View original document

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

The Demographics
of Wealth
2018 Series
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

Executive Summary

About the Center for Household Financial Stability
The Center for Household Financial Stability at the Federal Reserve Bank of St. Louis focuses on family
balance sheets, especially those of struggling American families. The Center researches the determinants
of healthy family balance sheets, their links to the broader economy and new ideas to improve them.
The Center’s original research, publications and public events aim to impact future research, community
practice and public policy. For more information, see www.stlouisfed.org/hfs.

Staff
Ray Boshara is an assistant vice president at the St. Louis Fed and director of the Center.
He is also a senior fellow in the Financial Security Program at the Aspen Institute.
William R. Emmons is an assistant vice president and economist at the St. Louis Fed and the lead
economist with the Center.
Lowell R. Ricketts is the lead analyst for the Center.
Ana Hernández Kent is a policy analyst for the Center.

Research Fellows
Barry Z. Cynamon is a research associate at the Weidenbaum Center at Washington University in St. Louis.
Michael Stegman, senior research fellow, also holds senior fellow positions at the Milken Institute Center for
Financial Markets, and the Center for Community Capital at the University of North Carolina.

Visiting Scholars
Fenaba R. Addo is an assistant professor of consumer science at the University of Wisconsin-Madison.
Emily Gallagher is an assistant professor of finance and real estate at the University of Colorado at Boulder.
Bradley L. Hardy is an associate professor of public administration and policy at the American University in
Washington, D.C., and a nonresident senior fellow in economic studies at the Brookings Institution.

2 Federal Reserve Bank of St. Louis

Authors
William R. Emmons is the lead economist with the Center for Household Financial Stability at the Federal Reserve Bank of St. Louis, where
he also serves as assistant vice president. His areas of focus at the Center
include household balance sheets and their relationship to the broader
economy. He also speaks and writes frequently on banking, financial
markets, financial regulation, housing, the economy, and other topics.
His work has been highlighted in major publications including The New
York Times, The Wall Street Journal and American Banker, and he has
appeared on PBS NewsHour, Bloomberg News, and other national
programs. Emmons received a Ph.D. in finance from the Kellogg School
of Management at Northwestern University. He received his bachelor’s
and master’s degrees from the University of Illinois at Urbana-Champaign.

Ana Hernández Kent is a policy analyst for the Center for Household
Financial Stability at the Federal Reserve Bank of St. Louis. She conducts
primary and secondary research and data analysis on household balance
sheet issues. Her primary research interests at the Fed include economic
disparities and opportunity, wealth outcomes, class and racial biases, and
the role of psychological factors in making financial decisions.
Kent received her Ph.D. in experimental psychology with concentrations
in social psychology and quantitative methods in behavioral sciences
from Saint Louis University. Kent received her Master of Science in experimental psychology from Saint Louis University and her bachelor’s degree
in psychology from the University of Notre Dame.

Lowell R. Ricketts is the lead analyst for the Center for Household Financial Stability at the Federal Reserve Bank of St. Louis, where he conducts
primary and secondary research and policy analysis on household balance sheet issues. His primary research focus has centered on household
liabilities and wealth outcomes. Prior to joining the team, he worked in
the Research division of the Federal Reserve Bank of St. Louis as a senior
research associate. Ricketts received a bachelor’s degree in economics
with a math emphasis from the University of Wisconsin-Madison. He
continues to be involved with the university’s Department of Economics
as a member of the Wisconsin Economics Young Alumni Council.

The Demographics of Wealth 3

An Introduction to the Series

The Demographics of Wealth
How Education, Race and Birth Year
Shape Financial Outcomes
By William R. Emmons, Ana H. Kent and Lowell R. Ricketts

I

ncome and wealth rebounded for many
families between 2013 and 2016, the dates
of the two most recent waves of the Federal
Reserve’s Survey of Consumer Finances
(SCF). Groups that had struggled the most
during and after the Great Recession, including less-educated, Hispanic and black, and
young families, participated in the recovery.
Nonetheless, long-standing income and
wealth gaps across education levels, races
and ethnicities, and age groups remain large.
This is the second in a series of new
essays that the Center for Household Financial Stability is publishing on how a family’s
demographic characteristics—including
educational attainment, race and ethnicity,
and birth year—are related to the family’s
financial outcomes. Like the previous essay
series published in 2015, the 2018 series will
focus on these three key demographic
dimensions in turn. An important new
feature of the 2018 series is the inclusion of
two generations of educational data for each
family. In addition to the educational attainment of the SCF respondent, the 2016 SCF for
the first time contains detailed information
on the respondents’ parents’ education. This
new information reveals even more clearly
that inherited demographic characteristics—
your race or ethnicity, your age and birth

4 Federal Reserve Bank of St. Louis

year, and even your parents’ level of education—profoundly shape the economic and
financial opportunities you have and the
outcomes you achieve.
As before, our primary data source is
the triennial SCF, which provides the most
comprehensive picture available of American
families’ balance sheets and financial behavior over time. In some of our analyses, we
use information from 47,776 families, each
of which was surveyed in one of 10 survey
waves between 1989 and 2016. When we
focus on the education of SCF respondents’
parents, we draw upon data collected from
6,248 families in 2016. In every case, the SCF
has been designed to be nationally representative, so we can safely generalize about the
population as a whole.
As we documented three years ago,
demographic characteristics remain
remarkably powerful in predicting a family’s
income and wealth. By expanding the scope
of inherited demographic characteristics to
include parents’ education, we believe the
2018 Demographics of Wealth series sheds
additional light on the deeply rooted sources
of economic and financial disparities. Fruitful
approaches to policy should be based on the
facts established here.

Executive Summary of Essay No. 2

T

his essay explores the connections between a
person’s birth year and measures of his or her
family’s financial well-being, including income and
wealth. We found that wealth losses occurred across
the age spectrum around the Great Recession but
that families younger than retirement age suffered
the most and have rebounded slowly. Based on
data from nearly 48,000 families born throughout
the 20th century, we found that families headed by
someone born in 1960 or later were less likely to
have recovered by 2016 than older families.
We focused on six groups of families based
on the birth decade—from the 1930s through the
1980s—of the family heads. We chose these decadal
cohorts because they were the only ones whose
typical (situated in the middle) family head was
between 24 and 80 years old both before and after
the financial crisis of 2008-09. We compared the
median inflation-adjusted wealth of these groups to
predicted levels achieved at various ages based on
data from all families responding to the Survey of
Consumer Finances between 1989 and 2016.
Our examination of the links between birth year
and wealth revealed three important findings:
• There is a pronounced life cycle of wealth. The
typical family’s wealth traces out an upward sloping
arc over most of its life cycle, beginning around
zero in the early 20s and reaching a peak of about
$228,000 at age 72. The range of actual wealth
accumulation across families is very large, but the
typical family’s experience is well-described as
rapid initial growth in percentage terms followed
by steady deceleration and eventual decline—albeit
slight—throughout the rest of the life cycle. The
shape of the wealth life cycle is influenced by economic and financial developments over time.

