View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

ST. LOUIS

2018:Q2 | VOL. 26 | NO. 2

FEDERAL RESERVE BANK

Insights on economic issues in today’s headlines

Hispanics and Their
Contribution to
America’s Human Capital
President Bullard

A Different Gender Gap

Millennial Finance

Reflections on key monetary
policy themes during his
first decade as president

Women now appear
to be at less risk of
unemployment than men

America’s youngest
working generation
lags behind

PAGE 3

PAGE 12

PAGE 14

IN THIS ISSUE

2018:Q2 | VOL. 26, NO. 2
The Regional Economist is published
quarterly by the Research and
Public Affairs divisions of the Federal
Reserve Bank of St. Louis. It addresses
the national, international and
regional economic issues of the day,
particularly as they apply to states in
the Eighth Federal Reserve District.
Views expressed are not necessarily
those of the St. Louis Fed or of the
Federal Reserve System.
Director of Research
Christopher J. Waller
Senior Policy Adviser
Cletus C. Coughlin
Deputy Director of Research
David C. Wheelock
Director of Public Affairs
Karen Branding
Editor
Subhayu Bandyopadhyay
Managing Editor
Gregory Cancelada
Art Director
Joni Williams
Please direct your comments
to Subhayu Bandyopadhyay
at 314-444-7425 or by email at
subhayu.bandyopadhyay@stls.
frb.org. You can also write to him
at the address below. Submission
of a letter to the editor gives us
the right to post it to our website
and/or publish it in the Regional
Economist unless the writer states
otherwise. We reserve the right to
edit letters for clarity and length.
To subscribe to the RE, go to
www.stlouisfed.org/subscribe/
regional-economist.
Regional Economist
Public Affairs Office
P.O. Box 442
St. Louis, MO 63166-0442

4

ST. LOUIS

2018:Q2 | VOL. 26 | NO. 2

REGIONAL
ECONOMIST

FEDERAL RESERVE BANK

Insights on economic issues in today’s headlines

Hispanics and Their Contribution
to America’s Human Capital
Despite making gains in professional and
other skilled occupations, Hispanics are more
likely to end up in lower-skill jobs. This disparity may be due to Hispanics lagging behind
their non-Hispanic peers in education.

Hispanics and Their
Contribution to
America’s Human Capital
President Bullard

A Different Gender Gap

Millennial Finance

The lessons learned
during his first 10 years
leading the St. Louis Fed

Women now appear
to be at less risk of
joblessness than men

America’s youngest
working generation
lags behind

PAGE 3

PAGE 12

PAGE 14

PRESIDENT’S MESSAGE ............................................................................................................. 3

Housing Costs and Regional Income Inequality
When accounting for housing costs, how does regional income inequality fare
in China versus the U.S.? ................................................................................................10
Unequal Pink Slips? Gender and the Risk of Unemployment
Women now appear to be less exposed to increased joblessness during recessions
than men. .....................................................................................................................................12
Accounting for Age: The Financial Health of Millennials
Millennials’ lagging net worth may be due to longer schooling and delays
in marriage and other life events. ................................................................................... 14
St. Louis Fed Steps in to Provide More-Timely Jobs Data
Our estimates on regional job growth can alert local policymakers of likely
revisions well ahead of time. ........................................................................................... 16
DISTRICT OVERVIEW

Gauging Debt Levels in the U.S. and Eighth District
Consumer indebtedness has exceeded a 2008 peak, but the situation isn’t bleak. ..... 19
NATIONAL OVERVIEW

U.S. Economic Growth Appears Solid This Year
Forecasters expect the economy to sustain above-trend growth for the
remainder of 2018. .......................................................................................................... 22
ECONOMY AT A GLANCE................................................................................................ 23

COVER IMAGE:
©THINKSTOCK/iSTOCK/monkeybusinessimages

ONLINE EXTRA
CEO Succession at Rural Banks

In recent years, new CEOs at rural banks have gotten younger.
2 REGIONAL ECONOMIST | Second Quarter 2018

Read more at www.stlouisfed.org/publications/re.

PRESIDENT’S MESSAGE

Reflections on Crisis to Recovery

M

y 10-year anniversary as president
and CEO of the Federal Reserve
Bank of St. Louis seemed like an appropriate time to reflect on the lessons learned
over this period, which has been anything
but ordinary.1
By the time I stepped into this role in
April 2008, the financial crisis was already
underway. The Federal Open Market
Committee (FOMC) reacted by lowering
the policy rate (i.e., the federal funds rate
target) several times in late 2007 and early
2008. In March 2008, the rescue of Bear
Stearns showed that the crisis had entered
a new—and a more difficult—phase.
During the summer of 2008, there
was still a case to be made that the U.S.
economy would muddle through the crisis.
However, the doubling of oil prices since
the summer of 2007 contributed to slower
economic growth during the second half
of 2008. With the collapse of Lehman
Brothers and AIG (American International
Group) in September 2008, the crisis was in
full swing.
In response, the FOMC lowered the
policy rate further, hitting the so-called
zero lower bound in December 2008. The
rate remained near zero for seven years. In
my view, the most important element of
this whole era has been encountering the
zero lower bound and then trying to decide
what to do, if anything, given that interest rates could not be reduced further in
response to poor economic conditions.
The crisis ultimately changed the nature
of how we think about central banking and
how a central bank should conduct monetary policy at the zero lower bound.
Against this backdrop, some of the key
themes and policy positions over my first
10 years as St. Louis Fed president are
briefly discussed below:
1) The limits of fiscal policy: Once the
policy rate hit the zero lower bound,
calls for fiscal approaches to stabilization policy gained popularity. However,
the FOMC was not out of ammunition; it turned to quantitative easing
(QE) and forward guidance. In a 2012
paper, I argued that stabilization policy
should be viewed the same way after the

crisis as it was before—monetary policy
should still be used to respond to shortterm fluctuations in the economy.
2) Fear of a deflationary trap: Many
inflation measures were low and declining in 2010. During that summer, I
released a paper that concluded the best
course of action for turning inflation
around—thus avoiding a Japanese-style
deflation—was to implement QE. The
FOMC implemented its QE2 program in
November of that year.
3) QE3—data-driven, not date-driven:
As early as 2009, I had advocated for
balance sheet policy to be state-contingent and adjusted depending on economic conditions.2 While QE1 and QE2
were associated with fixed end dates,
the FOMC’s QE3 program was openended—a form of state contingency. The
end of QE3 depended on certain labor
market conditions being met.
4) A preferred approach to normalization: Monetary policy normalization
began in December 2015 with “liftoff”
of the policy rate. The FOMC chose to
raise the policy rate before starting to
shrink the balance sheet, but I favored
the opposite sequence—a last-in, firstout approach. Choosing liftoff first has
forced the FOMC to raise the policy rate
in a world of superabundant reserves,
causing the Fed to adopt new operating
procedures for raising interest rates.
5) A regime-based view of the economy:
At the St. Louis Fed, we changed our
approach to near-term forecasts of the
macroeconomy and monetary policy in
June 2016. We now assume the macroeconomy could switch between regimes
(or steady states) and, therefore, could
have a set of possible long-run outcomes.
Projections for monetary policy are
calibrated for the current regime.
6) A push for more transparency:
Improving Fed communications became
a central focus of the FOMC during and
after the financial crisis. Still, more can
be done. One such improvement, which
Fed Chairman Jay Powell announced
this month, is a press conference after
every FOMC meeting3 rather than the

current practice of after every other
meeting. Another would be to replace
the FOMC’s Summary of Economic
Projections with a quarterly monetary
policy report that better explains the
FOMC’s actions and projections on a
regular basis.
7) The road to an inflation target: The
Fed lagged many other central banks in
adopting an explicit inflation target. In
early 2011, an ad hoc group of Federal
Reserve bank presidents (five of us)
drafted a one-page statement that not
only would name an inflation target
for the U.S. but would touch on other
important issues. This proposed statement was similar to the one the FOMC
adopted in January 2012.
8) Alternatives to inflation targeting:
Central banking around the world has
been primarily focused on inflation targeting as a way to keep inflation low and
stable. Alternative approaches such as
price-level targeting and nominal GDP
targeting could be an improvement on
inflation targeting and may be a wave of
the future in central banking.

James Bullard, President and CEO
Federal Reserve Bank of St. Louis

ENDNOTES
1

2
3

This column is based on the St. Louis Fed’s latest
annual report. See www.stlouisfed.org/annualreport/2017.
State-contingent policy means reacting to economic
events and not doing things according to the calendar.
The new approach will begin in January 2019.
REGIONAL ECONOMIST | www.stlouisfed.org 3

Hispanics and Their
Contribution to America’s
Human Capital
By Alexander Monge-Naranjo and Juan Ignacio Vizcaino

KEY TAKEAWAYS
• Hispanics are a growing share of the U.S.
workforce. Yet they are more likely to
work in lower-skill occupations than
non-Hispanics.
• Data show that the education level of
Hispanic workers lags behind that of
non-Hispanic workers. This may explain
the disparity in occupations.
• The country’s aggregate productivity
would improve if Hispanics could
develop their talent and skills.

I

©THINKSTOCK/ISTOCK/ Todd Warnock

4 REGIONAL ECONOMIST | Second Quarter 2018

mmigrants and native-born Americans. Farmworkers and professionals.
High school graduates and doctors. With
backgrounds from Mexico to Chile.
Much of this diversity is often ignored
in the frequent discussion of the everrising weight of Hispanics in both the
U.S. population and labor force. But the
composition of the Hispanic population
and labor force has been changing, in
some areas dramatically, and a deeper
understanding of these changes is needed
to assess the contribution of Hispanics to
the country’s overall human capital.
In this article, we explore the transformation in the human capital of Hispanics
and how these shifts have impacted their
occupations and integration into the
American workplace. We describe not
only the substantial increase in the numbers but also the significant diversity and
assimilation of Hispanic workers in the
U.S., how they compare with their peers
in terms of education, and their participation in different occupations. We also

Figure 1

Hispanics have sharply increased their presence in all occupations,
but they have an outsized share in lower-skill jobs

Professional and Technical Workers

0.8

Managers, Officials and Proprietors

0.9

1950

9.6

1.3

Clerical and Kindred

1.2

Craftsmen

1.2

11.3
14.5
16.8

Service Workers

2.0

Operatives

2.1

18.8
18.4

1.8

Unskilled Laborers

2016

8.0

Sales Workers

Farmers and Farm Laborers

Hispanics represented 13.4 percent
of the U.S. workforce in 2016.

LOWER SKILL HIGHER SKILL

OCCUPATIONS

Hispanics represented 1.6 percent
of the U.S. workforce in 1950.

24.8
3.2

25.0

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
SOURCES: IPUMS USA and authors’ calculations.

put emphasis on the presence of Hispanics
in the higher-earning occupations and
describe the increased role of Hispanic
women in those occupations.
Our review of the data helps us elucidate
and discuss some of the key challenges
faced by the Hispanic population—if they
are to fully assimilate and catch up with
their peers in the U.S. labor force.
The Growing Hispanic Presence
We collected individual level-data on
the age, gender, race, education level and
current occupation of workers from a
data set, IPUMS USA.1 For simplicity, we
grouped workers into two bins: Hispanics
and non-Hispanics, according to their
self-reported characterization. From these
data, we found that the percentage of
U.S. residents who identify themselves as

PERCENTAGE OF WORKERS WHO ARE HISPANIC

Hispanic or Latino have grown dramatically—more than eight times—within the
last seven decades. Hispanics represented a
mere 1.9 percent of the population in 1950,
compared to almost 16 percent in 2016.
The fastest growth is between 1970 and
2000, when the percentage more than
tripled—from 4 percent in 1970 to 12.3
percent in 2000. After 2000, the growth has
remained substantial, but has been much
slower. Very similar numbers hold when
we restrict attention to the working-age
population. As with the overall population,
the percentage of Hispanic workers grew
by a factor greater than eight, from barely
being just 1.6 percent (1 in 63) in 1950 to
being 13.4 percent (1 in 7.5) in 2016.
The IPUMS USA database can be
used to classify workers according to
11 educational attainment categories.

