Digital Commons @ George Fox University
Faculty Publications School of Business School of Business
2010
Fueling the Credit Crisis: Who Uses Consumer
Credit and What Drives Debt Burden?
Diane K. Schooley
Boise State University
Debra Worden
George Fox University, dworden@georgefox.edu
Follow this and additional works at: h5p://digitalcommons.georgefox.edu/gfsb
Part of the Business Commons
4is Article is brought to you for free and open access by the School of Business at Digital Commons @ George Fox University. It has been accepted for
inclusion in Faculty Publications School of Business by an authorized administrator of Digital Commons @ George Fox University. For more
information, please contact arolfe@georgefox.edu.
Recommended Citation
Schooley, Diane K. and Worden, Debra, "Fueling the Credit Crisis: Who Uses Consumer Credit and What Drives Debt Burden?"
(2010). Faculty Publications School of Business. Paper 31.
h5p://digitalcommons.georgefox.edu/gfsb/31
Fueling the Credit Crisis: Who Uses Consumer Credit and
What Drives Debt Burden?
DIANE K. SCHOOLEY and DEBRA DRECNIK WORDEN
n
Excessive household debt contributed to the worst
recession in decades. Insights about borrowing and
spending behavior can inform economic recovery
forecasts, policy decisions, and financial education.
This study identifies life cycle and credit attitude as
key determinants of who uses debt. Younger house-
holds are more likely to borrow for consumption,
as are those who believe that it is all right to borrow
to purchase luxury goods or cover living expenses.
Furthermore, households that condone borrowing
for these purposes have a higher consumer debt
burden. Debt capacity (or creditworthiness) and
financial discipline are also significant factors in
determining household debt use.
Keywords: consumer debt, credit attitude, personal
savings
T
he United States appears to be slowly recovering
from the worst recession since the Great De-
pression. Many factors contributed to the condi-
tions that led to the bursting of the consumer debt
bubble and the resulting collapse in consumption.
Access to credit eased during the booming housing
market, expanding home ownership, especially in
the lower brackets of the income distribution.
Mortgage debt was used to finance consumer
spending, which comprises 70 percent of the U.S.
GDP. However, since the downturn that began in
2007, jobs have been lost, housing markets are
down, and default rates are up. With traditional
sources of credit drying up, people have made recent
moves to cash out retirement and life insurance
plans. A generation of Americans is experiencing,
for the first time, depression-era economic condi-
tions [Laise 2008].
Given the likely fundamental shift in the
mindset of consumers, some wonder whether
households will return to past patterns of spending
and borrowing, or whether the changes in market
structures are so substantial that a recovery will not
resemble any others that have been experienced.
Economic theory suggests that the way to get the
economy back on its feet is accelerated spending.
However, after nearing zero in early 2008, the
personal savings rate briskly rose to levels unseen
in over a decade. Consumers have just only begun
to cautiously increase their spending. Conventional
wisdom says that until concerns about job security
are alleviated and consumer confidence increases,
consumers and businesses will maintain their new-
found austerity and reluctance to spend.
Although the future of the credit market is
unknown, some insight into how household
spending will respond in a new era of asset markets
and access to credit can be gained by examining
consumer credit use and attitudes about credit
before the bubble burst. This study employs uni-
variate and multivariate analyses to examine how
household characteristics may be related to the
incidence of consumer debt use and size of debt
burden (consumer debt/annual income). Char-
acteristics examined include stage of life cycle,
attitudes about credit use, debt capacity, financial
discipline, and economic expectations. Section 1
presents utility optimization models of household
borrowing/spending behavior. Section 2 describes
the survey data, Sections 3 and 4 present the ana-
lyses, and Section 5 provides conclusions.
n
Diane K. Schooley is Associate Dean and Professor of Finance at Boise State University. Her research interests include
corporate governance, corporate finance, and consumer finance. She is a Certified Treasury Professional. She earned a Ph.D. in
finance from the University of Colorado-Boulder. Debra Drecnik Worden is Professor of Business and Economics in the School of
Business at George Fox University. She earned a Ph.D. in Economics from Purdue University. Her research interests include
quantitative methods and consumer finance.
Business Economics
Vol. 45, No. 4
r
National Association for Business Economics
266
1. Household Saving and Borrowing
Saving and borrowing link household consumption
and income. Households save when income exceeds
spending and borrow (dissave) when spending ex-
ceeds income. Many theories of household saving/
borrowing motives and behavior that link con-
sumption and income have been developed over the
past 70 years. This section reviews several optimi-
zation models, where saving/borrowing behavior
optimizes a household’s utility over time.