• Members of all birth cohorts lost wealth around
the Great Recession, but only typical families
headed by someone born in 1960 or later had
failed to get back on track by 2016. Median wealth
levels of all six decadal cohorts we studied were
comfortably above their respective age-specific
wealth benchmarks in 2007. The Great Recession
reduced median wealth substantially among all six
groups. The four youngest cohorts (1950s and later)
dropped below their age-specific wealth benchmarks. The three youngest cohorts (1960s and later)
remained below those benchmarks in 2016.
• The 1980s cohort is at greatest risk of becoming
a “lost generation” for wealth accumulation.
Wealth in 2016 of the median family headed by
someone born in the 1980s remained 34 percent
below the level we predicted based on the experience of earlier generations at the same age. The
corresponding shortfalls of the 1960s cohort (–11
percent as of 2016) and the 1970s cohort (–18 percent) are worrying but are much smaller than their
respective 2010 and 2013 shortfalls. Alone among
the six decadal cohorts we studied, the typical 1980s
family lost ground between 2010 and 2016, falling
even further behind the typical wealth life cycle.
This represents a missed opportunity because asset
appreciation is unlikely to be as rapid in the near future as it was during the recent period. Two reasons
for optimism are that the 1980s cohort has many
years to get back on track and it is the most educated—hence, also potentially the highest-earning—
group ever.

The Demographics of Wealth 5

Essay No. 2

A Lost Generation?
Long-Lasting Wealth Impacts of
the Great Recession on Young Families
By William R. Emmons, Ana H. Kent and Lowell R. Ricketts

6 Federal Reserve Bank of St. Louis

Figure 1: Median Family Net Worth and Income
160

Thousands of 2016 $

140

Median Net Worth

140

Median Income

120

97

100

85

84

80
55

60

51

53

48

40
20
0

2007

2010
2013
Survey Year

2016

NOTE: See Sidebar 1 for more details on how income and net worth are
measured in the Survey of Consumer Finances.

Figure 2: Change in Median Net Worth,
Relative to 2007
0
–5

–6

–10
Percent Change

T

he Great Recession of 2008-09 inflicted deep
and widespread losses of income and wealth
on the typical American family, leaving both measures lower in 2016 than they were in 2007.1 (See
Figure 1.) Wealth losses occurred across the age
spectrum, but families younger than retirement
age suffered the most and have rebounded slowly.2
(See Figure 2.) In contrast, typical incomes dropped
much less than wealth among young (under 40)
and middle-aged families (40-61). In the case of
older families (62 and older), typical incomes never
declined below the 2007 level.3 (See Figure 3.)
The fact that many families suffered large
wealth setbacks during their prime earning and
wealth-accumulation years raises the question of
whether they will be able to rebuild their wealth
to meet major saving goals, including for a home
purchase, college tuition for their children and
retirement. Will the typical family that was young
or middle-aged at the time of the Great Recession
become part of a “lost generation” that struggles to
achieve life cycle milestones?
To judge whether particular birth cohorts—
that is, groups of families whose heads were born
during the same decade—are on track to meet
their wealth-accumulation targets, we estimated
typical life cycle wealth trajectories using data collected from 1989 to 2016. In other words, how much
wealth would we expect a typical family to have at
each age? We then compared the actual wealth levels
for six groups of families—those headed by someone
born in the 1930s, the 1940s, the 1950s, the 1960s, the
1970s and the 1980s—to these life cycle benchmarks.
Using data from 47,776 families in the Survey of
Consumer Finances (SCF) between 1989 and 2016,

–9

–15
–17

–20
–25

–27

–30
–35
–40
–45
–50

–36

–37

–37

–43
2010
Young (<40)

–47
2013

2016

Middle-aged (40-61)

Old (62+)

NOTE: Each age group’s median net worth in 2010, 2013 and 2016 was
compared to the same age group’s respective level in 2007.

Figure 3: Change in Median Income,
Relative to 2007

Table 1: Worst Wealth Shortfalls in a Single Year

30

Birth
Decade

Percentage from
Age-Specific
Benchmark

1

2013

39

1970s

–42.7

24

25
20
Percent Change

Rank

Year
Observed

Average
Age of
Family
Head

15

9

10

2

2013

29

1980s

–41.0

5

3

2010

35

1970s

–35.1

0

4

2016

32

1980s

–34.1

7

–5

–6

–10

–11 –11

–15

–10

–12

46

1960s

–29.1

2013

49

1960s

–27.5

7

2010

26

1980s

–24.7

2016

8

2016

42

1970s

–17.8

Middle-aged (40-61)

Old (62+)

9

2013

59

1950s

–15.5

10

1992

28

1960s

–14.9

2010
Young (<40)

2010

2013

–16

–20

5
6

NOTE: See note to Figure 2, replacing “net worth” with “income.”

NOTES: A wealth shortfall is the percent difference between the actual
median wealth for a birth cohort in a particular survey year and the
predicted wealth at that age based on data from all families in all survey
years of the SCF. All figures were adjusted for inflation. There were 50
cohort-year observations.

we found that typical families headed by someone
born in the 1960s, 1970s and 1980s were significantly below their wealth benchmark levels in
2016—by about 11, 18 and 34 percent, respectively.
Despite also having suffered wealth losses during
the recession, typical families headed by someone born in the 1930s, 1940s or 1950s were slightly
above their age-specific wealth benchmarks in
2016. (See Figure 4.)
Can the cohorts born in 1960 or later get back
on track? The younger the group, the more uncertain long-range wealth predictions must be. Nonetheless, we believe many families in the youngest
cohort we studied here—respondents born in the
1980s—are at substantial risk of accumulating less

wealth over their life spans than the members
of previous generations. Not only is their wealth
shortfall in 2016 very large in percentage terms,
but the typical 1980s family actually lost ground in
relative terms between 2010 and 2016, a period of
rapidly rising asset values that buoyed the wealth of
all older cohorts. (See Table 1 for a list of the worst
wealth shortfalls experienced by any birth cohort in
a single survey year.)
In common with the 1970s cohort—which, as of
2016, ranked as the second most-at-risk cohort for

Figure 4: Deviation of Median Wealth from Predicted Value
80
61

60

Percent

40
20
0

1989

56
33

31
9 5

17
1

1

13

4

2

6

2013

2016

4

–4

–11

–15
–29 –27

–40
1930-1939

2010

20
2 4

–20

–60

2007

1940-1949

1950-1959

1960-1969

–18
–35

–25

–34

–43

–41

1970-1979

1980-1989

Birth Cohort
NOTES: Predicted value was based on life cycle. For information on how median net worth was predicted, see Sidebar 2. Appendix 2 offers moretechnical details.
The sources for all the tables and figures are the Federal Reserve’s Survey of Consumer Finances and authors’ calculations.