ABOUT THE AUTHORS
Alexander Monge-Naranjo (left) is an economist and research officer at the Federal Reserve
Bank of St. Louis. His research interests include growth and development, labor and applied
contract theory. He joined the St. Louis Fed in 2012. Read more about the author and his
research at https://research.stlouisfed.org/econ/monge-naranjo.
At the time this was written, Juan Ignacio Vizcaino was a technical research associate at the
Federal Reserve Bank of St. Louis. He is currently a Ph.D. student in economics at
Washington University in St. Louis.

REGIONAL ECONOMIST | www.stlouisfed.org 5

Figure 2

Participation of Hispanics across Occupations
30

Percent

25
20

Professional and Technical
All Occupations
Unskilled Laborers

15
10
5
0
1950

1960

1970

1980

1990

2000

2010

2016

SOURCES: IPUMS USA and authors’ calculations.

Figure 3

Hispanic Workers: Relative Wages and Labor Share
90

Percent

As of 2016, about 1 in 20 Hispanic
female workers were professionals
compared to only 1 in 30 for the males.

14

85

12
10

80

8
75

6
4

70
65
1950

Percent

©THINKSTOCK/iSTOCK/monkeybusinessimages

16

2
0
1960

1970

1980

1990

2000

2010

2016

Average Hispanic Wage/Average Non-Hispanic Wage (Left Axis)
% of Hispanics in the Labor Force (Right Axis)
SOURCES: IPUMS USA and authors’ calculations.

For tractability, we grouped these categories into five broader groups: primary
school or less (nursery school through
eighth grade), secondary incomplete
(ninth to 11th grade), secondary complete
(12th grade), college incomplete (one to
three years of college), and college complete or more (four or more years of higher
education).
The data set also allows us to group
workers according to broad occupational
groups. Specifically, IPUMS USA uses the
1950 Census Bureau occupational classification, aggregating three-digit occupations into the following nine broad groups,
ordered by their skill intensity:2 professional and technical workers; managers,
officials and proprietors; sales workers;
clerical and kindred; craftsmen; service
6 REGIONAL ECONOMIST | Second Quarter 2018

workers; operatives; farmers and farm
laborers; and unskilled laborers.3
Hispanic Workers across Occupations
We start by exploring the changes in
what the Hispanic-American workers
do in the marketplace. First of all, the
presence of Hispanic workers has grown
in all occupations. For each of the nine
occupations available in IPUMS USA,
Figure 1 (see page 5) shows the percentage of workers who identify themselves as
Hispanic. In all the categories, the growth
has been substantial—in some cases by
much more than a tenfold growth in the
participation of Hispanics—including in
the highest paid occupations, i.e., professionals and managers, which we discuss
further below.

The Education Effect
A number of obvious questions arise.
The first one is the following: How does
much of the growth directed toward
lower-skill occupations translate into
wages? It turns out quite a bit! In Figure 3,
we display the behavior of the average
wages of Hispanic workers relative to
the wages of non-Hispanics. The figure
clearly shows that the average wages of
Hispanic workers have been substantially lower than those of non-Hispanic
workers during the entire sample
period. More interestingly, the figure
also shows that this ratio has fallen
substantially with the rise in the share
of Hispanic workers, most notably from

Figure 4

Educational Attainment of Hispanic and Non-Hispanic Workers
70

Hispanics 1950

60

Non-Hispanics 1950

50
Percent

A second key finding that is evident from Figure 1 is that the growth
in the Hispanic participation across
occupations has been far from uniform. Instead, it has been substantially
inclined toward lower-skill jobs. Indeed,
as of 2016, the participation of Hispanics in occupations such as service
workers, operatives and, most strongly,
farmworkers and unskilled laborers is
much higher than their weight in the
population and labor force. By contrast,
the participation of Hispanics in professional occupations and managerial
occupations and, to a lesser extent, sales
occupations is substantially lower than
in the aggregate of all occupations.
Figure 2 further illustrates these
asymmetries, displaying the fraction of
Hispanics in the occupations ranked
top and bottom in terms of skills
(i.e., professional and technical, and
unskilled laborers), along with the overall Hispanic presence in the labor force.
The figure shows that these disparities not only are non-negligible but also
have been growing over time. Accelerating first in 1970, when the Hispanic
presence was growing the fastest, these
disparities have increased even faster
since 1990, when the Hispanic population growth was starting to slow down.
The end result of these asymmetric
growth rates is that, for 2016, the last
year in our sample, Hispanics were only
8 percent of professional workers, who
are employed in highly paid occupations,
while they accounted for 25 percent of
unskilled laborers, which are occupations with much lower earnings.

Hispanics 2016

40

Non-Hispanics 2016

30
20
10
0

Primary
or Less

Secondary
Incomplete

Secondary
Complete

College
Incomplete

College Complete
or More

SOURCES: IPUMS USA and authors’ calculations.
NOTE: Educational attainment is defined by the following groups: primary school or less (nursery school
through eighth grade), secondary incomplete (ninth to 11th grade), secondary complete (12th grade),
college incomplete (one to three years of college), and college complete or more (four or more years of
higher education).

1970 to 2000, precisely when the population of Hispanics grew the fastest in
the U.S. For the later years, the ratio has
settled to around two-thirds of the average wage for non-Hispanic workers.
The next obvious question is: What
explains these large gaps? Education and
skill formation in general is the No. 1
candidate for the driver. Indeed, Figure 4
shows the distribution of workers across
levels of educational attainment for U.S.
Hispanic workers (red) and non-Hispanic
workers (black) for both 1950 and 2016.
Each bar represents the percentage of the
workers, Hispanics and non-Hispanics, in
each education group.
We must first recognize the substantial
progress in the educational attainment of
both groups of workers. The most dramatic improvement can be seen in the
drastic reduction of the population with
primary education or less: While 63.5
percent of the Hispanic workers in 1950
had primary education as their highest
level attained, this percentage plunged to
only 16.4 percent in 2016. On the other
extreme, the fraction of Hispanic workers
with some college moved from 3.6 percent
to 16.2 percent, while those who have at
least completed college surged from 1.8
percent to 8.1 percent.
While substantial, the improvements
in the schooling attainment of Hispanic
workers are far from enough to catch
them up with their non-Hispanic peers.

While substantial, the
improvements in the
schooling attainment of
Hispanic workers are far
from enough to catch
them up with their
non-Hispanic peers.

REGIONAL ECONOMIST | www.stlouisfed.org 7

Figure 5

Educational Attainment of Hispanic Female and Male Workers
70

Hispanic Females 1950

60

Hispanic Males 1950

Percent

50

Hispanic Females 2016

40

Hispanic Males 2016

30
20
10
0

Primary
or Less

Secondary
Incomplete

Secondary
Complete

College
Incomplete

College Complete
or More

SOURCES: IPUMS USA and authors’ calculations.
NOTE: Educational attainment is defined by the following groups: primary school or less (nursery school
through eighth grade), secondary incomplete (ninth to 11th grade), secondary complete (12th grade),
college incomplete (one to three years of college), and college complete or more (four or more years of
higher education).

©THINKSTOCK/iSTOCK/PATTIES

As of 2016, the participation of
Hispanics in occupations such as
service workers, operatives and,
most strongly, farmworkers and
unskilled laborers is much higher
than their weight in the population
and labor force.

8 REGIONAL ECONOMIST | Second Quarter 2018

Indeed, from 1950 to 2016, the fraction
of non-Hispanics with no more than
a primary education collapsed from
47.7 percent to just 3.6 percent. The fraction of non-Hispanics with some college
surged from 7.2 percent to 20.6 percent,
while the fraction with college completed
or more jumped from 3.3 percent to
21.5 percent.
The Hispanic education gap can explain
the lower earnings for two different reasons. First, as shown in Figures 1 and 2,
workers with lower levels of schooling
are more likely to end up in lower-skill
occupations.
This form of assignment is simply driven
by comparative advantage: Despite having
lower productivity in absolute terms in all
occupations, lower-education workers are
relatively more productive in lower-skill
occupations. By contrast, workers with
a higher level of education may be more
productive in all occupations, but their
productivity would be relatively higher in
higher-skilled occupations.
In these cases, the assignment of workers to jobs and occupations would exhibit
positive sorting: Highly skilled workers
would be assigned in higher proportions to higher-skilled occupations, and
less-skilled workers would be assigned in
higher proportions to lower-skill occupations. The IPUMS USA data clearly
indicate that because of the group’s lower

educational attainment, Hispanic workers
in the U.S. have comparative advantage in
those lower-skill occupations, as revealed
by their underrepresentation in higherskilled occupations (professionals, managers) and by their overrepresentation in
lower-skill occupations.
The second mechanism by which
education could explain why Hispanic
workers earn less than their non-Hispanic
peers would be that, despite both being
in the same occupation, they earn less
because of their inferior training and
lower skills. That is, education determines
not only the assignment of workers to an
occupation but also their absolute advantage in each occupation.
We explored this second mechanism,
using the same IPUMS USA data to
control for the impact of observable factors such as age, education, gender and
experience on the income of workers.4
Once we controlled for the other observable factors, we found that the earnings
of Hispanics and non-Hispanics were
fairly similar, even in the highest-earning
occupations. Indeed, some gaps between
Hispanics and non-Hispanics persisted
after controlling for age and education,
but the magnitude of those gaps was too
small to account for much of the observed
differences in the raw data.
In sum, the education gap of Hispanic
workers relative to their peers in the U.S.