Optimization models include variations of
the permanent income hypothesis (PIH) model,
attributed to Friedman [1957] and variations of
the Modigliani and Brumberg [1954] life-cycle (LC)
model. For PIH/LC models, the objective of saving
or borrowing is to smooth consumption over pre-
dictable fluctuations in income in order to achieve
constant marginal utility of consumption over
time. Although generally similar, the models differ
in their assumptions about length of planning
period. This period is infinite in PIH but finite over
the life of the household for LC.
Over the life cycle of a typical household, var-
iation in income is greater than the variation in
spending or consumption. Income is low in the
early stages of the life cycle, increases to a max-
imum before retirement, and then decreases during
retirement. According to LC, in order to maintain
a constant level of marginal utility, households will
borrow during the early stages of life, save in the
middle stages, and spend savings during retirement.
Expected income plays a key role in these
models, where permanent income has much more
influence on consumption than current or tempor-
ary income. Households expecting a permanent
increase in income will reduce their savings and/or
borrow against that higher income, realizing an
increased level of consumption that they expect
to sustain over time. Households anticipating a
permanent decrease in income will decrease con-
sumption—saving more and/or paying down debt.
Temporary changes in income—such as might
result from fluctuations in the economy—have no
sustained impact on consumption. However, a
temporary increase in income yields more saving
and/or debt repayment; a temporary decrease yields
less saving and/or more debt.
A household’s savings and borrowing decisions
are influenced by its preference for future vs. pre-
sent consumption. A decision to borrow implies
that current consumption is preferred to future
consumption; a decision to save implies the opposite.
If interest rates are expected to increase, the asso-
ciated increase in prices, and the implied higher
discount rate, reduces the present value of future
consumption. This should generate a stronger
preference for current consumption and encourage
borrowing.
Baek and Hong [2004] outline several limita-
tions of the PIH/LC models and the lack of em-
pirical evidence supporting them. The basic models
do not recognize marital status nor allow for the
presence of children in the household, determining
life cycle stage solely by age of the household
head [Browning Deaton and Irish 1985]. The basic
PIH/LC models also do not provide for liquidity
constraints. Although households may desire to
borrow in order to smooth consumption, they may
not qualify for adequate amounts of credit needed
to smooth consumption [Deaton 1992]. Finally,
the basic PIH/LC models do not recognize a pre-
cautionary motive of saving [Carroll and Summers
1991; Deaton 1991]. Prudence can explain why
households in early life cycle stages may not bor-
row as much as the PIH/LC model would predict
and why households in later life cycle stages
may not draw down assets as quickly as would be
predicted.
The PIH/LC models predict that households
will borrow during a recession in order to maintain
a level of consumption. When income declines, the
savings rate should be low or at least decline. In
other words, these models predict that borrowing
increases and saving decreases during a recession.
However, although the personal savings rate—
defined by the Bureau of Economic Analysis as
personal savings as a percent of personal dis-
posable income—declined throughout 2007 and
early 2008 to a low of 1.2 percent, it increased
substantially as the recession deepened, reaching a
high of 5.4 percent in the second quarter of 2009. In
this regard, household behavior may better be ex-
plained by the buffer-stock model, discussed below.
According to the buffer-stock savings model
[Deaton 1991; Carroll 1992], consumers hold assets
to protect consumption against unpredictable
fluctuations in income. Households can be both
impatient—borrowing to finance current con-
sumption if income is known with certainty—and
prudent, by holding precautionary balances. A
tradeoff of impatience and prudence ensures
a target level of wealth held as a sufficient buffer
against income fluctuations. Holding a target level
of wealth to weather fluctuations in incomes
FUELING THE CREDIT CRISIS
267
implies a precautionary motive for saving. Unlike
in the buffer-stock model, uncertainty in future
incomes plays no role in the PIH/LC models.
The possibility of unemployment leads to un-
certainty in household income and so affects cur-
rent consumption and saving. The more uncertain
household income, the higher the buffer stock of
wealth required; saving increases relative to con-
sumption. This precautionary motive can explain
the increases in personal saving rates in 2008 and
2009. With employment and income uncertainty
rising during the recession, households save more
in order to increase their buffer stock. And, as
the economy entered its slow recovery, consumer
spending increased cautiously and the personal
savings rate began to decline, reaching 3.5 percent
in the first quarter of 2010.
The PIH/LC models hold that stage of life
cycle, expected interest rates, and expected changes
in permanent income affect saving and borrowing
as individuals seek to smooth consumption over
their lifetimes. According to the buffer-stock
model, uncertainty of future income and employ-
ment drive the size of buffer-stock assets, and
thereby saving and borrowing, needed to smooth
consumption. Both models assume rational trade-
offs, but neither address attitudes toward credit.
The easy credit climate (until recently) and the
seemingly unlimited appetite for consumption may
influence households to consume now regardless
of consequences later. Unwary consumers may
underoptimize utility over their lifetimes by over-
consuming today and limiting future consumption.