The Demographics of Wealth 7

Sidebar 1: Family Income and Wealth
o measure income for the SCF, the interviewers
requested information on the family’s cash income,
before taxes, for the full calendar year preceding
the survey. The components of income in the SCF
are wages, self-employment and business income,
taxable and tax-exempt interest, dividends, realized
capital gains, food stamps and other related support
programs provided by government, pensions and
withdrawals from retirement accounts, Social Security,
alimony and other support payments, and miscellaneous sources of income for all members of the primary
economic unit in the household. All income figures
were adjusted for inflation to be comparable to values
recorded in 2016.

lifetime wealth underperformance, with a shortfall
of 18 percent—the typical 1980s family also carries
an extraordinarily high debt burden compared to
previous cohorts. This debt, only some of which
finances productive assets, increases the financial
fragility of many young families, too.
This essay has four parts. Section I describes the
life cycle of wealth and several broad changes in
typical wealth trajectories that have occurred since
1989. Section II identifies the birth-year cohorts hit
hardest by the Great Recession. Section III explores
the prospects for hard-hit cohorts to rebuild their
wealth in the coming years. Section IV concludes.
Four sidebars provide additional details on our data,
methodologies and related topics beyond the scope
of this essay. Two appendixes explain how we
chose the ages and birth years to study and provide technical details of how we estimated typical
wealth at each adult age.

I. The Life Cycle of Wealth
Young families typically have very little wealth.
In fact, if someone starts out adult life with student
loans or other debt, net worth even could be negative. (See Sidebar 1 for definitions of income and
wealth.) Near the end of the life cycle, families in
their early 70s typically have accumulated a significant amount of wealth before spending down
8 Federal Reserve Bank of St. Louis

Wealth is a family’s net worth, consisting of the
excess of its assets over its debts at a point in time.
Total assets include both financial assets (such as bank
accounts, mutual funds and securities) and tangible
assets (including real estate, vehicles and durable
goods). Total debt includes home-secured borrowing,
or mortgages, other secured borrowing (such as
vehicle loans) and unsecured debts (such as credit
cards and student loans). Debt incurred in association
with a privately owned business or to finance investment real estate is subtracted from the asset’s value,
rather than being included in the family’s debt. All
wealth figures were adjusted for inflation.

Figure 5a: Predicted Net Worth by
Age of Family Head: Ordinary Scale
250

Thousands of 2016 $

T

200
150
100
50
0

20

30

40
50
60
Age of Family Head

70

80

NOTES: See Sidebar 2 and Appendix 2 for more details regarding life
cycle predictions. The figures present the predicted median life cycle of
net worth using an ordinary scale (Figure 5a) or a log scale (Figure 5b)
for the y-axis.

some of it in retirement. Of course, typical wealth
life cycles differ somewhat depending on demographic factors such as education levels, and race
and ethnicity, which we discuss in more detail
elsewhere in this series.4
A plot of typical wealth levels at each age between
the beginning and end of adult life traces out the life
cycle of wealth—rising rapidly at first before peaking
in the early 70s.5 (See Figure 5a.) Transformation of
the vertical axis into a logarithmic (log) scale facilitates
visual comparisons at very different wealth levels.

Figure 5b: Predicted Net Worth by
Age of Family Head: Log Scale

Sidebar 2: Predicting Life Cycles

W

Thousands of 2016 $

1,000

e predicted a life cycle pattern for several
variables (family income and net worth,
cohort homeownership, saving and delinquency
rates) using data from all families across all waves
of the SCF. The life cycle is modeled as a function
of the family respondent’s age. For example, for
the life cycle of wealth, we looked at each family’s
wealth and its family head’s age; then we plotted
the line that best fit all the data. This helped us
predict median wealth across families at each age.
Importantly, because this relationship can vary
between time periods, we controlled for the year
in which respondents completed the survey. This
effectively removed the influence of time
period, and allowed us to look at a purer relationship between age and each variable. For moretechnical details of our estimation method, see
Appendix 2.

100

10

1

20

30

40
50
60
Age of Family Head

70

80

NOTES: In a semi-logarithmic chart like Figure 5b, equal vertical distances represent equal percentage differences. See Sidebar 3 for more
information about using log scales.

Figure 6: Differences from 1989 to 2016
in Predicted Wealth
80

1989-1998

60

2001-2007

2010-2016

40
20
Percent

(See Figure 5b.) See Sidebar 3 for a discussion of why
we use the log scale in some charts.
The flattening of the predicted wealth trajectory
in Figure 5b across the life cycle clearly shows that
a typical family’s wealth accumulation is most rapid
(in percentage terms) early in life. This reflects several
factors unique to early adulthood:
• Rising income, which makes saving easier;
• Development of regular saving habits, sometimes
including “forced saving” in the form of a monthly
mortgage payment;
• Better cash management and fewer delinquencies,
which are costly due to penalties, higher borrowing costs and less access to credit; and
• The larger impact of a dollar of additional saving
on (low) accumulated wealth.
The downward trend in wealth very late in life
is more gradual than a theoretical life cycle model
predicts: The dynamics of wealth at older ages are
complex. Spending down accumulated wealth
begins very late in life, if ever, for the typical family.
In this essay, we predicted wealth through age 80
but focused primarily on wealth before retirement
age. (See Sidebar 4 for a discussion of wealth at
advanced ages.)
Trends in the life cycle of wealth. Figure 6
depicts three distinct subperiods in which the
predicted wealth life cycle departed notably from its
long-run (1989-2016) average shape. (See Sidebar 2
and Appendix 2 for explanations of our estimation
methods.)

0
–20
–40
–60

20

30

40
50
60
Age of Family Head

70

80

NOTES: The lines show percent differences at each age between the
predicted median wealth level for a subperiod and the corresponding
predicted median wealth level using all sample data. The subperiods
include data from 1989, 1992, 1995 and 1998; 2001, 2004 and 2007; and
2010, 2013 and 2016, respectively. See Sidebar 2 and Appendix 2 for
more details on life cycle predictions.
The sources for all the tables and figures are the Federal Reserve’s Survey
of Consumer Finances and authors’ calculations.

The Demographics of Wealth 9

Sidebar 3: Charts with a Logarithmic
Vertical Axis

Figure 7: Change in Predicted Wealth
between 1989 and 2016
90

W

10 Federal Reserve Bank of St. Louis

70
50
Percent

e used a logarithmic (or log) scale in some
charts because wealth accumulation, like
other forms of growth, is exponential—that is,
it compounds over time. We often care about
growth rates or percentage differences rather than
the level of or absolute differences in wealth. The
log scale allowed us to illustrate growth rates
consistently. This is because a log scale straightens
the compound-growth curve, making a constant
growth rate appear as a straight line. If we did not
make this adjustment, it would be impossible to
compare growth rates or percentage differences at
different places in a chart.
When the vertical axis is a log scale, equal
vertical distances represent equal percentage
differences wherever they occur. In addition, the
slope of any line segment is proportional to the
rate of change between its endpoints—steeper
lines correspond to faster growth or larger
percentage differences. These features are
particularly useful when a chart represents a
broad range of values.
For example, we estimated that the typical
net worth of a family headed by someone who is
24 years old is about $5,072. The net worth of the
family of a typical 30-year-old is about $25,989,
or 412 percent more. The family of a typical
48-year-old has a net worth of $130,454, which is
402 percent more than the 30-year-old level.
On an unadjusted graph, the dollar difference
between the 24- and 30-year-olds’ net worth
would appear small—just $20,917. The dollar
difference between the 30- and 48-year-olds’ net
worth would appear large—about $104,465. But, in
fact, the percentage differences noted above are
essentially the same. On a log scale, the vertical
distances between the wealth of the typical
24- and 30-year-old and between the 30- and
48-year-old would be virtually identical.