labor force is the key candidate factor
driving the gaps in the occupations and
earnings. The key question is: Why does
the education level of Hispanics lag
behind? Can this gap be accounted for by
intergenerational persistence, given the
fact that it may take a while for children
of poorly educated immigrants to catch
up with the rest of the population? In
any event, policies aiming to improve the
standing of Hispanics in the U.S. labor
markets—and in general—are likely to fail
unless they address the lower educational
attainment of that population.
Gender Differences
We finish our exploration of the
Hispanic labor force by looking at the
evolution of gender differences over
time. Figure 5 decomposes the educational attainment of Hispanics, showing
separately the educational attainment
of female and male workers in 1950 and
2016. The solid bars are for the year 1950,
and the striped bars are for 2016. The
lavender and black bars represent females
and males, respectively.
The first fact observed in the figure is
that both males and females have advanced
substantially. In 1950, less than one-tenth
of workers in both genders had at least
some college education. By 2016, more than
40 percent of workers in both genders had
at least some college education.
The second clear fact is that in both
periods—and over the entire sample
period—the educational attainment
of Hispanic women is higher than the
educational attainment of men. Indeed,
men outnumber women in the groups of
workers with an education in the lower
categories: primary or less, secondary
incomplete and secondary complete.
Meanwhile, women outnumber men in
the upper categories: college incomplete
and college complete or more.
Can this gap between female and male
Hispanic workers explain differences in
their labor market experience? A formal
answer is outside the focus of this article,
but it will be the focus for an article in an
upcoming issue of the Regional Economist. But a hint can be provided here:
The data show that since 1980, female
Hispanic workers have overtaken their
male counterparts in professional and
technical occupations. As of 2016, about
1 in 20 Hispanic female workers were

professionals compared to only 1 in 30 for
the males. This trend is in line with the
observed evolution for the overall population, as we detailed in our previous article
in the Regional Economist.5
Conclusions
In this article, we explored the substantial shifts in the participation of Hispanic
workers in the American labor force since
1950. We show that the presence of Hispanic workers has increased dramatically
over these years, more than eight times.
We also documented that the presence
of Hispanics has increased in all occupations. However, we found big differences
in their expansion across the different
occupations.
In particular, we found that Hispanic
workers are assigned in higher proportions to lower-skill occupations. To
explain these findings, we argue that the
education of Hispanic workers lags behind
that of non-Hispanic workers, so the
observed pattern is consistent with recent
assignment models, in which workers
choose occupations on the basis of their
comparative advantage (e.g., Costinot
and Vogel).6
The data suggest that the observed disadvantages of Hispanic workers in the U.S.
labor markets do not seem to be the result
of labor market frictions, e.g., sheer discrimination or lack of information. Once
we control for education, age and gender
differences, the Hispanic/non-Hispanic
gaps mostly disappear. Instead, the factor
holding back Hispanic workers in the U.S.
seems to be their educational attainment.
To be sure, progress has been made
there, but the Hispanic population is
still lagging behind the rest of workers.
Regardless of the reason why Hispanics
remain behind, the aggregate productivity of the country would improve if that
source of talent and skills is fully developed and exploited.7
Finally, we found that women have
advanced at a faster pace than men have. The
patterns for Hispanic females are in line with
the findings for all the workers in general.8

ENDNOTES
1

2

3
4
5
6
7
8

See IPUMS USA, University of Minnesota, www.ipums.
org. We discard individuals whose employment status
is unknown, and those who are unemployed or not in
the labor force, as classified by the variable EMPSTAT
codes 0, 2 and 3.
Skill intensity is measured by the percentage of
workers in an occupation with the highest year of
school degree completed in 1950 being college or
more. Thus, the higher the percentage of workers in
an occupation with at least a college degree in 1950,
the more skill-intensive an occupation is. The order of
the top four occupations is preserved if we use 2016
instead of 1950 to measure skill intensity.
Observations of individuals with unclassified, missing
or unknown occupations are discarded.
This is what is known in the literature as “Mincer”
regressions.
See Monge-Naranjo and Vizcaino, 2017.
See Costinot and Vogel.
This point is forcefully made by Hsieh et al.
See Rendall.

REFERENCES
Costinot, Arnaud; and Vogel, Jonathan. Matching and
Inequality in the World Economy. Journal of Political
Economy, Vol. 118, No. 4, 2010, pp. 747-86.
Hsieh, Chang-Tai; Hurst, Erik; Jones, Charles I.; and
Klenow, Peter J. The Allocation of Talent and U.S.
Economic Growth. Unpublished manuscript, 2018.
See http://klenow.com/HHJK.pdf.
Monge-Naranjo, Alexander; and Vizcaino, Juan Ignacio.
Shifting Times: The Evolution of the American Workplace. Regional Economist, Fourth Quarter 2017,
Vol. 25, No. 4, pp. 4-9.
Rendall, Michelle. Brain versus Brawn: The Realization
of Women’s Comparative Advantage. Unpublished
manuscript, 2018. See https://sites.google.com/site/
mtrendall/research.
Ruggles, Steven; Genadek, Katie; Goeken, Ronald;
Grover, Josiah; and Sobek, Matthew. Integrated Public
Use Microdata Series [IPUMS] USA: Version 6.0 [data
set]. Minneapolis: University of Minnesota, 2015.
http://doi.org/10.18128/D010.V6.0.

Research assistance was provided by Hee Sung
Kim, a senior research associate, and Qiuhan
Sun, a research associate, both at the Federal
Reserve Bank of St. Louis.
(This article was published online June 27.)
REGIONAL ECONOMIST | www.stlouisfed.org 9

Housing Costs and Regional
Income Inequality in China
and the U.S.
By Brian Reinbold and Yi Wen
©THINKSTOCK/ISTOCK/EmJcox

KEY TAKEAWAYS
• Living standards within a country
can vary greatly due to differences in
regional housing costs.
• Adjusting income by regional housing prices provides a better picture of
income inequality.
• When accounting for housing prices,
the degree of inequality among China’s
provinces improves. For U.S. states, this
measure of inequality worsens.

M

easuring income inequality has
long been a key focus in welfare
economics. Economists have documented
that income inequality has increased in
virtually all advanced economies, but it has
remained mixed for developing and emerging economies.1
However, solely looking at income tells
only part of the story about the difference
in people’s living standards because income
does not reveal information about the cost
of living. For example, housing prices can
vary tremendously across a nation; they
also vary based on urban, suburban and
rural geography.
Since housing typically takes a large share
of an individual’s income and living space
is one of the most important and scarce
commodities, housing prices can greatly
affect living standards. In other words, the
purchasing power of a dollar is not the same
across regions due to variations in the cost
of living, especially housing. Therefore,
factoring in cost of living can yield fruitful
insights about true inequality.
In this article, we look at regional per
capita disposable income data for China,
a developing nation, and the U.S., an
advanced economy, to see how income
inequality compares between two large
countries with a substantial income gap.
Disposable income is income less taxes,
which is the income an individual has
10 REGIONAL ECONOMIST | Second Quarter 2018

available for consumption. Looking at
regional data will also allow us to see how
income varies across a geographic region.
Note that we are using regional per capita
disposable income within a country, which
allows us to characterize the degree of
inequality for a region’s average household
in relation to different regions. Thus, we
are unable to say anything about inequality
across individuals.
We will then adjust regional average
income by housing prices in a region to
see how that affects inequality in living
standards.
We calculate the Gini coefficient for
each country using each region’s per capita
disposable income to measure income
inequality. The Gini coefficient takes values
between 0 and 1. A value of 1 indicates
maximum inequality, while a value of 0
means perfect equality. Since we are using
average income in a region, the Gini coefficient will be less than if we were using each
individual’s own income in each country.2
Per Capita Disposable Income
First, we look at per capita disposable
income by province and the municipalities
of Beijing and Shanghai for China (excluding
Hong Kong and Taiwan). Income tends to
be concentrated along the rapidly developing
eastern coast, especially in the municipalities
of Beijing and Shanghai. Both municipalities
have the highest average per capita disposable income, at just under 50,000 yuan. (The
exchange rate is about six yuan per dollar.)3
The innermost provinces in western China,
Gansu and Xizang, have the lowest per capita
disposable income, at around 12,000 yuan.
The highest regional income per capita is four
times greater than the poorest. The crossregional average of per capita disposable

income is about 22,000 yuan, and the crossregional median of per capita disposable
income is about 18,600 yuan. The Gini of
cross-regional income per capita is 0.19.
Now looking at U.S. states and the District of Columbia (D.C.), we see a similar but
less pronounced skew in income distribution. The richest region is D.C., with per
capita disposable income of around $63,000.
The poorest state is Mississippi, with per
capita income of around $32,000. In the
U.S., the highest regional income per capita
is two times greater than the poorest. This is
less than half of the difference in China. The
cross-regional average of per capita disposable income is $42,027, the cross-regional
median of per capita disposable income is
$40,829, and the cross-regional Gini of per
capita disposable income is 0.08.
We see that regional income inequality
is much greater in China than in the U.S.
The Gini coefficient in China is more than
twice as large as that in the U.S.
China’s rapid development has contributed to inequality so far, but the historical
experience of the U.S. suggests that China’s
regional inequality may start to shrink as
China further develops.
Adjusting Regional Income
by Regional Housing Prices
Since housing usually represents a large
portion of consumer expenditures, housing
prices can greatly affect a household’s living standards. Although it may not account
for most of consumption, it is the most
important component of spending.
For example, given a certain income, a
person could afford either a small apartment in New York City or a large, singlefamily house in St. Louis.
Because most daily consumption goods

ABOUT THE AUTHORS
Yi Wen is an economist and assistant vice president at the Federal Reserve
Bank of St. Louis. His research interests include macroeconomics and the
Chinese economy. He joined the St. Louis Fed in 2005. Read more about the
author and his research at https://research.stlouisfed.org/econ/wen.
Brian Reinbold is a research associate at the Federal Reserve Bank of St. Louis.

are tradable and mobile across a country
through a nationwide grocery market, the
cost of living in terms of daily consumption goods does not change dramatically
across regions. However, housing is a specially localized good, and it is not tradable
or mobile. Therefore, the main source of
the cross-regional difference in living standards comes from the difference in housing
prices rather than grocery prices, even if
grocery consumption accounts for a larger
proportion of consumer spending.
In addition, housing prices have
increased both in the U.S. (except during
the recent financial crisis) and in China, so
people’s living standard—the purchasing
power of their income—must have changed
in recent years. To measure this effect,
we use regional housing prices to adjust
per capita disposable income. Notice that
cross-country comparison is meaningful in
our context only if we use nominal housing
prices instead of a housing price index.
In 2015, housing on average accounted
for about 22 percent of consumption
expenditure in China,4 and it represented
about 33 percent of household expenditure in the U.S. in 2016.5 We used housing
prices of over 100 Chinese cities to construct regional housing price levels. The
average price of new housing across China’s
provinces is 855 yuan per square foot, and
the median price is 649 yuan. Home prices
in the most expensive city, Shanghai, are
nearly nine times greater than in the least
expensive province, Shaanxi.
For the U.S., we look at Zillow’s median
house listing price by state for a mix of new
and existing homes by state. The crossregional average is $143 per square foot,
and the median price is $124. The price of
housing in D.C., the most expensive region,
is over five times higher than home prices
in Indiana, the least expensive state.
Housing prices are unequal across regions
in both China and the U.S., but the disparity is greater in China. Interestingly, regions
with high per capita disposable income
also tend to have high housing prices. This
suggests that, everything else being equal,
a high-income region does not necessarily
have a high living standard when the cost of
housing is taken into account.
To create a measure of living standard,
we adjusted disposable income by housing
prices for both China and the U.S. Namely,
we divided the regional nominal per capita
disposable income by its respective

Gini Coefficients: China and the U.S.
2015 Regional Per Capita
Disposable Income

2015 Regional
Housing Prices

2015 Regional
Standards of Living*

s Gini in China

0.19

0.30

0.16

Gini in U.S.