Contrary to theoretical assumptions, they may not
be looking beyond today. The analysis that follows
examines the impact of credit attitudes on debt use,
as well as factors proposed by the models.
2. Survey Data
The public database of the 2007 Federal Reserve
Board Survey of Consumer Finances (SCF) is used
for this analysis.
1
The purpose of the SCF is to
provide a comprehensive view of the financial
behavior of a cross-section of U.S. households.
Information is gathered on all assets and liabilities
of the household, as well as demographic char-
acteristics such as home ownership, employment,
income, and makeup of the household unit.
Attitudes about the use of credit and savings
behavior are also measured.
The SCF is distinguished in its sample design.
Two sampling techniques are employed to obtain
more detail on the financial behavior of those
households holding a disproportionate share of
the nation’s wealth [Aizcorbe Kennickell and
Moore 2003].
2
Two-thirds of the households
included in the data set are randomly selected
from across the United States; the remaining
are wealthy households selected from a tax-return
derived list. Although this sampling design prohi-
bits the use of the sample as representative of
the U.S. population, inferences can be made about
the relationships among variables within house-
holds.
The 2007 SCF data were gathered mostly be-
tween May and December, with a small fraction
of interviews conducted in early 2008. Although
the survey data may not reflect the substantial
decline in asset prices that followed, the economy
was visibly slowing during the latter half of 2007.
Returns in the stock market went from double
digit growth in 2006 to relatively flat annual returns
in 2007, with markets on the decline by year-end.
3
Housing prices had begun to decline in mid-2006
[S&P/Case-Shiller] and sales of existing homes
were on a steady downturn since the second quarter
of 2007. In the last quarter of 2007, household net
worth fell for the first time in over five years
[Gongloff 2008]. The national jobless rate was
creeping upward. Personal consumption, and so
real GDP, was rising at a slower pace, and the
media were anticipating the release of pessimistic
data for the fourth quarter of 2007. All of these
signs of economic slowdown would impact house-
holds’ responses about attitudes toward credit use
and financial decisions.
Variable definitions and descriptive statistics
for the sample are provided in Tables 1 and 2. The
measure of consumer debt utilized is expanded
beyond the standard definition. As individuals
increasingly face a “credit crunch,” they access
1
The data are available at www.federalreserve.gov/pubs/
oss/oss2/2007/scf2007data.html.
2
The SCF also treats nonresponses differently. The
method of multiple imputation replaces each missing value
with a set of five values that represent a distribution of pos-
sibilities. Thus, the final database consists of five complete
observations for each respondent, which are combined for the
analysis [Rubin 1987 and Kennickell 1991].
3
The S&P 500 Index rose 15.8 percent in 2006 and 5.5
percent in 2007. The DJIA rose 16.3 percent in 2006 and 6.4
percent in 2007.
Diane K. Schooley and Debra Drecnik Worden
268
credit in new ways—including tapping home equi-
ty, pension plans, and life insurance. The distinc-
tion between credit card debt and installment
loans is blurred—credit limits are so high that
cardholders can use their lines of credit to purchase
durable goods and finance vacation expenses.
Second mortgages and home equity lines of credit
can be utilized for a variety of purposes beyond
home improvement and remodeling. This analysis
attempts to capture the use of borrowings against
Table 1. Variable Definitions and Statistics
Dependent Variables Definition Descriptive Statistics
Consumer Debt Total of credit card balances, installment credit, other debt
such as loans against pensions, life insurance, and margin
loans, as well as second mortgage and home equity lines
of credit borrowings used to purchase consumer goods
Median value for those with
consumer debt: $15,000
Consumer Debt Burden Consumer debt divided by annual household income Median value for those with
consumer debt: 0.218
Consumer Debt Use Household has consumer debt Proportion of Sample: 0.605
Explanatory Variables Definition Descriptive Statistics
Life Cycle Proportion of Sample:
Young Single Single, age under 40, no children under 18 years 0.055
Young Couple Married or with partner, age under 40, no children under
18 years
0.043
Young Family Married or with partner, age under 40, children under 18 years 0.118
Single Parent Single, age under 50, children under 18 years 0.065
Middle-Aged Couple Married or with partner, age 40–49, no children under 18 years 0.041
Mature Family Married or with partner, age 40 plus, children under 18 years 0.189
Mature Couple Married or with partner, age 50 plus, no children under
18 years
0.326
Mature Single Single, age 50 plus, no children under18 years 0.164
Credit Attitude Respondent indicates that it is all right to (1) borrow money to
cover vacation expenses or purchase a fur coat or jewelry
and/or (2) cover living expenses when income is cut
0.556
Debt Capacity