30
10
0
–10
–30
–50

25

30

35

40

45 50 55 60
Age of Family Head

65

70

75

80

NOTE: The columns represent the percent change between 1989 and
2016 at each age. For example, predicted wealth at age 25 in 1989 was
$13,230; in 2016, it was $6,951—a decline of 47.5 percent.

The subperiods are:
• The pre-housing-bubble period (1989, 1992, 1995,
1998), when young and middle-aged families
typically had wealth above their long-term trend,
and old families were slightly below (blue line);
• The housing-bubble period (2001, 2004, 2007),
when all ages, but especially young families, had
wealth above their long-term trend (red line); and
• The post-Great Recession period (2010, 2013,
2016), when the typical wealth of old families
remained above their long-term trend, but the
wealth of middle-aged and especially young
families dropped below their respective longterm average levels (green line).
Figure 7 shows the percent change between
1989 and 2016 at each age in predicted wealth. The
changes are striking—almost a 50-percent decline
among the youngest families, while the very oldest
families enjoyed a 100-percent-plus increase.
The shifting life cycles of wealth. The patterns
depicted in Figure 6 provide two important insights
into the shifting fortunes of various age groups. First,
there are notable differences across subperiods
in predicted levels of wealth—unusually high in
the housing-bubble period and much lower at
other times. Second, there is a distinct steepening
of the relationship between age and wealth from
the pre-housing-bubble period to the post-Great
Recession period. On balance, wealth has shifted
away from younger families toward older families.
Figure 8 summarizes long-term cumulative
changes in predicted levels of wealth at several ages

Figure 8: Change in Estimated Age-Specific Wealth Levels since 1989
100
80

Percent Change

60

25

35

45

55

65

75

40
20
0
–20
–40
–60
–80
1989

1992

1995

1998

2001

2004

2007

2010

2013

2016

Survey Year
NOTE: Each age group’s predicted net worth at each SCF year was compared to the predicted level for the group of people who were that age in 1989.
For example, the predicted wealth at age 25 in 2010 was $4,504, which was 66 percent below the predicted wealth of a typical family headed by someone
age 25 in 1989.
The sources for all the tables and figures are the Federal Reserve’s Survey of Consumer Finances and authors’ calculations.

between 25 and 75. The typical 65- and 75-year-oldheaded families were richer in 2016 than in 1989,
while the reverse is true for all younger ages shown.
Together, Figures 6, 7 and 8 clearly show that the
life cycle of wealth has steepened dramatically,
with families aged 60 at the inflection point. Even
before 2007, the steepening trend was visible, but
the Great Recession and its aftermath significantly
widened the wealth gap between young and old.

II. The Hardest-Hit Generations
Comparisons over time of typical wealth life
cycles provide an informative look at the conditions
facing families at different life stages but provide little
direct insight into the fates of particular families as they
traverse their own life courses. A 25-year-old family
respondent in 1989, for example, was 52 in 2016; a
55-year-old family respondent in 1989 was 82 in 2016.
How do trends at specific ages affect individual families over time as they themselves reach different ages?
To understand how the members of particular
birth years have fared, we tracked six decadelong
cohorts over almost three decades—families headed
by people born in the 1930s, 1940s, 1950s, 1960s, 1970s
and 1980s. This method creates “quasi-panels” of families grouped by birth decade that are sampled in each
SCF wave between 1989 and 2016. To be clear, we
do not track individual families across time; instead,
each sample group is selected using birth year as the
sole criterion for inclusion in a decadal group.6 (See
Appendix 1 for details of our sample selection.)

Sidebar 4: Wealth in Old Age

T

he conventional life cycle model of spending and
saving taught in a beginning economics course
traces out a hump-shaped wealth trajectory that
peaks when a person enters retirement and then declines toward zero near the end of life. In fact, typical
life cycle wealth trajectories revealed by the SCF are
almost flat during retirement rather than declining
rapidly toward zero.
Rather than disproving the basic life cycle model,
this evidence reflects several factors unique to old
age that are not incorporated in the simplest model.
Some of the factors that explain why old families
typically do not spend all of their wealth include:
• Bequest motives: intentions to leave wealth to
heirs or charities;7
• Unpredictable lifespans and low rates of wealth
annuitization (the purchase of insurance contracts
that hedge against outliving your savings);8
• Concerns about medical expenses, including
gaps in health insurance coverage and uncertain
out-of-pocket expenditures;9
• Asset illiquidity (the difficulty of accessing wealth
in the form of housing equity or equity in a small
business or real estate); and
• Survivorship bias (the distortion in surveys like
the SCF resulting from the fact that people who
have survived to an old age are more likely to
have been well-off when younger than the average member of their birth cohort).10
The Demographics of Wealth 11

Thousands of 2016 $, Natural Log Scale

Figure 9a: Predicted vs. Actual Median Net Worth, Family Heads Born 1930-1959
1,000

= year 2007

100

10
Predicted

1930-1939

1940-1949

1950-1959

1
20

25

30

35

40

45

50

55

60

65

70

75

80

Age of Family Head
NOTES: See Appendix 1 for more details on how birth cohorts were constructed and how their actual observed wealth was estimated and matched to
predicted values.
In this and all subsequent figures, the observation for the year 2007 is highlighted for each cohort with a triangular data point in red.

Thousands of 2016 $, Natural Log Scale

Figure 9b: Predicted vs. Actual Median Net Worth, Family Heads Born 1960-1989
1,000

= year 2007

100

10
Predicted

1960-1969

1970-1979

1980-1989

1
20

25

30

35

40

45

50

55

60

65

70

75

80

Age of Family Head
The sources for all the tables and figures are the Federal Reserve’s Survey of Consumer Finances and authors’ calculations.

Three post-1960 cohorts. First, consider a
family headed by someone who was 52 years old
in 2016; this person was born in 1964. The typical
52-year-old-headed family in 2016 was about
11 percent, or some $16,800, below the wealth level
we would predict based on the experiences of all
similarly aged SCF families.11 At first glance, this
shortfall does not seem extreme or insurmountable.
However, peak inflation-adjusted income and saving typically occur in one’s 40s or 50s. Thus, people
in their 50s face a limited number of remaining
high-saving years. Moreover, the high returns on
housing and financial assets in recent years are unlikely to continue in future years.12 Thus, catching
up to the wealth benchmarks established by earlier
generations is possible but no simple feat for the
typical family respondent born in the 1960s.
12 Federal Reserve Bank of St. Louis

Next, consider a typical 42-year-old family
respondent in 2016, who was born in 1974. This
respondent’s family was 18 percent ($16,400) short
of the wealth level we predicted for age 42. The
likelihood that asset returns will not be as high in
the future as they were in recent years is a major
concern for this group, too.13
Finally, consider a typical 32-year-old family
respondent in 2016 (born in 1984). This respondent’s
family was 34 percent ($12,000) below the 32-year-old
benchmark established by earlier generations. Like
the members of the 1970s cohort, the typical 1980s
family also had higher debt in relation to both
income and assets than any previous generation
at the same ages, creating headwinds to wealth
accumulation and risks to financial stability when
setbacks occur. On the optimistic side, young