0.08

0.21

0.12

*A region’s per capita disposable income adjusted by its average housing price
SOURCES: National Bureau of Statistics of China, U.S. Bureau of Economic Analysis, China Index
Academy/Soufun, Zillow, Haver Analytics and authors’ calculations.

regional housing price in each region.
Across China’s provinces, the standard
of living varies much less than income
alone. The province with the highest living
standard is Shaanxi, and the province with
the lowest living standard is Hainan.
Indeed, the Gini for living standard is
0.16. This is less than the previous measure based solely on disposable income in
China, suggesting improved distribution
and equality. Municipalities with seemingly high income (Shanghai and Beijing)
now have low standards of living in terms
of housing affordability.
In the U.S., however, inequality in living
standard actually increases across states.
The Gini is 0.12, which is over a 40 percent
increase compared to the Gini in disposable income. Indiana is now the “richest”
state in terms of living standard, and
Hawaii is now the “poorest” state in terms
of living standard, or housing affordability. The living standard is about 4.5 times
greater in Indiana than in Hawaii. Much
of this inequality is driven by states where
high home prices greatly reduce living
standards relative to the median of states.
One caveat is that not everyone is a
homeowner. So alternatively we could use
the absolute rental cost in each region to
adjust regional average disposable income to
capture the renter population’s inequality. But
to the extent that rental cost is proportional
to housing prices, our measure of living standard may not change dramatically if disposable income is adjusted by rental cost instead.

same consumption bundle.
We see that regional inequality is substantially less severe in the U.S. than in
China when considering only disposable
income. However, this gap in inequality
between the two countries shrinks significantly once regional variations in the cost
of housing are taken into account.
In terms of purchasing power of income
on nontradable goods like housing, the
cross-regional inequality in China is not
much more extreme than that in the U.S.,
although housing is much more affordable
in the U.S. than in China—thanks to much
higher per capita income and significantly
more arable land in the U.S.
Still, this gap in per capita income
remains enormously large. It will take
China 60 years—about two generations—
to erase the difference, assuming the
country can maintain a growth rate that
is 4 percentage points higher than the U.S.
rate during that time.
(This article was published online June 8.)

ENDNOTES
1
2

3
4

Conclusion
Adjusting income by cost of living
can provide useful insights about living
standards because, ultimately, household
income means only as much as the purchasing power of that income. Yet the cost of
living is not equal across regions, especially
with respect to nontradable and nonmobile
consumption goods such as housing. People
in high-income regions may have to pay a
disproportionately higher cost to enjoy the

5

See Dabla-Norris et al.
For example, an individual whose income is in the
90th percentile will have an income many times
greater than that of someone in the bottom 10th
percentile. This huge difference in income between
the top and bottom raises the Gini coefficient. When
looking at regional income data, the relative difference between wealthiest regions and the poorest will
not be nearly as large. Therefore, the Gini coefficient
will be smaller in this case.
As of March 29, 2018.
This is based on authors’ calculations using
consumption expenditure data from the National
Bureau of Statistics of China.
This is based on authors’ calculations using
consumption expenditure data from the
Bureau of Labor Statistics. See BLS.

REFERENCES
Bureau of Labor Statistics. Consumer Expenditures—
2016. News release, Aug. 29, 2017. See www.bls.gov/
news.release/cesan.nr0.htm.
Dabla-Norris, Era; Kochhar, Kalpana; Suphaphiphat,
Nujin; Ricka, Frantisek; and Tsounta, Evridiki. Causes
and Consequences of Income Inequality: A Global
Perspective. IMF Staff Discussion Note, International
Monetary Fund, June 2015.
REGIONAL ECONOMIST | www.stlouisfed.org 11

Unequal Pink Slips? Gender and
the Risk of Unemployment
By Guillaume

Vandenbroucke and Heting Zhu
©THINKSTOCK/iSTOCK/DIMA_SIDELNIKOV

12 REGIONAL ECONOMIST | Second Quarter 2018

Changes in the Gender Gap
during Recessions
Changes in unemployment, which are
large during recessions, can have important welfare consequences. But, how do
recessions affect the unemployment gap?
We address this question in Figure 2. To
build this figure, we considered the last
seven recessions identified by the NBER.1
First, we collected the data on gender gap
in unemployment that start with the beginning of each recession and end 24 months
later. Then, to allow for an easy comparison
between each of the seven series of numbers, we “normalized” the gap to zero at the
beginning of each recession. This explains
why the lines in Figure 2 all start at zero.
An example can help. Take the case of
the 1980 recession. When the recession
started in January 1980, the unemployment

Figure 1

The Gender Unemployment Gap
3
2
1
0
–1

SOURCES: FRED (Federal Reserve Economic Data), National Bureau of Economic Research, Bureau of
Labor Statistics and authors’ calculation.
NOTES: The gap is the female unemployment rate minus the male unemployment rate; a positive gap
means women were more exposed to joblessness than men. Data are for workers aged 16 and older.
Shaded areas indicate a recession.

ABOUT THE AUTHORS
Guillaume Vandenbroucke is an economist and research officer at the Federal
Reserve Bank of St. Louis. His research focuses on the relationship between economics
and demographic change. He joined the St. Louis Fed in 2014. Read more about the
author and his research at https://research.stlouisfed.org/econ/vandenbroucke.
Heting Zhu is a senior research associate at the Federal Reserve Bank of St. Louis.

2018

2013

2008

2003

1998

1993

1988

1983

1978

1973

1968

–3

1963

–2
1958

en and women fare differently in the
labor market. There is, for instance,
a large literature documenting earning
differences between men and women who
work the same job and have comparable
education and experience. Similarly, it is
well-known that progression and promotion in the workplace often seem more
difficult for women than for men.
In this article, we discuss another
facet of the difference between men and
women in the labor force: their exposure
to unemployment. We complement our
analysis with a discussion of the blackwhite exposure to unemployment and
show that it behaves noticeably differently
than male-female exposure.
Figure 1 shows the difference between the
unemployment rates of women and men
since the late 1940s. We call this the “gender
unemployment gap.” The shaded areas
represent recessions dated by the National
Bureau of Economic Research (NBER).
A few observations are worth noting.
First, the gap tended to be positive before
the 1980s; it was arguably large during
the 1960s and 1970s, when the gap was
between 1 and 2 percentage points. This
means that for a long period of time, the
unemployment rate of women was above
that of men, i.e., women faced higher
unemployment risk than men.
Second, throughout the 1980s and until
the last recession, the gap was no longer as
large as it had been. Instead, the gap seemed

1953

M

1948

• The U.S. jobless rate for women had
been higher than that for men for more
than three decades after World War II.
• Starting in the 1980s, the gender
unemployment gap shrank.
• Women now appear to be less exposed
to increased unemployment during
recessions than men.

Percentage Points

KEY TAKEAWAYS

to hover just above or below zero, suggesting
that women and men faced a similar risk of
unemployment during this period.
Finally, the unemployment gap exhibits
a tendency to decrease during recessions.
This is particularly clear in the last recession. The unemployment rates of men
and women were very close in the months
leading up to the recession. In June 2007,
for instance, the unemployment rate was
4.7 percent for men and 4.4 percent for
women. But the unemployment rate of
men rose to 11 percent in January 2010
versus 8.4 percent for women, causing
a gap of almost –3 percentage points
between them at the end of the recession.
The decline of the gender gap in the
unemployment rate indicates that women
appear to be less exposed to increased
unemployment during recessions than men.

1960

1969

1973

1990

2001

2007

from 1973 to 2001, the Great Recession was
followed by a reduction in the unemployment risk of women relative to men. But
the magnitude of the reduction is dramatically different: Two years after the start of
the recession, the gender unemployment
gap was about 2 percentage points lower. In
summary, these post-1970 recessions imply
a lasting reduction in the unemployment
risk of women relative to men, but the last
recession stands out in the magnitude of
this reduction.

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

The Racial Gap

Figure 2

Changes in the Gender Unemployment Gap during Recessions

Percentage Points

1

0

–1

–2

–3
0

1

2

3

4

5

6

7

8

1980

Months since Start of Recession

1

A similar analysis can be conducted
across race instead of across gender. Figure
3 is analogous to Figure 2, but the gap analyzed there is the difference between black
and white. A positive gap means that the
black unemployment rate is higher than
the white unemployment rate.2
The lesson from Figure 3 is remarkably
different from that of Figure 2. First, the 2007
recession does not particularly stand out.
Second, all the plotted recessions exhibit an
increasing gap in the two years following the
start of the recession. Black workers become
relatively more exposed to unemployment
than white workers after a recession.

0

Conclusion

SOURCES: FRED (Federal Reserve Economic Data), National Bureau of Economic Research, Bureau of
Labor Statistics and authors’ calculation.
NOTES: A positive number indicates that the gap increased, i.e., the risk of unemployment rose more for
women than men as the recession progressed. Conversely, a negative number indicates that the gap fell,
i.e., the risk increased less for women than men.

Figure 3

Changes in the Racial Unemployment Gap during Recessions
3

Percentage Points

2

1973

1980

2001

2007

1990

–1
–2
0

1

2

3

4

5

6

7

8

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Months since Start of Recession

SOURCES: FRED (Federal Reserve Economic Data), National Bureau of Economic Research, Bureau of
Labor Statistics and authors’ calculation.
NOTES: A positive number indicates that the gap increased, i.e., the risk of unemployment rose more for
blacks than for whites as the recession progressed. Conversely, a negative number indicates that the gap
fell, i.e., the risk increased less for blacks than whites.

gap between men and women was 1.1 percentage points, i.e., the women’s unemployment rate was 1.1 percentage points higher
than that of men. In May 1980, which was
the fourth month after the start of the
recession, the gap was 0.1 percentage point.
Thus, the gap decreased by 1 percentage
point. Hence, the –1 can be seen in Figure
2 at the fourth month after the start of the
1980 recession.
There are three groups of recessions
that stand out in Figure 2. Consider
first the 1960 and 1969 recessions. The

unemployment gap did not decrease significantly; it remained positive or near zero
for one to two years after the start of the
recession. This confirms the first observation made about Figure 1, that is, women
faced higher unemployment risk than men.
The second group comprises the recessions from 1973 to 2001. These recessions
show an approximately 0.5 to 1 percentage
point decrease in the unemployment gap
two years after the start of the recession.
Finally, the Great Recession—that is, the
2007 recession—stands out. Like recessions

We do not have a theory of the different patterns exhibited across recessions in
Figure 2. Similarly, we do not have a theory
of the difference between Figure 2 and
Figure 3. We have documented the patterns
of these gaps, but an explanation of these
patterns is beyond the scope of our article.
Yet the patterns raise important questions.
Why are women relatively less exposed to
the unemployment risk after recessions?
Why are black workers relatively more
exposed? Why does the Great Recession
appear so different for the gender gap but
not the race gap? Further research aimed at
explaining these patterns would be of great
interest.
(This article was published online May 7.)

ENDNOTES
1

2

We do not consider the recession that started in July
1981 since it is subsumed in the two-year period after
the start of the recession that began in January 1980.
Data for the black unemployment rate are not
available for the 1960s.

REGIONAL ECONOMIST | www.stlouisfed.org 13

Accounting for Age:
The Financial Health of Millennials
By YiLi Chien and Paul Morris
©THINKSTOCK/iSTOCK/Andrea Obzerova

KEY TAKEAWAYS
•		Millennials and Generation X were the
youngest working generations in 2016
and 2001, respectively. How do their
balance sheets compare?
• Because of fewer assets and more
debt, millennial households had an
average net worth of about $90,000 in
2016 versus $130,000 for Generation X
households in 2001.
• Spending more time in school and
delaying marriage and other major life
events may explain why millennials
have a lower net worth.