FT Income Household earns at least one full-time income 0.719
Uncertain Income Household reports next year’s income is uncertain 0.366
Housing Status Household owns home 0.742
Financial Discipline
No Saving Rule Household has no consistent plan for saving income 0.484
Revolve Household only sometimes or hardly ever pays total balance
owed on credit cards
0.324
Late Pay Household got behind or missed loan payments in the past year 0.228
Economic Expectations
Higher Interest Rates Respondent expects interest rates to increase over the next
5 years
0.633
Lower Interest Rates Respondent expects interest rates to decrease over the next
5 years
0.077
Control Variables Definition Descriptive Statistics
Net Worth The value of all real and financial assets owned, including
business equity, less the value of all mortgage and consumer
debt outstanding; in $000s
Median value for all households:
$302.15
Income Total gross income received by the household in 2006 from
all sources, including withdrawals from IRAs and pension
accounts; in $000s
Median value for all households:
$70.00
FUELING THE CREDIT CRISIS
269
home equity as a substitute for the usual install-
ment or consumer credit. The amounts owed on
second mortgages or home equity lines of credit
that were used to purchase durable goods (for
example, cars, recreational vehicles, major appli-
ances, furniture) or for consumption (for example,
entertainment equipment, vacations, general living
expenses) are included in the measure of consu-
mer debt. According to this measure, just over 60
percent of the households in the sample use con-
sumer debt. For those households with consumer
debt, the median consumer debt outstanding is
$15,000 and the median debt burden—measured
by the ratio of consumer debt outstanding to
annual household income—is 21.8 percent.
4
Although no standard definition of household
life cycle stages exists, it is generally accepted
that age, marital status, and the presence of chil-
dren should be taken into consideration. The life
cycle variables in this analysis follow the construct
presented by Bojanic [1992], where the non-
traditional Single Parent and Middle-Aged Child-
less Couple are included to capture changes in
today’s household structure. The most common
stage, Mature Couple, comprise almost one-third
of the sample, followed by Mature Family and
Mature Single at roughly half their size.
Age, income, consumer debt use, and debt
burden by stage of life cycle are presented in Table 2.
Young Single households (median age 28 vs. 51
for the overall sample) have the highest median
consumer debt burden at just over 39 percent, with
74 percent of such households having consumer
debt. These households only have one income
earner, if that, and are likely to use consumer debt
to meet current consumption, borrowing against
expectations of debt repayment out of future
earnings. Although a greater proportion of Young
Couples and Young Families, who are a slightly
older, have consumer debt (78 and 79 percent,
respectively), they have higher income; and their
median consumer debt burdens are less than 30
percent. Even with two income-earners, the con-
sumption needs of young households in their
family formation years render consumer debt an
attractive option. About 72 percent of Single
Parents (median age 37) have consumer debt, and
their low median income of $26 thousand results
in a consumer debt burden of nearly 35 percent.
Like Young Singles, these households only have
one income earner, and like Young Families,
they have children with voracious consumption
appetites.
The Middle-Aged Couple has an older house-
hold head (median age 45) and no children at
home. Although 70 percent of these households
have consumer debt, their median income of $89
thousand results in a median debt burden of 22.5
percent. The Mature Family is similar in age but
still have children at home. Two-thirds of these
households have consumer debt; a high median
income of $130 thousand yields the lowest median
consumer debt burden at just 16.5 percent. Both of
Table 2. Characteristics of Life Cycle Stages
Life Cycle Stage Median Age
Median Income
($000 s)
Have Consumer
Debt (%)
Median Consumer
Debt Burden*(%)
Young Single 28 32.0 73.7 39.1
Young Couple 29 58.5 78.1 29.5
Young Family 34 60.0 79.3 26.6
Single Parent 37 26.0 71.8 34.6
Middle-Aged Couple 45 89.0 70.0 22.5
Mature Family 48 130.0 65.8 16.5
Mature Couple 63 160.0 49.8 17.4
Mature Single 65 31.0 45.6 18.1
Overall (n=4,418) 51 70.0 60.5 21.8
*For those with consumer debt.
251 households (5.7 percent of the original sample) could not be classified in Bojanic’s life cycle measure, for example, a single
person over 50 with young children does not fit into any category.
4
Influenced by high outliers, the mean amount of con-
sumer debt outstanding is $151,996 and the mean debt burden
is 48.5 percent.
Diane K. Schooley and Debra Drecnik Worden
270
these types of households are in a life cycle stage
of earning income and building assets, and they
may be able to borrow against more than one
income stream.
The Mature Couple and the Mature Single are
similar in age (low to mid-60s). Less than half
of these households use consumer debt, resulting
in median debt burdens of 17 to 18 percent.