Table 2: Relative Wealth Positions of All Cohorts in SCF Years
Age

1940-1949

1950-1959

1960-1969

1970-1979

6.2

(12.3)

1980-1989

24

19.9

25
26

(41.0)
9.5

34

2.3

7.1

35
36
(1.1)

13.6
(42.7)
10.0

41

22.1

33.0

31.3
28.7

(29.1)

(5.2)

13.2
13.3
4.2
1.8
8.2

70
71

4.2
5.1

74
75

48

0.8

76

49
50

61.1

73

46
47

3.8

67

72

45

(15.5)

63

69
(17.8)

43

(5.3)

68

(2.5)

42
44

61

66

39

64.6

65

38
40

18.1

60

64

37

45.9

62
(35.1)

(10.8)

(4.1)

59
(34.1)
4.1

56.3

58

9.5

32
33

57

1960-1969

9.4

56

29
31

54

1950-1959

4.8

55

(14.9)

30

1940-1949

52

5.4

28

1930-1939

53
(24.7)

27

Age

35.0

(27.5)

18.5

51

77
78

0.8

79
80

NOTES: Each entry shows the percent difference between a cohort’s median wealth in an SCF year and the corresponding predicted median wealth
level at the average age of the cohort at that time. For example, when the average family head in the 1950-59 cohort was 34 (survey year 1989),
median wealth of the cohort was 2.3 percent above the predicted level. Only the 1940s, 1950s and 1960s cohorts were on average between the ages
of 24 and 80 in every one of the 10 SCF waves.

families today have more education on average,
which can translate into higher earning potential
if high income returns on education hold up.
Quantifying wealth shortfalls by birth decade.
Figures 9a and 9b superimpose actual median wealth
levels of families born in each of the six decades we
tracked on the predicted wealth trajectory derived
from the entire sample. Actual median wealth levels of
the cohorts born before 1960 generally were above the
corresponding predicted levels throughout the years
we observed them. The 1930s and 1940s cohorts lost

wealth after 2007, but remained above their agespecific benchmarks. (See Figure 9a.) The 1950s cohort
fell below the predicted levels in 2010 and 2013, but
exceeded it again by 2016.
Cohorts born in 1960 or later, on the other hand,
were moved well below benchmark levels by the
Great Recession and remained below them through
2016. (See figures 4 and 9b.) Table 2 displays percent
deviations from age-specific benchmark wealth
levels for all six cohorts as they traversed their life
cycles between 1989 and 2016.
The Demographics of Wealth 13

Thousands of 2016 $, Natural Log Scale

Figure 10: Predicted vs. Actual Median Income
100

= year 2007

90
80
70
60
50
40
30

Predicted

20
10

20

25

30

35

40

45

50

55

1930-1939

1940-1949

1950-1959

1960-1969

1970-1979

1980-1989

60

65

70

75

80

Age of Family Head
NOTE: For figures 10 through 14, see Appendix 1 for more details on birth cohorts and estimated and predicted values.

Figure 11: Predicted vs. Actual Saving Share
60

= year 2007

50

Percent

40
30
20
Predicted

1930-1939

1940-1949

1950-1959

1960-1969

1970-1979

1980-1989

10
0

20

25

30

35

40

45

50

55

60

65

70

75

80

Age of Family Head
NOTES: Members of a family are considered actively saving if they reported that over the past year, they spent less than their income. This spending
does not include any investments they had made. The figure shows the actual or predicted share of a cohort that actively saved in the year of the
survey. This question was not asked during the 1989 survey wave. For more information, see question X7510 in the SCF codebook.

The sources for all the tables and figures are the Federal Reserve’s Survey of Consumer Finances and authors’ calculations.

III. Why Were Young Families Hit So Hard?
To shed more light on why families headed by
someone born in 1960 or later typically are below
their age-specific wealth benchmarks—and to gauge
their potential to recover—we looked at several
financial indicators, and trends in income and saving.
Income and saving trends appear to be relatively
unimportant, while several financial indicators—
especially debt and homeownership—loom large.
Income and saving. A family’s income is a key
determinant of wealth, as higher incomes allow
greater saving. Higher income also may signal the
existence of other traits that could lead to greater
wealth accumulation, such as patience, cognitive
14 Federal Reserve Bank of St. Louis

and noncognitive abilities, and specific knowledge
(e.g., numeracy), that could improve financial decision-making. Higher income also may be associated
with access to good wealth-building institutions,
such as employer-provided and -subsidized health
insurance and retirement plans. Unusually low
incomes among members of young cohorts therefore would be a plausible source of wealth shortfalls
if they had occurred.
SCF evidence does not support the hypothesis
that low incomes have contributed to low wealth
among families whose heads were born after 1960.
Figure 10 shows that, relative to the predicted income
life cycle we estimated using all SCF families, the

Figure 12: Predicted vs. Actual Median Debt to Income
120

= year 2007

Predicted

100

1930-1939

1940-1949

1950-1959

1960-1969

1970-1979

1980-1989

Percent

80
60
40
20
0

20

25

30

35

40

45

50

55

60

65

70

75

80

Age of Family Head
NOTES: A household’s “usual” income is used for the denominator in the debt/income ratio. Respondents were asked whether their household
incomes in the past year were unusually high or unusually low. Given either response, the respondents were asked to provide their household
incomes in a “normal” year. We used that measure where relevant as a type of permanent income, insulated from yearly income fluctuations that
were perceived as temporary. For more information, see question X7650 in the SCF codebook.

Figure 13a: Predicted vs. Actual Homeownership Rates, Family Heads Born 1930-1959
90

= year 2007

80
70
Percent

60
50
40
30
20

Predicted

10
0

20

25

30

35

40

45

50

55

1930-1939
60

65

1940-1949
70

1950-1959
75

80

Age of Family Head
NOTE: The homeownership rate was calculated as the percentage share of households in each group reporting that they had any housing assets.

typical family headed by someone born in the 1960s,
1970s and 1980s has fared well. To be sure, families
headed by someone born before 1960 have done
even better, but there is no reason to believe that
income shortfalls either before or after the Great
Recession are an important source of wealth shortfalls.
A low propensity to save also would be a plausible source of low wealth. However, Figure 11 suggests that, as with income, there is nothing unusual
about the saving habits of younger cohorts. Indeed,
most groups appear to save somewhat less than the
predicted level, but this is true of all the cohorts we
studied. The Great Recession appeared to lower the
share of families that saved across all birth cohorts,

but these rates then recovered. In light of the large
wealth shortfall facing the 1980s cohort, it is
encouraging that this cohort saved at higher rates
than the 1970s cohort at the same ages.
The likely culprits for 1960s and 1970s families: houses and debt. Debt and homeownership
are more likely culprits in explaining why families
whose heads were born in 1960 and later were hit
so hard by the Great Recession and have failed to
recover completely. Beginning with families whose
heads were born in the 1930s, each successive
decadal cohort generally has piled up more debt
relative to income at a given age than the preceding cohort. (See Figure 12.) This trend was interThe Demographics of Wealth 15

Figure 13b: Predicted vs. Actual Homeownership Rates, Family Heads Born 1960-1989
90

= year 2007

80
70
Percent

60
50
40
30
20

Predicted

10
0

20

25

30

35

40

45

50

55

1960-1969
60

65

1970-1979
70

1980-1989
75

80

Age of Family Head

Figure 14: Predicted vs. Actual Delinquency Rates
14

= year 2007

12

Predicted

Percent

10

1930-1939

1940-1949

1950-1959

1960-1969

1970-1979

1980-1989

8
6
4
2
0

20

25

30

35

40

45

50

55

60

65

70

75

80

Age of Family Head
NOTES: The delinquency rates were calculated as the percentage share of households in each group reporting that they were behind on their debt
obligation payments for two months or more in the past year. For more detail, see question X3005 in the SCF codebook.