T

here is no shortage of news articles
written on the saving and investment
behaviors of millennials. What is lacking,
however, is a clear picture of what these
young people are doing with their money.
The Wall Street Journal has reported
concerns about low levels of saving related
to mounting student loan and credit card
debt. By contrast, the personal finance website NerdWallet pointed out that some millennials are saving considerable amounts
for retirement.
There are also conflicting reports on
their homebuying behavior. Real estate
news website The Real Deal noted that
millennials are not buying homes because
of high student loan balances, but Business
Insider reported that they are waiting longer to purchase their first homes and tend
to purchase homes that are more expensive
when they do buy.
Given that these articles fail to come to
any consensus, we aim to offer a glimpse
into the state of millennials’ household
finances. To see how millennials fared
relative to the previous young generation,
Generation X, we compared millennial
households’ finances in 2016 to those of
Gen X back in 2001.1 We analyzed the
average asset and liability positions and
14 REGIONAL ECONOMIST | Second Quarter 2018

their compositions using householdlevel data from the Survey of Consumer
Finances (SCF).2
Overall, our analysis indicates that
reductions in both financial assets and
nonfinancial assets (e.g., a home) contributed to millennials’ having fewer overall
assets than Gen Xers had in 2001. In terms
of liabilities, the millennials were slightly
more indebted on average, as they held
a higher amount of student loans that
outweighed reductions in mortgage and
credit card debt.3
A Lower Net Worth
The average value of total assets was
lower among millennials than Gen Xers.
As shown in Figure 1, millennials held an
average of $162,000 of assets relative to
Gen X’s average of $198,000.4
The reduction occurred in both financial and nonfinancial assets. The average
financial asset position was around $15,000
lower than in 2001, dropping from $65,000
to $50,000. The reduction of nonfinancial
assets was $22,000, dropping from $133,000
to $111,000.
Part of the reduction of the nonfinancial
asset position occurred in housing. Millennials held an average of $69,000 in their
primary residence, while Gen Xers held an
average of $78,000. While millennials held
lower levels of assets on average, they had
a slight advantage in average retirement
account balance, at $15,500 relative to
Gen X’s $13,600.
Millennials held a slightly higher level of
total debt, at an average of $72,000 compared to Gen X’s average of $67,000. While
the average levels of debt were similar
across the two generations, the composition
was markedly different. Average student

loan levels surged from $4,200 for Gen X
to $14,700 for millennials. Because of the
smaller average value of housing assets
for millennials, the level of mortgage debt
was also smaller at $43,000 compared to
$49,000 for Gen X.
We also observed that the burden of
credit card debt among millennials was
actually lower than that of the previous
generation. The unpaid credit card balance
stood at $1,800, which was lower than Gen
X’s average $2,700 (not shown in Figure 1).
In short, we see that millennials’ average
asset position was lower, while they held
slightly more debt, which led to an average
net worth of $90,000 for millennials and
$130,000 for Gen X.
A Robustness Check
The prices of some asset categories may
have changed significantly from 2001 to
2016. While the dollar values in the SCF
are inflation-adjusted to 2016 dollars, this
does not account for swings in the relative
prices between asset categories that could
make one category disproportionately
more expensive in one year than another.
To alleviate this concern, we performed
a simple robustness check. For each asset
category, we computed the ratio of the
average value for each generation to the
average of all households in those respective years. We report the results in Figure 2.
These ratios are best interpreted as a
percentage of the average value for all
households. For example, a ratio of
0.2 indicates that the generation in question held assets or liabilities equal to
20 percent of the average value across
all households in that year.
The orange bar in the total assets category
of Figure 2 represents the ratio of the average

ABOUT THE AUTHORS
YiLi Chien is an economist and research officer at the Federal Reserve Bank of
St. Louis. His areas of research include macroeconomics, household finance
and asset pricing. He joined the St. Louis Fed in 2012. Read more about the
author and his research at https://research.stlouisfed.org/econ/chien.
Paul Morris is a senior research associate at the Federal Reserve Bank of St. Louis.

Figure 1

Changing Priorities

The Changing Balance Sheet across Generations

The net worth of the youngest working generation fell since 2001, as they held
fewer assets and more debt on average.
However, this is not necessarily an
indictment of millennials’ spending and
saving habits. Society is in a state of transition as the life cycle continues to extend.
People have been living longer and retiring
later, and with that has come a multitude of
other demographic shifts.
Relative to previous generations, more
millennials have opted to delay entering
the labor market, with many deciding to
pursue higher levels of education. The labor
force participation rate for 20- to 24-yearolds dropped to 70.5 percent in 2016 from
77.1 percent in 2001. Over the same period,
the share of those ages 25 to 29 with four
years of college or more increased from
28.4 percent to 36.1 percent. In addition,
a higher percentage of young adults are
living with their parents, and the median
age at first marriage has been increasing for
both men and women.6
We observe that millennials have been
going to school longer and delaying major
life events. Thus, it makes sense that they
hold lower levels of assets. They have had
less time in the labor force, and a smaller
share of them have moved out on their
own, which contributes to the lower levels
of residential assets. However, they have
shown a higher propensity to save for
retirement and to avoid credit card debt.
While millennials hold higher levels
of student loans, education is often an
investment that improves productivity
and future earnings. Given these considerations, the concerns regarding millennials’ spending and saving habits may be
at least partially eased, as they will likely
have more time in the labor force to accrue
assets and pay off their debts.

$200,000

Millennials in 2016

$150,000

Generation X in 2001

$100,000

Asset Categories

Education
Loans

Mortgage

Total
Debt

Primary
Residence

Nonfinancial
Assets

Retirement
Accounts

Financial
Assets

$0

Total
Assets

$50,000

Liabilities
Categories

Net
Worth

SOURCES: Survey of Consumer Finances and authors’ calculations.
NOTES: Data are for the average household with a head between ages 20 and 35 in 2016 (millennials)
and 2001 (Gen Xers). The chart displays total assets and types of assets, total debts and types of debts,
and net worth, which is total assets less total debts. Financial assets include retirement accounts, and
nonfinancial assets include primary residence.

Figure 2

Ratio of Balance Sheet Value of Young Generations to Average Value of All Ages
1.0
0.8
0.6

Millennials in 2016
Generation X in 2001

0.4

Asset Categories

Mortgage

Total
Debt

Primary
Residence

Nonfinancial
Assets

Retirement
Accounts

Financial
Assets

0.0

Total
Assets

0.2

Liabilities
Categories

Net
Worth

SOURCES: Survey of Consumer Finances and authors’ calculations.
NOTES: The chart represents the ratio of balance sheet values for the average household with a head
between ages 20 and 35 in 2016 (millennials) and 2001 (Gen Xers) relative to the average value for all
households that year. The data are for total assets and types of assets, total debts and mortgage debts,
and net worth, which is total assets less total debts. Financial assets include retirement accounts, and
nonfinancial assets include primary residence. Education loans are excluded because the average value
for all households is much smaller than the average value for households headed by young adults, leading
to ratios well over 1.

total assets of Gen X to those of all households in 2001, while the blue bar represents
a similar ratio for millennials in 2016.
In this relative measure, the millennials had a significantly smaller asset ratio
(21 percent) than Gen Xers (32 percent).
The financial assets, nonfinancial assets
and housing ratios for millennials each
dropped about 10 percentage points, and
the retirement account ratio fell by about

5 percentage points.
By contrast, the average debt ratio was
lower for millennials. Compared to those
of Generation X, the total debt and mortgage ratios were down around 15 and 23
percentage points, respectively. Yet, these
lower debt ratios were outweighed by lower
asset ratios, thus pushing millennials’ net
worth ratio down to 13 percent from 24
percent for Generation X.5

(This article was published online May 16.)

ENDNOTES
1

2

We define millennial households as those whose
heads are between ages 20 and 35 as of 2016, and
we define Generation X households as those whose
heads were in the same age range back in 2001. While
there is no clear demarcation of generational boundaries, our definitions roughly match those popularly
referenced.
The survey provides cross-sectional data on U.S.
households’ demographic characteristics, incomes,
balance sheets and pensions every three years.

(continued on Page 21)
REGIONAL ECONOMIST | www.stlouisfed.org 15

INDUSTRY PROFILE

St. Louis Fed Steps in
to Provide More-Timely Jobs Data
By Charles Gascon and Paul Morris

• Bureau of Labor Statistics revises
state and local employment data just
once a year.
• The St. Louis Fed analyzes the data
and reports on it four times a year.
• This extra reporting can take the
surprise out of a once-a-year revision.

O

n March 12, the U.S. Bureau of
Labor Statistics (BLS) released its
annual revision to its monthly state and
local employment data. The latest revision shows weaker growth across the
Federal Reserve’s Eighth District (based
in St. Louis) than initially reported.1 For
example, growth from the fourth quarter
of 2016 to the third quarter of 2017 in
Arkansas was revised down from 2 percent to 0.8 percent; for another example,
Kentucky’s growth was revised down
from 1.6 percent to 0.5 percent.
Although these revisions were significant, they didn’t come as a surprise
to us at the St. Louis Fed. We have been
releasing our own estimates of regional
employment growth since mid-2017, and
they have generally matched up well with
the revisions released every March by the
BLS. Our estimates can alert policymakers of likely revisions well ahead of time,
allowing them to make decisions based on
information that is often more accurate
than the initial releases from the BLS.
The BLS uses its Current Employment
Statistics (CES) program to produce
monthly estimates of nonfarm payroll
employment. Once a year, it revises these
figures, relying largely on data from its
Quarterly Census of Employment and
Wages (QCEW) program. Rather than
wait for the annual revision, we have
been producing our own quarterly job
figures based on the most recent QCEW
data. Back in December, we released
our estimates that showed weaker
16 REGIONAL ECONOMIST | Second Quarter 2018

Figure 1

We Weren’t As Surprised When the Data Were Revised
An Example: Total Nonfarm Employment in Kentucky
1960
Thousands of Employees

KEY TAKEAWAYS

1940
1920
1900
1880

CES (December 2017)

1860

St. Louis Fed (December 2017)

1840

CES (March 2018)

1820
1800
2013

2014

2015

2016

2017

2018

SOURCES: Bureau of Labor Statistics (BLS) and authors’ calculations.
NOTE: Using Kentucky as an example, the figure shows the initially released data (blue dashed line) as
of December 2017, alongside the St. Louis Fed’s estimate at the same point in time (red line). The black
dotted line shows the “true” revised values released by the BLS in March 2018. The initial release and the
revised data are from the BLS’ Current Employment Statistics (CES) program; our estimates use data from
the BLS’ Quarterly Census of Employment and Wages program.

employment growth across the Eighth
District than was being reported by the
BLS at the time.2
Predictable Data Revisions
Figure 1 plots time series of the initially
reported BLS data as of early December for
Kentucky, along with our estimates and
the revised data released in March by the
BLS. Our estimates provide us with more
up-to-date information on the expected
direction and magnitude of the revision
in a particular area. In December, we
expected the BLS to revise the initial CES
release down from year-over-year growth
of 1.6 percent to 1.1 percent. In March, the
BLS revised growth down to 0.5 percent.
While we were not expecting such a large

revision, our estimates correctly indicated
a downward revision and were closer to the
revised CES data than was the BLS release
available in December.
Figures 2 and 3 report year-over-year
growth for the states and four largest
metropolitan statistical areas (MSAs) in
the Eighth District. We excluded Illinois
because the majority of economic activity
in the state occurs in the Chicago area,
which is part of the Federal Reserve’s
Seventh District. The BLS revised employment growth down in every state and in
each of the four largest MSAs with the
exception of Little Rock, Ark. In addition
to capturing the single upward revision,
our December estimates correctly predicted downward revisions across most

ABOUT THE AUTHORS
Charles Gascon (left) is a regional economist and a senior coordinator in the Research
Division of the Federal Reserve Bank of St. Louis. His focus is studying economic
conditions in the Eighth District. He joined the St. Louis Fed in 2006. Read more
about the author and his research at https://research.stlouisfed.org/econ/gascon.
Paul Morris is a senior research associate at the St. Louis Fed.