However, the couple has a median income of
$160 thousand, compared with $31 thousand for
the single household. These households are in
the life cycle stage where they have accumulated
assets. Either their consumption needs are not as
extensive, or they can liquidate assets rather than
incur debt.
Debt Capacity or creditworthiness is measured
with three variables—whether the household
head and/or partner earns a full-time income,
whether next year’s household income is uncertain,
and whether the household owns their home vs.
renting or otherwise. Nearly 72 percent of the
households in this sample have at least one full-
time income and over 74 percent are homeowners.
Both of these indicate ability to qualify for con-
sumer credit. However, almost 37 percent of the
households are uncertain about next year’s in-
come—perhaps a sign of the slowing economy,
rising unemployment, and falling asset values.
Financial Discipline is measured with three
variables—the existence of a consistent savings
plan, paying off credit-card balances in full each
month, and making loan payments on time. The
variables are coded to represent the lack of dis-
cipline. A consistent plan for saving includes setting
money aside each month or spending the income
of one family member while saving all other
income. Forty-eight percent of the households
report no savings plan or profess not to save at all.
Nearly one-third of households report that they
typically revolve their credit card balances, and 23
percent report that they have either missed or made
late loan payments in the past year.
3. Univariate Analysis of Credit Attitude
Until the recent recession, the personal savings
rate in the United States had been on a steady
decline while consumer borrowing had increased.
Responses to two survey questions can provide
insight into households’ attitude toward the use of
credit to finance consumer spending. One question
addresses borrowing to purchase luxury items: “Is
it all right for someone like yourself to borrow
money to cover vacation expenses or purchase a
fur coat or jewelry?” Another considers borrowing
to meet day-to-day expenses: “Is it all right for
someone like yourself to borrow money to cover
living expenses when income is cut?”
Overall, 15 percent of the households answer
yes to borrowing to purchase luxury items whereas
almost 50 percent respond yes to borrowing to
cover living expenses. The univariate analyses
presented in Tables 3a and 3b illustrate whether
attitude about credit use differs significantly across
household Life Cycle stages and measures of Debt
Capacity and Financial Discipline.
Responses to both survey questions vary
significantly across stage of life cycle. Compared
with the Mature Single, more than twice as many
Young Singles condone borrowing to pay for lux-
ury items. The focus on immediate gratification—
purchasing luxury items on credit rather than
saving for the purchase—is deemed more accep-
table by the young. Borrowing to cover living
expenses is also significantly more acceptable to
those households in younger stages of the life
cycle. Although Single Parents and Young Families
may find it necessary to borrow in the face of a cut
in income to meet the needs of their dependents,
Young Singles and Young Couples have similar
attitudes without the pressure. Older households
are more likely to have accumulated assets that
can be liquidated in the face of income cuts, so they
find borrowing to pay for living expenses less
acceptable.
Examining credit attitude across Debt Capacity
measures reveals insights into what fueled the
recent credit crisis. Households with less debt
capacity—next year’s income is uncertain or they
do not own a home—are less likely to qualify for
consumer credit. Surprisingly, a significantly
greater percentage of these households condone
borrowing to purchase luxury items or to cover
living expenses when income is cut. Perhaps these
attitudes are driven by the abundance of consumer
credit that has been available (until recently) and
the desire to meet current consumption wants and
needs.
Financial Discipline is another factor related to
responses to these credit attitude questions.
Households who do not exemplify sound financial
practices may deem borrowing for any reason to
be appropriate. Almost twice the percentage of
those who carry a balance on their credit cards
indicate that it is all right to borrow to purchase
luxury items compared with those who pay their
FUELING THE CREDIT CRISIS
271
balance in full each month. A higher percentage
of households demonstrating no Financial Dis-
cipline—those who have no savings plan, revolve
their credit balances, or have been late or missed
loan payments—condone borrowing when income
is cut.
4. Multivariate Analysis of Consumer Debt Use
The decision to participate in the consumer
debt market
This section begins with an analysis of the de-
terminants of the household decision to utilize
Table 3. (a) Credit Attitude about Luxury Items by
Household Characteristics; (b) Credit Attitude about
Living Expenses by Household Characteristics
Percent
Responding
Yes
(a)
Is it all right for someone like yourself to borrow money to
cover vacation expenses or purchase a fur coat or jewelry?