The sources for all the tables and figures are the Federal Reserve’s Survey of Consumer Finances and authors’ calculations.

rupted by 1980s families after the Great Recession,
but one should not exaggerate the improvement
this represents. The typical 1980s family, alongside
1970s families, is on track for higher debt burdens at
any given age than any previous cohort. Although
1970s and 1980s families were young during the
housing-bubble period, they also show signs of
being highly debt-burdened. For example, compared to the median 1960s family, which had a
debt-to-income ratio of about 50 percent in its
early 30s, the median 1970s family had a 107percent ratio and the median 1980s family had
an 80-percent ratio at that age.
For many families, the largest amount of debt is
in the form of a mortgage. Moving into homeownership at an early age made many young families
16 Federal Reserve Bank of St. Louis

vulnerable to the economic and financial shocks
of the Great Recession. Figures 13a and 13b contrast the homeownership rates of families whose
heads were born before and after 1960, respectively.
Families headed by someone born in the 1960s and
1970s had homeownership rates above predicted
levels before the Great Recession. By 2016, those
groups’ homeownership rates had fallen significantly below predicted levels. Corresponding to the
abrupt flattening of homeownership trajectories for
these groups, debt-delinquency rates have been
high for all cohorts born in 1960 and later, a sign of
burdensome debt. (See Figure 14.)
Together, high debt ratios, high homeownership
rates and high delinquency rates among 1960s
and 1970s families point to housing and mortgage

debt as probable sources of important wealth losses
during the Great Recession. Conversely, as home
values recovered in recent years, many of these
homeowners benefited, as demonstrated by significant reductions in typical wealth shortfalls relative
to benchmark levels. (Recall Figure 4 and Table 2.)
Not like their elders. Families whose heads
were born in the 1980s are different. They generally were too young to be homeowners during
the housing bubble; in fact, only 19 percent of
1980s families were homeowners in 2007. Even
by 2016, fewer than 45 percent of 1980s families
were homeowners. The predominant type of debt
they owe is non-mortgage debt, including student
loans, auto loans and credit card debt. Because
none of these types of debt finance assets that have
appreciated rapidly during the last few years—such
as stocks and real estate—they have received no
leveraged wealth boost like that enjoyed by older
cohorts. The 1980s cohort was unique in falling
even further behind its wealth benchmark between
2010 and 2016. Given the prospect of lower asset
returns in the future than in the recent past, 1980s
families face a formidable challenge in building
wealth rapidly enough to reach benchmark levels
set by earlier generations.

Historically, high asset returns in recent years have
not prevented the 1980s cohort from falling further
behind its wealth benchmark between 2010 and
2016. And, with the exception of the 1970s cohort,
this group may be on track to bear the heaviest debt
burden ever.
Income, saving and homeownership trends
have been unexceptional for the 1980s cohort so
far. Efforts to enhance the first two and a measured
approach to the third—including careful management of mortgage debt—would serve this generation well. Cautious use of non-mortgage debt also
will be important.
It is far too soon to know whether families
headed by someone born in the 1980s will become
members of a lost generation for wealth accumulation. To be sure, there are grounds for optimism. Yet
there are reasons to be very concerned about the
financial outlook for many young Americans.

IV. Will Young Families Become a Lost
Generation for Wealth Accumulation?
Two key factors on the side of 1980s families
are time and education. The oldest member of the
1980s generation was just 36 in 2016, while the
youngest was only 27. These families have many
more years to earn, save and accumulate wealth.
At the same time, this is the most highly educated
generation we evaluated. Pursuing a college degree
can be very expensive and offers no guarantees.
Nonetheless, the average return is substantial.14 It is
possible that the income and wealth trajectories of
this generation will be steeper than those of earlier generations, allowing many families to achieve
their wealth goals in the end.
Yet the task faced by the typical 1980s family
should not be underestimated. This cohort has
been the slowest to recover from the Great
Recession. In fact, its wealth shortfalls (relative
to the age-specific benchmark levels we predicted)
were the only ones to worsen from 2010 to 2016.

The Demographics of Wealth 17

Appendixes
Appendix 1:
Sample Selection and Focus of Analysis
Table 3 provides the average age and sample
representation for all decadelong cohorts born in
the 20th century. We used all SCF families when
estimating life cycles for all variables of interest but
truncated the sample in two ways in our analysis.
First, we focused only on family respondents
between the ages of 24 and 80. Family heads under
24 are “immature” both because many are in the
process of transitioning to a new family structure
(marriage or long-term partnering) and because
their formal education may not be complete. To be
sure, these transitions also occur after age 24 for
some people, but we avoided restricting our sample any more than was necessary. Families over 80
are very likely to be unrepresentative of their birth
cohorts at earlier stages in their lives due to survivorship bias, so we excluded them, too. Again, an
argument could be made that a different (probably
younger) cutoff age might be desirable, but we
believe 80 is a defensible choice.
The second way we limited our analysis was
to follow only families headed by someone born
between 1930 and 1989. In other words, we did not
discuss results for families whose heads were born

in the 1900s, 1910s, 1920s and 1990s, even though
they were included when we estimated life cycle
trajectories. Family heads born between 1900 and
1929 already were over 80 by 2010, the first SCF
year after the Great Recession. Thus, we could not
track the Great Recession’s effects on their wealth
within our chosen age range. Family respondents
born in 1990 or later were under 18 in 2007, the last
SCF year before the Great Recession. Therefore, we
could not draw before-and-after comparisons for
this group either.
Many different potential birth cohorts were
considered prior to settling on the ranges presented
here. The 10-year bands offer easier identification
(born in the ’80s, born in the ’70s, etc.), as well as a
robust sample size. Along with the benefits, there is
one complication with this approach: Our life
cycles were predicted from age; at the same time,
the median net worth and similar statistics
estimated for birth cohorts were generated using
the responses for all households within a 10-year
range. Comparing the estimate for the cohort to
the prediction from the life cycle trend requires a
single age to identify the cohort and place it in our