Figure 2

Breaking Down the Data across the Eighth District
District and State Employment Growth
2016:Q4 to 2017:Q3
2.0

CES (December 2017)
St. Louis Fed Estimates (December 2017)
CES (March 2018)

Percent

1.5

1.0

0.5

0.0
Eighth
District

Arkansas

Kentucky

Indiana

Missouri

Mississippi

Tennessee

SOURCES: Bureau of Labor Statistics and authors’ calculations.
NOTE: We excluded Illinois because the majority of economic activity in the state occurs in the Chicago
area, which is part of the Federal Reserve’s Seventh District. The Eighth District is based in St. Louis.

time, allowing them to

Does the Pattern Hold Up for the District’s Biggest Cities?
Metropolitan Statistical Area Employment Growth
2016:Q4 to 2017:Q3

1.5

make decisions based

CES (December 2017)
St. Louis Fed Estimates (December 2017)

on information that is

CES (March 2018)

often more accurate

1.0
Percent

policymakers of likely
revisions well ahead of

Figure 3

2.0

Our estimates can alert

than the initial releases

0.5

from the BLS.

0.0
–0.5
–1.0

Little Rock, Ark.

Louisville, Ky.

St. Louis

Memphis, Tenn.

SOURCES: Bureau of Labor Statistics and authors’ calculations.
NOTE: The four cities are the largest MSAs in the Fed’s Eighth District, which is based in St. Louis.

of the Eighth District. The exceptions
were Indiana, where the BLS’ revision was
negligible, and Mississippi.
One way to think of the initial release
of employment data is as an estimate of
the “true” value that will be released in
March, similar to our estimate produced
at the same time. You can see that our
estimates are much closer to the values
released in March, on average.3 The average error of the initially reported CES data
was 0.6 percent; our error was half that, or
0.3 percent. Our estimates improved upon
the initial release in four of the six states

and three of the four MSAs.
In addition to improving upon the
initially reported data in Kentucky, our
estimates performed well in Arkansas.
While our estimates overshot the downward revision, we brought the prediction
error down to 0.4 percent from an initially
reported 1.1 percent.
Memphis was a unique case. Our
estimates suggested a steep downward
revision. The direction of the revision was
correctly predicted but was of a much
smaller magnitude than anticipated,
resulting in a larger prediction error.
REGIONAL ECONOMIST | www.stlouisfed.org 17

Are data revisions worthy of your attention?

Decisions Based on Early Data:
Be Careful
Analyzing initially reported employment data can lead to incorrect conclusions
when significant revisions occur, as we saw
in Arkansas and Kentucky. Thus, an awareness of the expected revisions is important.
Revisions can occur for a variety of
reasons, including: sample size may be
small, new firms may not complete the BLS
surveys when the firms are initially formed
(leading to understating employment) and
closing firms may not respond (leading to
overstating employment).
While our December estimates do not
perfectly match up with the BLS’ revisions, ours serve as useful indicators
of where we might expect employment
growth to be when March arrives. Continue to check back, as we intend on
releasing our employment estimates
regularly in the future.
Calculating Our Estimates
We use a process developed at the Dallas
Fed known as early benchmarking.4 It uses
the same administrative data that the BLS
uses for its annual benchmark revision.
Around the 20th of each month, the
BLS releases estimates of state and local
employment for the previous month
produced from its CES survey. This is a
voluntary survey of businesses and samples
about 7 percent of establishments. The BLS
relies heavily on its QCEW data for its revision, which is less timely but is collected
from all establishments with employees
covered by unemployment insurance.
Because the BLS releases QCEW data with
a six-month lag but only benchmarks in
March, we have been able to produce early
18 REGIONAL ECONOMIST | Second Quarter 2018

estimates of revised state and local employment after each release of the QCEW.
Keep Up with ALFRED
The BLS’ revised employment data for
states, MSAs and industries across the
nation far exceed the scope of the estimates
that we produce. Fortunately, you can
examine how the revisions changed the
story of employment growth in a particular
region using archived data. The St. Louis
Fed maintains records of data revisions in
its ALFRED (ArchivaL Federal Reserve
Economic Data) database, which allows
you to retrieve vintage versions of data that
were available on specific dates in history.
This means that you can compare initial
releases with revised data for any of the
nonfarm payroll series available in FRED
(Federal Reserve Economic Data, which is
our signature database).
Step by Step Instructions,
Using Missouri as an Example
To produce a line graph showing yearover-year growth rates of the data before
and after the revision for a particular state
or MSA, follow the procedure we outline
below for the state of Missouri.
1. Start at the GeoFRED map of total
nonfarm employment for states
(http://geof.red/m/9m7). A corresponding map for MSAs is also available
(http://geof.red/m/9m6).
2. Access the FRED page containing nonfarm payroll data for Missouri by clicking on the state. Then select the Details
and Data tab and follow the link directly
under the tab you just clicked.
3. Click on ALFRED Vintage Series in the

Related Content section underneath the
chart to access the series in the ALFRED
database.
4. Click on the Edit Graph button. Under
the Format tab, change the graph type
from bar to line.
5. Click on Edit Lines, select either Line 1
or Line 2, and change the units to
Percent Change from Year Ago and
copy to all.
6. Remain under the Edit Lines tab. Select
vintage date “2018-03-12” for Line 1 and
“2018-01-23” for Line 2, respectively.
7. Select a starting date for the graph dating
back to at least the beginning of 2016.
This ensures that the entirety of the BLS’
revision period is visible.
The final line graph is available at https://
alfred.stlouisfed.org/graph/?g=jXjb.
(This article was published online May 31.)

ENDNOTES
1

2
3

4

The Eighth Federal Reserve District covers all or parts
of Arkansas, Illinois, Indiana, Kentucky, Mississippi,
Missouri and Tennessee.
See Gascon and Morris.
We use prediction error as our performance metric,
which we define to be the absolute value of the difference between the growth rates of the initial release or
estimate and the revised data.
For more information on the early benchmarking
process, see the Federal Reserve Bank of Dallas.

REFERENCES
Gascon, Charles; and Morris, Paul. Employment Growth
in the Eighth District Appears Weaker Than Currently
Reported. On the Economy (a blog), Dec. 21, 2017. See
https://www.stlouisfed.org/on-the-economy/2017/
december/employment-growth-eighth-districtappears-weaker-currently-reported.
Federal Reserve Bank of Dallas. Early Benchmarking:
How Early Benchmarking Improves the Accuracy of
Payroll Employment Data. DataBasics. See https://
www.dallasfed.org/research/basics/benchmark.aspx.

DISTRICT OVERVIEW
ILLINOIS

Gauging Debt Levels in the U.S.
and Eighth District

INDIANA

MISSOURI
KENTUCKY

By James D. Eubanks and Don E. Schlagenhauf

TENNESSEE
ARKANSAS

MISSISSIPPI

KEY TAKEAWAYS

I

n May, the Federal Reserve Bank of
New York released its latest version of
the Household Debt and Credit Report,
which reported data for the first quarter
of 2018. A key finding in this report is the
continued increase in household debt. In
fact, in the first quarter, nominal household debt reached $12.8 trillion, exceeding the prior peak of $12.7 trillion in
2008. Of course, shortly after the peak in
2008, a period of deleveraging occurred
amid the Great Recession.
In this article, we look more closely at
the recent developments in household
debt accumulation nationally and in the
Eighth District.1 One of the key findings
is that household debt is increasing, but it
has not yet reached the level observed in
2008 if adjustments are made for inflation. Also, the cause of the debt run-up
in 2008 was mortgage debt. By contrast,
consumer credit card debt and auto debt
are the key drivers in the more recent
increase.
The Household Debt and Credit Report
(HDCR) is based on an anonymized
5 percent sample of credit files assembled
from data provided by the credit monitoring company Equifax. This data set
is named the Equifax/Federal Reserve
Bank of New York Consumer Credit
Panel (CCP). We use this data to examine

credit developments in the United States
as well as in the Eighth District.2
While we use the same data as in the
HDCR, we make two adjustments. First,
we use a different definition of total consumer debt. The New York Fed includes
student debt; in this research, total consumer debt does not include student debt.
The reason for the exclusion of individual
student debt data is that this data was not
consistently reported prior to 2006.
Second, we express all debt in inflationadjusted values, whereas the HDCR
reports nominal values of debt. We use
the personal consumption expenditures
(PCE) chain-type price index to adjust for
inflation. In addition, we normalize each
series so that the value is equal to 100 in
the first quarter of 2003.
Figure 1 shows total inflation-adjusted
consumer debt for the U.S. and the Eighth
District; the Great Recession (December
2007 to June 2009) is highlighted by the
gray bar.
Clearly, consumer debt was increasing
rapidly before it peaked during the recession. It is also clear that the run-up in
consumer debt in the Eighth District was
much smaller than in the entire economy.
The explanation lies with mortgage debt.
The housing sector boom in the Eighth
District was much smaller compared to

The Eighth Federal Reserve District is composed of four
zones, each of which is centered around one of the four
main cities: Little Rock, Louisville, Memphis and St. Louis.

Figure 1

The Amount of Real Consumer Debt
Has Yet to Peak
150
United States

Index, 2003:Q1=100

• In nominal terms, U.S. household debt continues to grow, exceeding a peak set in 2008.
Is this cause for concern?
• In real terms, debt isn’t at a new high. And the recent increase has been modest
compared to the debt run-up before the Great Recession.
• Though the rate of serious delinquency is rising in both the U.S. and Eighth District,
the increase doesn’t appear to be troublesome.

140
Eighth District
130
120
110
100
2003

2008

2013

2018

SOURCES: Equifax/Federal Reserve Bank of
New York Consumer Credit Panel and authors’
calculations.
NOTES: Consumer debt, which excludes student
loans, has been adjusted by the personal consumption expenditures (PCE) price index. The shaded area
represents the 2007-2009 recession.

ABOUT THE AUTHORS
Don E. Schlagenhauf (left) is an economist at the Federal Reserve Bank of St. Louis.
His research focuses on macroeconomics and policy, with emphasis on housing. He joined
the St. Louis Fed in 2017. Read more about the author and his research at https://research.
stlouisfed.org/econ/schlagenhauf.
James D. Eubanks is a senior research associate at the Federal Reserve Bank of St. Louis.

REGIONAL ECONOMIST | www.stlouisfed.org 19

that of other regions. As a result, the
increase in mortgage debt was smaller.
Since the latter part of 2013, total real
consumer debt has increased once again,
but this growth has been modest compared to the increase observed prior to
the Great Recession.
Is a Debt Crisis on the Horizon?

©THINKSTOCK/iSTOCK/VADIMGUZHVA

Figure 2

Real Consumer Debt by Category
Home Equity Line of Credit

Auto Debt
155

Eighth District

280

United States

260

Index, 2003:Q1=100

Index, 2003:Q1=100

145
135
125
115
105

220
200
180
160
140
120

85

100

2008

2013

2003

2018

Mortgage
170

Eighth District

United States

110

Index, 2003:Q1=100

Index, 2003:Q1=100

2008

2013

2018

Credit Card Debt
115

160
150
140
130
120
110

Eighth District

United States

105
100
95
90

The Delinquency Problem

85
80
75
70

100
2003

United States

240

95

2003

Eighth District

65
2008

2013

2018

2003

2008

2013

2018

SOURCES: Equifax/Federal Reserve Bank of New York Consumer Credit Panel and authors’ calculations.
NOTES: The debt value has been adjusted by the personal consumption expenditures (PCE) price index.
The shaded area represents the 2007-2009 recession.
20 REGIONAL ECONOMIST | Second Quarter 2018

The fact that real consumer debt is
increasing in both the entire economy
and the Eighth District leads to the
question: Is another debt crisis on the
horizon? In order to answer this question, we refer to Figure 2, where the
major categories of consumer debt—auto
debt, consumer credit card debt, home
equity line of credit (HELOC) debt 3 and
mortgage debt—are reported.
As can be seen, the real value of mortgage debt has increased very slowly compared to the pre-Great Recession trend. In
the Eighth District, the trends mirror the
national trends. Specifically, HELOC debt
has continued to decline. The real value
of auto debt has increased since 2012,
while consumer credit card debt has been
generally increasing since 2014.
In the first quarter of 2018, the latest
quarter for which data is available, consumer credit card debt increased nationally by 4.5 percent from the first quarter
of 2017 and rose 3.1 percent in the Eighth
District. In both cases, this increase
is partially explained by the effect of
the prior holiday season. Meanwhile,
mortgage debt increased by 2.4 percent
nationally; in the District, mortgage debt
essentially did not change.
Auto debt increased both nationally
and in the District, though growth has
been easing since the second quarter of
2017. In contrast, HELOC debt declined
nationally by 4.4 percent and by 1.4 percent in the District.