Overall sample 15.4
Life Cycle *po0.0001
Young Single 24.5
Young Couple 19.1
Mature Family 18.3
Single Parent 17.8
Young Family 14.4
Mature Couple 13.6
Middle-Aged Couple 13.0
Mature Single 11.6
n=4,167 (251 unclassified)
Debt Capacity
FT Income *po0.0001
Has full-time income 16.5
No full-time income 12.7
Uncertain Income *p=0.0003
Next year’s income is uncertain 16.6
Next year’s income is predictable 14.7
Housing Status *po0.0001
Homeowner 14.8
Renter or other 17.1
Financial Discipline
Saving Rule p=0.95
No saving rule 15.4
Has saving rule 15.4
Revolve *po0.0001
Revolve balance on credit cards 23.5
Pay off balance in full 12.4
n=3,498 (920 have no credit cards)
Late Pay p=0.49
Late/missed loan payment in past year 17.2
All payments on time 16.7
n=3,417 (1001 have no loans)
(b)
Is it all right for someone like yourself to borrow money to
cover living expenses when income is cut?
Overall Sample 49.7
Life Cycle *po0.0001
Single Parent 67.3
Young Single 66.4
Young Couple 63.5
Young Family 62.9
Table 3 (continued )
Percent
Responding
Yes
Mature Family 50.8
Middle-Aged Couple 50.5
Mature Single 42.8
Mature Couple 38.9
n=4,167 (251 unclassified)
Debt Capacity
FT Income *po0.0001
Has full-time income 52.1
No full-time income 43.5
Uncertain Income * po0.0001
Next year’s income is uncertain 54.3
Next year’s income is predictable 47.0
Housing Status *po0.0001
Homeowner 45.8
Renter or other 60.8
Financial Discipline
Saving Rule *po0.0001
No saving rule 53.9
Has saving rule 45.7
Revolve *po0.0001
Revolve balance on credit cards 58.8
Pay off balance in full 42.7
n=3,498 (920 have no credit cards)
Late Pay *po0.0001
Late/missed loan payment in past year 58.2
All payments on time 49.1
n=3,417 (1001 have no loans)
n=4,418 except where noted.
*The percent of households responding “Yes” is statis-
tically different across household characteristics at the 99
percent level of confidence.
Diane K. Schooley and Debra Drecnik Worden
272
consumer credit. Logistic regression is used to es-
timate the probability that households participate
in the consumer debt market. The model assumes
that the household’s choice to have consumer debt
is characterized by a logistic distribution, and the
maximum likelihood estimates of the regression
coefficients yield an estimated probability derived
from the cumulative logistic distribution function.
The odds ratio is the probability that an event
occurs divided by the probability that it does not
occur. In the logit model, the log of the odds is
linear:
log½Pr Consumer Debt=
ð1 Pr Consumer DebtÞ ¼ a þ
X
b
k
x
k
The explanatory variables that are hypothesized
to influence the probability that the household has
consumer debt are denoted by x
k
, and the regres-
sion coefficients from the model are denoted by b
k
.
The estimate of the odds ratio (derived from taking
the exponential of the maximum likelihood esti-
mates, b
k
) indicates the impact that a unit change
in x
k
has on the probability of an event, holding
all other factors constant. An odds ratio of 1.00
indicates equal odds, meaning the explanatory
variable has no significant impact on the event
probability.
5
The results of the estimated model
are presented in Table 4 and are interpreted as
follows.
6
For the indicator variables, the odds ratio es-
timate denotes the marginal effect on the prob-
ability that the household will participate in the
consumer debt market when the variable is turned
on, takes the value 1. If it is not turned on, the
value is 0. For the continuous variables that mea-
sure household net worth and income, the odds
ratio estimate indicates the marginal impact on
the probability that the household will participate
in the consumer debt market given a $1,000 change
in the variable.
As predicted by the PIH/LC models, a house-
hold’s Life Cycle stage significantly impacts its
probability of participating in the consumer debt
market.
7
Younger households (those with median
age under 40) are more than twice as likely to have
consumer debt than Mature Couples. For example,
a Young Single household is 117 percent more
likely, with the confidence interval estimate in-
dicating a 53 percent to 207 percent higher prob-
ability. Young Couples are 129 percent and Young
Families are 145 percent more likely to have
Table 4. Logistic Regression Analysis of Consumer
Debt Use
Explanatory
Variable
Odds Ratio Estimates
p-
value
Point
estimate
95% confidence
interval
estimate
Life Cycle
Young Single 2.17* 1.53–3.07 0.000
Young Couple 2.29* 1.54–3.39 0.000
Young Family 2.45* 1.87–3.20 0.000
Single Parent 2.12* 1.54–2.92 0.000
Middle-Aged Couple 1.52* 1.06–2.17 0.022
Mature Family 1.42* 1.16–1.72 0.001
Mature Single 0.97 0.79–1.19 0.798
Credit Attitude 1.45* 1.27–1.67 0.000
Debt Capacity
FT Income 2.47* 2.09–2.92 0.000
Uncertain Income 0.93 0.81–1.07 0.324
Housing Status 1.21* 1.00–1.45 0.048
Financial Discipline
No Saving Rule 1.30* 1.13–1.50 0.000
Economic Expectations
Higher Interest Rates 1.20* 1.03–1.39 0.018
Lower Interest Rates 1.13 0.86–1.49 0.390
Net Worth ($000) 0.99* 0.99–0.99 0.000
Income ($000) 0.99* 0.99–0.99 0.002
Intercept 0.38* 0.29–0.50 0.000
n=4,167.
