alan B. Krueger (Princeton University)
May 29, 2014
42
Alan B. Krueger (Princeton University)
May 29, 2014
43
Regional Analysis
Alan B. Krueger (Princeton University)
May 29, 2014
44
Even in Low Unemployment States, Long-Term
Unemployment Remains Elevated
Alan B. Krueger (Princeton University)
May 29, 2014
45
Transitions to Employment for Long-Term Unemployed
Are Similar in High and Low Unemployment States
Alan B. Krueger (Princeton University)
May 29, 2014
46
Labor Force Exits are Slightly Higher for Long-Term
Unemployed in Low Unemployment States
Alan B. Krueger (Princeton University)
May 29, 2014
47
Calibration Model
• Allow differential effect (and matching
efficiency) of short-term and long-term
unemployed, and NILF.
• Focus on prime-age workers to avoid issues
of demographic changes and increased
schooling, following Kroft et al., 2013
• Fit matching function from 2002-2007, using
flow rates from UL-E, US-E, and N-E.
Alan B. Krueger (Princeton University)
May 29, 2014
48
Calibration Model, Ctd.
• Using these matching function parameters
simulate forward from January 2008.
• Forcing variables are actual job finding rates,
layoff rates and labor force exit rates, and
Vacancies.
• Investigate whether matching function with
lower weight on the long-term unemployed
and differential labor force withdrawal
predict loops around the Beveridge Curve.
Alan B. Krueger (Princeton University)
May 29, 2014
49
Beveridge Curve
Vacancy Rate
(As a share of 25-54 year old Labor Force)
6%
5%
Dec. 2013
4%
3%
2%
Actual Data
1%
0%
0%
1%
2%
3%
4%
5%
6%
7%
8%
9%
10%
Unemployment Rate for 25-54 year olds
Alan B. Krueger (Princeton University)
May 29, 2014
50
Actual and Calibrated Beveridge Curve
Dec. 2013
RMSE = 0.8
Alan B. Krueger (Princeton University)
May 29, 2014
51
Predicted Sharp Run-Up in Long-term Unemployment
and Gradual Future Decline
Alan B. Krueger (Princeton University)
May 29, 2014
52
Counterfactual Calibration Exercise Imposing 2006-07
Labor Force Withdrawal Rates
• Used calibrated model to investigate the effect on
unemployment of labor force exits by duration of
unemployment. Labor force exits fell for the long-term
unemployed during the Great Recession and in the ensuing few
years, before rising toward their historical average. What
impact did this pattern have on the unemployment rate and
shift in the Beveridge Curve?
• Imposed the labor-force withdrawal rates for the long-term and
short-term unemployed from 2006-07, just before the
recession, each year going forward.
• The unemployment rate would have been under predicted by 3
percentage points under this scenario. The root mean squared
error in this counterfactual model rises to 1.6 percentage point,
substantially worse than the fit when the actual path of labor
force participation by duration of unemployment is used in the
model. Smaller and shorter lived loop around Beveridge Curve.
Alan B. Krueger (Princeton University)
May 29, 2014
53
Counterfactual Calibration Exercise based on Oct. 2009
Labor Force Withdrawal Rates
• To explore the role of the gradual increase in labor force
withdrawal by the unemployed since the unemployment rate
peaked October 2009, conducted a counterfactual exercise in
which maintained the labor force withdrawal rates at their
October 2009 levels and projected the unemployment rate
through December 2013 using the estimated matching
function and all of the other realized flow data.
• Compared to the calibrated model the unemployment rate was
1.3 percentage point higher in this counterfactual projection.
Thus, the increase in labor force withdrawal rates from
October 2009 through December 2013 appears to account for
a little over one percentage point of the 3.2 percentage point
drop in the unemployment rate for prime age workers.
Alan B. Krueger (Princeton University)
May 29, 2014
54
Simulation: Dec. 2013 Forward Assuming U to N
Returns to Pre-Recession Average
Alan B. Krueger (Princeton University)
May 29, 2014
55
Conclusions
• The Great Recession and Moderate Recovery were exceptional
because of the high-rate of long-term unemployment, which resulted
from collapse of vacancies and labor force withdrawal rates
• Even in good times, the long-term unemployed have great difficulty
finding steady, full-time employment
• Job finding rates for the long-term unemployed are less responsive to
the strength of the economy than they are for the short-term
unemployed
• The long-term unemployed are increasingly dropping out of the labor
force
• Suggestive evidence that the long-term unemployed exert little
pressure on the economy
• Would expect pick up in inflation and real wages as short-term
unemployment falls, but confidence band is wide
• Key Policy Challenge: What interventions can keep the long-term
unemployed in the labor force and raise their prospects of finding an
acceptable job? Stronger macro-economy probably necessary but not
sufficient.
Alan B. Krueger (Princeton University)
May 29, 2014
56