The increase in both credit card and
auto debt has raised concerns in the
popular press. No doubt these concerns
are partially the result of memories of the
role that mortgage debt played in causing
the Great Recession.
Yet, debt is only a problem if it is
defaulted upon. One approach to gauge
the risk of future defaults is to examine the
serious-delinquency rate, which is defined

as the share of debt that is past-due 90 days
or more. In the Eighth District, delinquency
rates did increase. However, the year-overyear increase was only 0.5 percentage points
for auto debt and 0.4 percentage points for
credit card debt. Changes in delinquency
rates were of similar size nationally.
These increases are not troublesome.
Yet, some may argue that these increases
may signal future increases in delinquency
rates, especially for auto debt. Their
reasoning is that the auto industry, like
the housing industry prior to the Great
Recession, is relying on subprime lending

to further increase auto sales. Our data set,
however, allows us to gain some insight into
this argument as credit scores are available
for the set of individuals in our sample.
We identify a subprime debt as an
individual with a credit score in the 280619 range. In the first quarter of 2018, the
most recent data available, we found that
the total real value of new auto loans to
subprime borrowers dropped by 10.1 percent year-over-year in the Eighth District,
while they increased by 2.4 percent in the
U.S. If the declining pattern continues,
the subprime concern over how auto sales

are being financed should be lessened in
the Eighth District.
(This article was published online June 18.)

ENDNOTES
1

2

3

Headquartered in St. Louis, the Eighth Federal
Reserve District includes all of Arkansas and parts of
Illinois, Indiana, Kentucky, Mississippi, Missouri and
Tennessee.
Later this year, we will write a companion piece that
looks at developments in the major metropolitan
statistical areas within the Eighth District.
HELOC debt is simply borrowing on the equity a
homeowner has accrued on the home owned.

Financial Health of Millennials
(continued from Page 15)
3

4

5

6

In addition to average asset and liability positions, we
also compared median asset and liability positions
across generations. The results are qualitatively
similar to the averages. However, the median levels
of housing assets, retirement account balances and
mortgage debt were zero, making comparisons
infeasible. For example, more than half of millennials
had no housing assets.
The dollar numbers reported in the SCF data are
inflation-adjusted to 2016 dollars and therefore can
be compared directly. In this article, dollar amounts
of $20,000 and greater have been rounded to the
nearest $1,000; those lower than $20,000 have been
rounded to the nearest $100.
As always, we cannot rule out that other underlying
factors could possibly bias the results shown in both
Figures 1 and 2. However, we see similar patterns
across both figures, implying a consistent story that
millennials hold lower levels of assets and have lower
net worth than Generation X on average.
The labor force participation rate data are from the
Bureau of Labor Statistics, and the demographic data
are from the Census Bureau.

The Dem
ogra
of Wealthphics
2018 Seri
es
The Demographics
How Educ
ation
of Wealth
Shape Fin , Race and Birth
Ye
Essay No.

ancial Ou
tcomes

2018 Series

ar

2: A Los
t Genera
tion? Lon
of the Gre
g-Lasting
at Recess
ion on You
Wealth Imp
ng Familie
acts
s | May
2018
How Education, Race and Birth
Year

Shape Financial Outcomes
Essay No. 2: A Lost Generation? Long-Lasting Wealth Impacts
of the Great Recession on Young Families | May 2018

Cryptocurrencies
and Fintech
Can’t make head or tail of
Bitcoin? Wonder what block-

REFERENCES

chain is? Check out a new

Hudson, E.K. Here’s Why Millennials Aren’t Buying Houses.
The Real Deal, Sept. 24, 2017. See https://therealdeal.
com/2017/09/24/heres-why-millennials-arent-buyinghouses/.
Lebowitz, Shana; and Shontell, Alyson. Millennials Aren’t
Buying Starter Homes—They’re Splurging on MillionDollar Places Instead. Business Insider, Oct. 31, 2017. See
www.businessinsider.com/millennials-dont-buy-starterhomes-2017-10.
O’Shea, Arielle. How Millennials Got a 6-Figure Start on
Retirement Saving. NerdWallet, Sept. 11, 2017. See www.
nerdwallet.com/blog/investing/the-key-to-six-figuresavings/.
Zumbrun, Josh. Younger Generation Faces a Savings
Deficit. The Wall Street Journal, Nov. 9, 2014. See www.
wsj.com/articles/savings-turn-negative-for-youngergeneration-1415572405.

webpage offered by the
Research Division of the
St. Louis Fed. There you can
read more about cryptocurrencies, blockchain and their
possible impact.
https://research.stlouisfed.org/
publications/cryptocurrenciesand-fintech/

DOOMED TO BE A
"LOST GENERATION"?
You didn't choose your birth
year, but it could affect your
wealth. The Great Recession
hit groups born in certain
decades harder than others:
Was yours among them?
Find out by reading the
latest Demographics
of Wealth essay, www.
stlouisfed.org/householdfinancial-stability/thedemographics-of-wealth.

REGIONAL ECONOMIST | www.stlouisfed.org 21

NATIONAL OVERVIEW

U.S. Economic Growth Appears
Solid This Year
By Kevin L. Kliesen
©THINKSTOCK/iSTOCK/Worawee Meepian

KEY TAKEAWAYS
• Professional forecasters expect the
U.S. economy to sustain above-trend
growth, perhaps close to 3 percent,
for the remainder of 2018.
• Business fixed investment continued
to grow at a healthy pace in the first
quarter, but growth in household
spending slowed.
• Real gross domestic product is likely
to grow 2.5 to 3.0 percent for 2018
and 2019, while the unemployment
rate will probably decline further.

T

he U.S. economy continues to expand
at a rate that exceeds its potential
rate of growth (somewhere around 2.25
percent). The consensus of professional
forecasters is that above-trend growth,
perhaps close to 3 percent, will persist
over the remainder of 2018, likely leading to further reductions in the unemployment rate. However, inflation has
rebounded modestly and is now at the
2 percent target rate of the Federal Open
Market Committee (FOMC). Overall,
the near-term outlook for the national
economy appears solid.
Healthy Growth but Larger
Budget Deficits
After increasing at a 3 percent annual
rate over the second half of 2017, real gross
domestic product (GDP) advanced at a 2.3
percent annual rate in the first quarter. Of
note, business fixed investment continued
to expand at a healthy rate, increasing at a
rate of more than 6 percent for the fourth
time in the past five quarters. Surveys of
large and small businesses continue to
show considerable optimism as a result
of this year’s tax cuts and additional
increase in federal government outlays.
In response, most economists expected
22 REGIONAL ECONOMIST | Second Quarter 2018

Year-over-Year Percentage Change

Headline Personal Consumption Expenditures (PCE) Inflation: Actual and Forecast
3.0
Actual

Forecast

2% Target

2.5
2.0
1.5
1.0
0.5
0.0
Jan. ’12

Jan. ’13

Jan. ’14

Jan. ’15

Jan. ’16

Jan. ’17

Jan. ’18

Jan. ’19

SOURCES: Bureau of Economic Analysis, along with Laura Jackson, Kevin Kliesen and Michael Owyang,
the authors of the article in the notes below.
NOTES: The above forecast was calculated on April 25, 2018, before the release of actual PCE inflation
for March. The forecast is produced from the St. Louis Fed’s FAVAR model. For more details on the
model, see “A Measure of Price Pressures,” Federal Reserve Bank of St. Louis Review, First Quarter 2015,
Vol. 97, No. 1, pp. 25-52.

continued brisk growth of real personal
consumption expenditures (PCE), which
had advanced at a 3.1 percent annual
rate over the second half of 2017. However, real PCE growth slowed sharply in
the first quarter to the smallest increase
in about five years (1.1 percent). This
development has led some forecasters to
wonder whether the expected boost to
household expenditures from the tax cuts
was too optimistic.
A key question in the outlook is
whether the modest slowing in firstquarter consumption growth is temporary or whether it points to something
more persistent. Overall, fundamentals
suggest real PCE growth will probably rebound over the remainder of
the year. Importantly, the demand for
labor remains strong, helping to fuel
solid gains in labor compensation. For

example, the National Association for
Business Economics (NABE) Business
Conditions Survey released in May indicated that the wages and salaries net rising
index (percentage of firms reporting rising
wages less percentage reporting falling
wages) posted its highest level on record
(back to 1982). Indeed, there were signs
of a spring thaw in consumption spending, as real PCE increased at a robust
5.1 percent annual rate in March. With
consumer spending likely to rebound,
coupled with the still-healthy outlook in
construction spending and some strengthening in U.S. exports, forecasters generally believe that real GDP growth will
be modestly stronger over the final three
quarters of the year.
Professional forecasters also expect
some bump in government expenditures
this year and the next from the Bipartisan

ABOUT THE AUTHOR
Kevin L. Kliesen is a business economist and research officer at the Federal Reserve Bank
of St. Louis. His research interests include business economics, and monetary and fiscal
policy analysis. He joined the St. Louis Fed in 1988. Read more about the author and his
research at https://research.stlouisfed.org/econ/kliesen.

ECONOMY AT A GLANCE
All data as of May 11, 2018

Real GDP Growth

0

–2
’13

(This article was published online May 24.)

Q1
’14

’15

’16

’17

Percent Change from a Year Earlier

2

CPI–All Items
All Items, Less Food and Energy

2

0

April

–2
’13

’18

’14

’15

’16

’17

’18

NOTE: Each bar is a one-quarter growth rate (annualized);
the red line is the 10-year growth rate.

Rates on Federal Funds Futures on Selected Dates

Inflation-Indexed Treasury Yield Spreads
2.50

2.20
10-Year

2.25

2.00

2.00

1.80

1.75
1.50

03/21/18
05/02/18

11/01/17
12/13/17
01/31/18

20-Year

Percent

Percent

5-Year

1.60
1.40

1.25

1.20
May 11, 2018

1.00
’14

’15

’16

’17

1.00

’18

1st-Expiring
Contract

NOTE: Weekly data.

3-Month

6-Month

12-Month

Contract Settlement Month

Civilian Unemployment Rate

Interest Rates

8

4
10-Year Treasury
3

6

Percent

Percent

7

5

2
Fed Funds Target

1

4
3
’13

April
’14

’15

’16

’17

1-Year Treasury
April

0

’18

’13

’14

’15

’16

’17

’18

NOTE: On Dec. 16, 2015, the FOMC set a target range for the
federal funds rate of 0.25 to 0.5 percent. The observations
plotted since then are the midpoint of the range.