Table 3: Average Age and Sample Representation
Birth
Cohorts

Mean
Age in
1989

Mean
Age in
2007

Mean
Age in
2016

1900-1909

83.1

1910-1919

74.3

1920-1929

64.5

81.8

89.6

1930-1939

54.3

72.3

81.0

90.3

Sample Size in Each Survey Wave
1989

1992

1995

1998

2001

2004

2007

2010

2013

2016

121

100

57

35

17

5

0

0

0

0

342

306

270

195

117

68

58

29

10

0

545

619

550

449

417

292

257

217

152

89

593

608

654

594

544

515

435

510

417

370

1940-1949

43.7

62.2

71.0

686

814

870

894

854

920

836

995

875

853

1950-1959

34.4

52.2

61.3

571

811

933

965

1,086

1,078

1,056

1,484

1,298

1,370

1960-1969

25.5

42.5

51.7

273

555

732

722

797

887

868

1,402

1,310

1,299

1970-1979

18.8

32.6

41.6

7

87

233

438

543

575

606

1,097

1,015

1,062

23.9

31.6

0

0

0

14

66

178

301

705

768

878

23.3

0

0

0

0

0

0

0

43

170

327

1980-1989
1990-1999

NOTE: Sample size was rounded wherever appropriate.
The sources for all the tables and figures are the Federal Reserve’s Survey of Consumer Finances and authors’ calculations.

18 Federal Reserve Bank of St. Louis

Appendix 2:
Life Cycle Regressions
wealth life cycle graphs. We calculated the average
age among all households within the cohort and
rounded to the nearest integer. We paired that age
with the statistic estimated for the cohort. This is an
inexact matching process, and the value estimated
could be associated with a household that is slightly
younger or older than the average age in the cohort.
We believe the magnitude of the potential error
using this approach is relatively small and outweighed by the ease of exposition.

The life cycle trend for continuous variables
(net worth, debt-to-income ratio) was estimated
using median multiple regression. The trends used
in comparisons with birth cohorts were modeled as:

Yi = C + β1 Ai + β 2 Ai2 + β3 Ai3 + Yeari
where Y was the outcome of interest for household i;
A is the age of the respondent; and Year was a
vector of estimated coefficients and respective binary
variables for each of the SCF waves, omitting 1989.
For variables reflected as population shares
(owning a home, rates of delinquency of 60-plus
days in a year), we relied on probit regression, a
nonlinear regression model designed for binary
dependent variables. A probit regression models the
probability that Y (the dependent variable) equals 1.
Our full specification was modeled as follows:

Pr (Yi = 1) = Φ (C + β1 Ai + β 2 Ai2 + β3 Ai3 + Yeari )
where F was the cumulative standard normal
distribution.
The sample size was 47,776 families surveyed
across all years of the SCF. The comparison of life
cycle trends by subsets of survey years simply
dropped the vector of survey year binaries from the
specification and omitted all observations not in
the survey years of interest. All estimates incorporated nonresponse-adjusted sample weights.

The Demographics of Wealth 19

Endnotes
1 Throughout the essay “typical” describes the
family in the middle or at the median.
2 Household ages are defined as the age of the
survey respondent. We use the terms “family”
and “household” interchangeably. We sometimes
refer to family respondents as “family heads.”
3 All dollar figures in this essay are adjusted for
inflation and are expressed in terms of 2016
purchasing power.
4 The first essay in this series (published in February) discussed the role of both one’s own and
one’s parents’ education in determining one’s
income and wealth. (See Emmons, Kent and
Ricketts, February 2018.) The third essay will
discuss race, ethnicity and wealth. The 2015
series of The Demographics of Wealth analyzed
race and ethnicity, education, and age and birth
year, respectively, using SCF data through 2013.
(See Emmons and Noeth, February, May and
July 2015.)
5 We used data from all ages to estimate the life
cycle of wealth but focused on ages 24 through
80 in this essay. (See Appendix 1 for details of
our sample selection.)
6 True panels are available in other datasets, such
as the Panel Study of Income Dynamics (PSID).
But direct comparisons have shown that SCF
quasi-panels perform well in many respects and
are superior in capturing family wealth, especially at the high end. (See Bosworth and Anders,
2008, and Pfeffer et al., 2016.)
7 See Modigliani (1988).
8 The inherent asymmetry of insuring against a
scenario of outliving your savings leads to positive net worth upon death, on average, among
families seeking to self-insure.
9 See De Nardi, French and Jones (2010).
10 See Bosworth and Anders (2008).
11 We explain in Sidebar 2 and Appendix 2 how we
predict typical wealth at each age. The earliest
birth year included in our sample of 52-year-old
family heads was 1937—these family heads were
52 years old in 1989—and the latest was 1964.

20 Federal Reserve Bank of St. Louis

12 See Table 1 in Emmons (2017). The average
annual rate of household wealth accumulation
between 2011 and 2016 was more than three
times its average long-term rate.
13 Between 1986 and 2012, which roughly
corresponds to our sample period, almost half
of wealth accumulation was due to capital
gains—that is, rising asset prices. (See Table 2 in
the working paper version of Saez and Zucman
(2016).) Thus, a period of below-average
asset-price increases would slow wealth
accumulation significantly.
14 See Emmons, Kent and Ricketts (2018).

References
Bosworth, Barry P.; and Anders, Sarah. “Saving and
Wealth Accumulation in the PSID, 1984-2005.”
Center for Retirement Research at Boston College,
Working Paper 2008-2, February 2008.
De Nardi, Mariacristina; French, Eric; and Jones,
John B. “Why Do the Elderly Save? The Role of
Medical Expenses.” Journal of Political Economy,
February 2010, Vol. 118, No. 1, pp. 39-75.
Emmons, William R. “Is Homeownership Bad for
Wealth Accumulation?” Housing Market
Perspectives, Federal Reserve Bank of St. Louis,
December 2017, Issue 7.
Emmons, William R.; Kent, Ana H.; and Ricketts,
Lowell R. “The Financial Returns from College
across Generations: Large but Unequal.”
The Demographics of Wealth 2018 Series, Federal
Reserve Bank of St. Louis, February 2018, Essay No. 1.
Emmons, William R.; and Noeth, Bryan J. “Race,
Ethnicity and Wealth.” The Demographics of
Wealth, Federal Reserve Bank of St. Louis,
February 2015, Essay No. 1.
Emmons, William R.; and Noeth, Bryan J.
“Education and Wealth.” The Demographics of
Wealth, Federal Reserve Bank of St. Louis, May 2015,
Essay No. 2.
Emmons, William R.; and Noeth, Bryan J. “Age, Birth
Year and Wealth.” The Demographics of Wealth,
Federal Reserve Bank of St. Louis, July 2015,
Essay No. 3.
Modigliani, Franco. “The Role of Intergenerational
Transfers and Life Cycle Saving in the Accumulation of Wealth.” Journal of Economic Perspectives,
Spring 1988, Vol. 2, No. 2, pp. 15-40.
Pfeffer, Fabian T.; Schoeni, Robert F.; Kennickell,
Arthur; and Andreski, Patricia. “Measuring Wealth
and Wealth Inequality: Comparing Two U.S.
Surveys.” Journal of Economic and Social Measurement, June 2016, Vol. 41, No. 2, pp. 103-20.
Saez, Emmanuel; and Zucman, Gabriel. “Wealth
Inequality in the United States since 1913:
Evidence from Capitalized Income Tax Data.”
Working Paper No. 20625, 2014. Retrieved from
http://www.nber.org/papers/w20625.pdf.