*Odds Ratio Estimate differs from 1.00 at a 5 percent
significance level.
The p-value is the observed level of significance for the
maximum likelihood estimates of the regression coefficients,
b
k
.
The chi-square statistics for the likelihood ratio tests in
each of the five imputations are significant at less than the
1 percent level.
5
The confidence interval estimate of the odds ratio—
derived from the parameter estimates and their covariance
matrix—indicates whether the explanatory variable has a sig-
nificant impact at the 95 percent level of confidence. If the
value 1.00 is within the interval, then the estimated coefficient
is not significantly different from zero and the explanatory
variable has no statistically significant impact on the event
probability.
6
The original sample of 4,422 households is reduced to
4,167 for the logit analysis. Besides those eliminated because of
the inability to classify their life cycle, four observations were
excluded from the public database because of concerns about
confidentiality.
7
The most common group in this sample—the older
Mature Couple whose children are no longer at home—is in
the constant.
FUELING THE CREDIT CRISIS
273
consumer debt than Mature Couples. These results,
as expected, reflect that as households move into
their family formation years and children are
added, the demand for consumer spending in-
creases beyond income streams.
A household’s Credit Attitude has a significant
impact on its probability of participating in the
credit market. This finding, given that measures of
wealth, income, stage of life cycle, debt capacity,
and financial discipline are held constant, under-
scores the important role that mindset plays in the
borrowing decision. Households who believe that it
is all right to borrow for vacations and other luxury
items, or to cover living expenses when income is
cut, are 45 percent more likely to participate in the
consumer debt market than those who do not
condone borrowing for these purposes.
Financial Discipline als o significantly impacts
consumer debt use.
8
As expected, households with no
consistent saving rule are significantly more likely
(30 percentage points) to have consumer debt than
those households who do have a saving rule. When
spending wants/needs arise, a household that has
a saving rule is not as likely to participate in the
consumer debt market.
The results reveal opportunities for change in
consumers’ attitudes about borrowing and their
financial discipline. Education that impacts con-
sumers’ attitudes about credit use and increases
financial discipline may prevent a repeat of the
credit crisis. For example, consumers could be
more educated about the length of time required to
pay off a balance given a particular payment, or
about the benefits of saving regularly. Raising
awareness about the difference between needs and
wants and emotions attached to purchases may
provide a different perspective on consumption. As
households’ attitude toward borrowing becomes
more conservative and awareness of the benefits of
saving increases, they may borrow less, regardless
of the loose guidelines and incentives offered by
lenders to entice them.
A household’s Debt Capacity significantly im-
pacts its ability to participate in the credit market.
Households with at least one full-time income
stream are 147 percent more likely to have con-
sumer debt. Because the model is controlling
for household income, the results capture the
willingness of lenders to extend credit and the
household’s willingness to take on debt. According
to the odds ratio, homeowners are 21 percent more
likely to have consumer debt than others. Perhaps
this is a reflection of the use of home equity for
consumption, or the readiness of lenders to extend
credit to homeowners.
At the time of the survey, nearly two-thirds of
the households believed that the economy would
exhibit increasing interest rates in the near future.
Consistent with the PIH/LC models, those house-
holds believing that interest rates will rise are
20 percent more likely to participate in the con-
sumer debt market than those who think that in-
terest rates will remain level. Higher interest rates
often result from inflation, and so households are
more likely to borrow and spend today if they think
prices will rise. Higher interest rates also reflect a
higher opportunity cost of waiting to consume, so
that households consume today.
The analysis of consumer debt burden
This study next examines households’ debt burden,
which reflects the ability to sustain the debt and
repay it. Selecting only those households with
consumer debt, multiple linear regression analysis
is used to test the relationship between consumer
debt burden and stage of life cycle, credit attitude,
financial discipline, debt capacity, and economic
expectations, holding constant the household’s
wealth (net worth) and income. Results are pre-
sented in Table 5.
9
Although the household’s stage of life cycle is
very significant in explaining who participates in
the consumer debt market, it is not a powerful
determinant of the debt burden. Compared with
Mature Couples, only Young Singles have a sig-
nificantly higher consumer debt burden. Younger
households make less income than they expect to in
the future, so they borrow to smooth consumption
over their lifetime. Mature Couples typically earn
more income and are repaying debt, thereby
carrying a significantly lower debt burden than
Young Singles. Although the relationship is not as
strongly significant, Mature Singles have less debt
burden than Mature Couples. Mature Singles have
a higher median age than Mature Couples, and are
8
For the logistic model, only the No Saving Rule measure
of financial discipline is included. The variables Revolve and
Late Pay are undefined for those households with no credit
cards or loan payments.