U.S. Agricultural Trade
90

Average Land Values Across the Eighth District
15.0
Year-Over-Year Percent Change

Exports

Billions of Dollars

75
60
Imports

45
30
15
0

Brian Levine, a research associate at the Bank,
provided research assistance.

4

4

Inflation Developments
All else equal, higher inflation will
also increase interest rates. Inflation
was stronger than expected in the first
quarter. After increasing by 1.7 percent
in December 2017 from a year earlier,
the all-items PCE price index rose to
2 percent in March 2018 from a year
earlier. Some of this firming reflects the
recent run-up in crude oil prices, which
have more than doubled since their
recent low of just under $31 per barrel in
February 2016. In addition, firms appear
to be experiencing sizable materials cost
pressures, as noted in the aforementioned
NABE survey. With little discernible
erosion in profit margins thus far in 2018,
this finding suggests that many firms have
begun passing along a portion of these
increased costs to consumers.
Although the St. Louis Fed’s inflation
forecasting model projects a modest
further increase in inflation over the near
term, to about 2.25 percent in early 2019,
the model continues to indicate a small
probability (less than 10 percent) that
headline inflation will accelerate past
2.5 percent over the next 12 months.
At this point, the most likely outcome
for 2018 and 2019 is real GDP growth of
between 2.5 percent and 3 percent, inflation close to the Fed’s 2 percent target—
with perhaps some modest risk of an
overshoot if crude oil prices continue
to trend higher—and further declines
in the unemployment rate.

Consumer Price Index (CPI)

6

Percent

Budget Act that was signed into law in February. However, the trade-off from lower
taxes and increased government spending
is an expected erosion in the budget outlook. In April, the Congressional Budget
Office (CBO) released its Budget and Economic Outlook. In nominal dollar terms,
the CBO projects that the federal budget
deficit will average $1.2 trillion per year
from fiscal years 2019 to 2028 (roughly 5
percent of nominal GDP). Larger budget
deficits, stronger economic growth and
the FOMC’s plan to reduce the size of its
balance sheet all suggest rising long-term
interest rates over the near term.

March

Trade Balance
’13

’14

’15

’16

’17

NOTE: Data are aggregated over the past 12 months.

’18

12.5
10.0

Quality Farmland
Ranchland or Pastureland

7.5
5.0
2.5
0.0
–2.5

2017:Q1

2017:Q2

2017:Q3

2017:Q4

2018:Q1

SOURCE: Agricultural Finance Monitor.

On the web version of this issue, 11 more charts are available, with much of those charts’ data specific to the Eighth District. Among the
areas they cover are agriculture, commercial banking, housing permits, income and jobs. To see those charts, go to www.stlouisfed.org/
economyataglance.
REGIONAL ECONOMIST | www.stlouisfed.org 23

P.O. Box 442
St. Louis, MO 63166-0442
Change Service Requested

N E X T

I S S U E

Understanding the Trade Deficit

S

eeking to bolster the country’s
economic fortunes, U.S. policy-

makers have recently focused on
reducing the trade deficit with China.
But is the trade imbalance the cause
of job loss in America or a symptom
of deeper economic changes? We
take a closer look at the U.S. trade
deficit with China.

printed on recycled paper using 10% postconsumer waste
PHOTO CREDIT: ©THINKSTOCK/ISTOCK/hxdbzxy

ECONOMY AT A GLANCE
All data as of May 11, 2018.
SECOND QUARTER 2018

Real GDP Growth

Percent Change from a Year Earlier

4

4
Percent

VOL. 26, NO. 2

Consumer Price Index (CPI)

6

2

0

–2
’13

|

Q1
’14

’15

’16

’17

CPI–All Items
All Items, Less Food and Energy

2

0

April

–2
’13

’18

’14

’15

’16

’17

’18

NOTE: Each bar is a one-quarter growth rate (annualized);
the red line is the 10-year growth rate.

Inflation-Indexed Treasury Yield Spreads

Rates on Federal Funds Futures on Selected Dates
2.20

2.50
10-Year

20-Year

2.25

2.00

2.00

1.80
Percent

Percent

5-Year

1.75

1.60

1.50

1.40

1.25

1.20

May 11, 2018

1.00
’14

’15

’16

’17

1.00
1st-Expiring
Contract

’18

NOTE: Weekly data.

6-Month

12-Month

Interest Rates

8

4
10-Year Treasury

7

3

6

Percent

Percent

3-Month

Contract Settlement Month

Civilian Unemployment Rate

5

2
Fed Funds Target

1

4
3
’13

03/21/18
05/02/18

11/01/17
12/13/17
01/31/18

April
’14

’15

’16

’17

1-Year Treasury
April

0

’18

’13

’14

’15

’16

’17

’18

NOTE: On Dec. 16, 2015, the FOMC set a target range for the
federal funds rate of 0.25 to 0.5 percent. The observations
plotted since then are the midpoint of the range.

U.S. Agricultural Trade
90

Average Land Values across the Eighth District
15.0
Year-Over-Year Percent Change

Exports

Billions of Dollars

75
60
Imports

45
30
15
0

March

Trade Balance
’13

’14

’15

’16

’17

NOTE: Data are aggregated over the past 12 months.

’18

12.5
10.0

Quality Farmland
Ranchland or Pastureland

7.5
5.0
2.5
0.0
–2.5

2017:Q1

2017:Q2

2017:Q3

2017:Q4

SOURCE: Agricultural Finance Monitor.

2018:Q1

U.S. Crop and Livestock Prices
140

Index 1990-92=100

120

Crops
Livestock

100
80
60
40
’03

March
’04

’05

’06

’07

’08

’09

’10

’11

’12

’13

’14

’15

’16

’17

’18

COMMERCIAL BANK PERFORMANCE RATIOS

U.S. Banks by Asset Size/First Quarter 2018
$300 millionLess than
$1 billion
$300 million

Less than
$1 billion

$1 billion$15 billion

Less than
$15 billion

More than
$15 billion

1.20

1.17

1.29

1.24

1.27

3.88

3.86

3.87

3.86

3.87

3.10

1.00

0.84

0.91

0.88

0.83

0.86

1.18

1.35

1.36

1.29

1.32

1.10

1.18

1.25

All

$100 million­$300 million

Return on Average Assets*

1.26

1.16

1.12

Net Interest Margin*

3.23

3.89

Nonperforming Loan Ratio

1.11

Loan Loss Reserve Ratio

1.23

Return on Average Assets*

Net Interest Margin*
1.32

1.13

1.32

1.59

1.19

1.09
1.06
1.15
1.02

First Quarter 2018

1.00

Indiana
1.34

1.15

0.75

1.27

Mississippi

1.28

Missouri
Tennessee

1.25

1.50

1.75

Percent

0.96
0.91
0.93

0.71

Illinois

1.08

0.75

0.72
1.20
1.20

Kentucky
0.90

Mississippi

1.04
1.23
1.30

Missouri

0.82

1.00

0.79

Tennessee

0.94
1.25

First Quarter 2017

SOURCE: Federal Financial Institutions Examination Council
Reports of Condition and Income for all Insured U.S.
Commercial Banks.
NOTE: Data include only that portion of the state within
Eighth District boundaries.
*Annualized data.

0.61

Indiana

1.03

0.77

0.71

First Quarter 2018

Arkansas

0.79

0.62

1.03
1.13
1.05
1.09
1.09
1.17

Eighth District

0.90

0.69

0.50

First Quarter 2017

Loan Loss Reserve Ratio

0.70

0.25

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
First Quarter 2018

Nonperforming Loan Ratio

0.00

4.03
4.00
3.95
3.71
3.50
3.40
3.64
3.31

Kentucky

First Quarter 2017

0.60

3.56
3.45
3.63
3.59

Illinois

1.05

0.50

4.20
4.10

Arkansas

0.97
0.98

0.00 0.25

3.82
3.68

Eighth District

Percent

0.00 0.20

0.40

First Quarter 2018

0.60

1.10

0.80 1.00

1.20

First Quarter 2017

For additional banking and regional data, visit our
website at https://fred.stlouisfed.org.

1.40

REGIONAL ECONOMIC INDICATORS

Nonfarm Employment Growth/First Quarter 2018
Year-Over-Year Percent Change
United
States

Eighth
District †

Arkansas

Total Nonagricultural

1.5%

0.9%

0.4%

Natural Resources/Mining

9.0

–2.2

–5.7

Construction

3.9

0.4

Manufacturing

1.8

Trade/Transportation/Utilities
Information

Illinois

Indiana

Kentucky

0.6%

0.4%

–4.4

1.2

0.3

–0.5

1.3

2.8

1.4

1.4

3.2

0.9

0.6

–0.3

–1.9

–3.2

Financial Activities

1.6

Professional & Business Services

0.8%

Mississippi

Missouri

1.1%

Tennessee

1.2%

1.6%

–5.3

0.8

0.0

–1.4

1.9

–2.7

NA

0.9

0.3

0.4

1.8

0.3

–0.1

0.6

2.2

0.6

0.6

1.0

–8.2

–4.5

–9.4

–3.0

–5.4

2.0

0.3

1.8

1.6

2.4

0.4

–0.6

3.0

2.1

2.5

2.4

1.3

0.7

–0.2

1.6

–0.1

4.1

4.3

1.6

Educational & Health Services

2.0

1.3

1.0

0.6

2.2

0.3

3.0

1.7

1.9

Leisure & Hospitality

2.0

1.2

0.6

1.6

–0.8

–0.9

1.3

0.0

4.6

Other Services

1.6

0.4

0.5

–0.2

–0.9

2.5

0.8

–0.3

2.1

Government

0.0

0.3

0.3

1.0

–0.4

–0.3

–0.5

0.0

0.7

† Eighth District growth rates are calculated from the sums of the seven states. Each state’s data are for the entire state even though parts of six of
the states are not within the District’s borders.

Unemployment Rates

Eighth District Payroll Employment by Industry-2017

I/2018

IV/2017

I/2017

4.1%

4.1%

4.7%

United States

Professional and
Business Services
Financial
Activities

Educational and
Health Services
13.1%

15.0%

Leisure and
Hospitality

5.4%

Arkansas

3.8

3.7

3.6

Illinois

4.7

4.9

5.1

Indiana

3.2

3.4

3.7

Kentucky

4.1

4.5

5.2

Manufacturing

Mississippi

4.5

4.8

5.4

Construction 4.0%

Missouri

3.7

3.6

4.0

Tennessee

3.4

3.3

4.4

10.3%

Information
1.4%
19.9%
Trade
Transportation
Utilities

3.9%

Other Services

14.9%
11.7%

Government
Natural Resources
and Mining 0.2%

Housing Permits/First Quarter

Real Personal Income/Fourth Quarter

Year-Over-Year Percent Change in Year-to-Date Levels

Year-Over-Year Percent Change

6.8
11.1

United States
33.2

7.1
–19.8

Arkansas

–8.1

Illinois
10.2
13.5

–21.9

–50 –40 –30 –20 –10 0
2018

1.5

–1.2

Kentucky

1.0

6.2

1.2

–0.2

Missouri

1.3

–0.9

Tennessee
10

20

30 40 50 60

2017

NOTE: All data are seasonally adjusted unless otherwise noted.

Percent

1.3

0.7

–1.5

54.7 Mississippi
–10.2
–10.7
–12.0

1.8

0.0

Indiana
32.0

–41.6

2.2

–0.1

1.2
–2.0 –1.5 –1.0 –0.5 0.0
2017

0.5

1.0

1.5

1.8
2.0

2016

NOTE: Real personal income is personal income divided by the
personal consumption expenditures chained price index.

2.5