The Demographics of Wealth 21

Also from the St. Louis Fed’s
Center for Household Financial Stability
All about family finances
The role student loans play in racial wealth gaps. The difference
a little cash on hand makes. A record high for household wealth.
Those are a few of the subjects covered in the series In the Balance.
The short essays showcase research on American households’
balance sheets and what can be done to strengthen them. Read the
essays at www.stlouisfed.org/publications/in-the-balance.

Your house and your money
The quarterly Housing Market Perspectives has looked at the
potential effects of tax code changes on housing choice, the
impact of homeownership on a family’s wealth, how spending on
housing affects the economy, and more.
The latest on Center research and events is available at
www.stlouisfed.org/publications/housing-market-perspectives.

Perspectives on Household Balance Sheets

Cash on Hand Is Critical for Avoiding Hardship
By Emily Gallagher and Jorge Sabat

W

hy would someone keep $1,000 in
a low-earning bank account while
owing $2,000 on a credit card that charges
a double-digit percentage interest rate?
Our research suggests that keeping a
cash buffer greatly reduces the risk that
a family will miss a payment for rent,
mortgage or a recurring bill, will be unable
to afford enough food or will be forced to
skip needed medical care within the next
six months.
Many families struggle to make ends
meet. A Federal Reserve survey estimated
that almost half of U.S. households could
not easily handle an emergency expense
of just $400.1
Should more families be encouraged
to hold a liquidity buffer even if it means
incurring more debt in the short-term?

Linking Balance Sheets and
Financial Hardship

population of interest for understanding
the antecedents of financial hardship.
We tracked families who said in the first
survey that they hadn’t recently experienced any of four types of financial hardship: delinquency on rent or mortgage
payments; delinquency on regular bills,
e.g., utility bills; skipped medical care; and
food hardship, defined as going without
needed food.
To assess whether the composition of a
family’s balance sheet helped predict any
of these forms of hardship, we asked in
the initial survey if the family had any balances in the following categories:
• Liquid assets, such as checking and
saving accounts, money market funds,
and prepaid cards
• Other assets, including businesses,
real estate, retirement or education
savings accounts
• High-interest debt, such as that from
credit cards or payday loans
• Other unsecured debt, such as student
loans, unpaid bills and overdrafts
• Secured debt, including mortgages
or debts secured by businesses, farms
or vehicles.
More details on the categories can be
found in the methodology.
We controlled for factors such as income
and demographics and tracked whether the
roughly 5,000 families had suffered a financial shock that would affect the results.

The Center for Household Financial
Stability at the Federal Reserve Bank of
St. Louis focuses on family balance sheets,
especially those of struggling American
families. The Center researches the determinants of healthy family balance sheets,
their links to the broader economy and
new ideas to improve them. The Center’s
original research, publications and public
events aim to impact future research, community practice and public policy. For more
information, see www.stlouisfed.org/hfs.

Using a novel data set, we investigated
which types of assets and liabilities predicted
whether a household would experience
financial hardship over a six-month period.2
The survey data that we use is particularly apt to study this question, not only
because it asks the detailed financial and
demographic questions that are often
missing from public surveys, but also
because it includes two observations for
the same household. One observation is
collected at tax time and another observation is collected six months after tax time. ISSUE
8 | MARCBalance
Results:
Sheets Matter
H 2018
This feature of our data set is ideal for
Our results are summarized in the figure,
capturing the probability that a household
which displays the estimated effects of
that is currently financially stable falls into
variations in each balance-sheet category
financial hardship in the near term. FurF EDE RAL
on the risk of encountering financial
thermore, the survey samples only from
R E S E RVE
low-to-middle income
households, our
BAN
(continued on Page 2)
K

ST. LOU IS

HOUSING MAR

PERSPE

On the Level

KE T

CTIVES

wit
1 h Bil
l Em

mons

Fewer Tax Brea
ks for
Homeowners:
A Good Thing?
B

Bill Emmons is
an
assistant vice presid
ent and
economist at the
Federal
Reserve Bank of
St. Louis and
the senior econo
mic adviser for
the Bank’s Cente
r for Household
Financial Stabili
ty.

road indexes
of real, or infla
tionadjusted, hous
• State and
e prices gene
local taxes are
rise and fall
rally
no longer
fully deductibl
with economic
• The tax dedu
e, making it
activity.
This suggests
ctibility of mor
less likely
that a hous
that what’s
tgage
interest on
ehold’s item
good for
homeowners
second mor
ized deductions will exce
, vis-à-vis risin
tgages (i.e.,
home equity
ed the new
g house
prices, is also
loans) and seco
standard
deduction.
good for the
nd
homes was
economy.
(See the acco
scaled back.
mpanying figur
• The maximum
e.)
• Marginal
But could a
amo
tax rates were
unt of mortdecline in real
gage debt for
reduced,
house
cutting the
prices also be
which inter
value to an
good for the
est can be
deducted was
itemizer of
econthe
omy? If it’s
MID
redu
ced to $750
and all othe
the result of
,000
r deduction
from $1 milli
efficiencyenhancing chan
s.
on for joint
filers. (Any Likely Effec
ges in the tax
loans taken
ts on Housing
code,
many economis
out after Dec.
ts say yes.
15, 2017
are subject
As a result
to the new
of these chan
Recent Tax Law
rule; existing
mortgages have
ges, many
Changes
economists
been grandfath
expect hou
se prices to
in with the
ered
The Tax Cuts
trend somewh
old limit.)
and Jobs Act
at lower. 1 Mor
(TCJA) of
2017 places
tgage
borrowing
new limits on
and other aspe
deductions
for state and
cts of
Annual Growth
local taxes and
Rates: Real Hous
property
taxes, and scale
e Prices and Real
s back the mor
10
GDP
tgag
interest dedu
per Capita
e
ction (MID).
Many economists expe
ct these chan
ges to reduce
the number
of taxpayer
s who claim
5
the MID on
itemized retu
rns starting
with the 2018
tax year.
Several prov
isions of TCJA
will
affect taxpayer
0
s:
• The standard
deduction was
doubled, to $12,
000 for indiv
iduals and
$24,000 for
-5
joint filers,
making it
likely that mos
t low- and midd
Real house
price index
leincome taxp
ayers who item
Real GDP per
capita
ized in
the past will
-10
choose the
standard
deduction inste
1975
1980
ad.
1985
1990
Percent

Stay up to date
The latest on Center research and events is available at
www.stlouisfed.org/household-financial-stability.

ISSUE 18 | NOVEMBER 2017

1995

SOURCES: Federa

l Housing Financ
e Agency and

FEDERAL RESERV

E BANK OF

1

22 Federal Reserve Bank of St. Louis

ST. LOUIS

2000

Bureau of Econo

2005

mic Analysis.

2010

2015

2020

Credits
Editor: Heather Hennerich
Designer: Joni Williams

The Demographics of Wealth 23

All the essays in this series can be read on the website of the Center for Household Financial
Stability at www.stlouisfed.org/hfs.