9
Those households with no income are excluded because
the debt burden is undefined, reducing the sample to 2,086
observations.
Diane K. Schooley and Debra Drecnik Worden
274
half the household size, which may explain their
lower debt burden.
A household’s Credit Attitude drives its debt
burden. As expected, households indicating that it
is all right to borrow to purchase luxury items or to
cover living expenses carry higher consumer debt to
income relative to those who replied no to both
questions. Considering a household’s Financial
Discipline, only the measure of savings behavior is
significant in explaining consumer debt burden.
Households who do not have a consistent saving
rule carry a higher debt burden than those with a
saving rule. As with debt use, education that im-
pacts consumer attitudes about credit and the
benefits of regular saving may go far in reducing
households’ debt burden.
Consumer debt burden is related to a house-
hold’s Debt Capacity. Even while holding income
and wealth constant, households earning a full-
time income maintain a significantly lower con-
sumer debt balance to income. Uncertain income
is positively related to the household’s consumer
debt burden, even though their creditworthiness
and ability to repay is questionable. This result
reflects the eagerness of lenders to make loans as
well as the abundance of credit available even to
high risk borrowers during the time preceding
the survey. Given that the household’s income is
held constant, this result could indicate household
borrowing to smooth out consumption in times
of uncertainty. This result is contrary to the buffer-
stock model prediction that those with uncertain
income carry higher buffer stock, that is, less debt
and more savings.
5. Conclusion
The United States is experiencing the worst reces-
sion since the Great Depression, and many point
to the burden of too much household debt as a
cause. Booming asset markets and easy access to
credit markets led Americans to borrow in record
amounts. Although some believe that mindsets
are fundamentally changed forever and we will
never again experience those levels of borrowing
(and spending), it is important to understand
factors that influenced borrowing so that we can
begin to understand what will shape credit use in
the future.
This study utilizes survey data to examine
household consumer credit use on two dimensions:
factors that determine the likelihood of borrowing
and factors that determine how much is borrowed
relative to income (debt burden). The results
indicate that the likelihood of participating in
the consumer debt market depends upon stage of
life cycle and attitudes toward credit use. Not sur-
prisingly, younger households are more likely to
have consumer debt than mature couples. As
households move into their family formation years
and children are added, the demand for consumer
spending increases beyond income streams. As
households age and enter the later stages of life,
we expect that they will they be less likely to use
credit to meet consumption wants and needs.
More enlightening is the evidence of the sig-
nificant influence of the attitude toward credit use.
Those who believe it is all right to borrow for
luxury items or to cover living expenses are much
Table 5. Regression Analysis of Consumer Debt
Burden for Households with Consumer Debt
Explanatory Variable
Estimated
Coefficient
F
Statistic
p-
value
Life Cycle
Young Single 0.67* 24.61 0.000
Young Couple 0.12 0.75 0.387
Young Family 0.03 0.07 0.786
Single Parent 0.16 1.49 0.223
Middle-Aged Couple 0.07 0.23 0.633
Mature Family 0.03 0.10 0.753
Mature Single 0.17** 2.99 0.084
Credit Attitude 0.14* 6.06 0.014
Debt Capacity
FT Income 0.25* 9.97 0.002
Uncertain Income 0.12* 4.34 0.037
Housing Status 0.06 0.58 0.448
Financial Discipline
No Saving Rule 0.11** 3.60 0.058
Revolve 0.08 1.91 0.167
Late Pay 0.06 0.62 0.432
Economic Expectations
Higher Interest Rates 0.02 0.06 0.813
Lower Interest Rates 0.03 0.09 0.770
Net Worth ($000) 0.00 2.08 0.151
Income ($000) 0.00 2.19 0.139
Intercept 0.49* 15.38 0.000
n=2,086.
Overall F statistic=4.99*.
The adjusted R
2
ranged from 0.033 to 0.036 across the
five separate imputation regressions.
*Significant at the 95 percent level of confidence or
higher. An F statistic, rather than the traditional t-statistic, is
calculated from the estimated parameters and parameter
variances across the five imputations. The p-value is the ob-
served level of significance associated with each F statistic.
**Significant at the 90 to 95 percent confidence level.
FUELING THE CREDIT CRISIS
275
more likely to use consumer credit, and have a
higher consumer debt burden. Another notable
finding is that households with no consistent
savings plan are both more likely to borrow and to
carry a higher debt burden when they do borrow.
Data as of 2010 suggest that “deleveraging”
and “retrenchment” is spreading throughout the
economy. Total consumer credit outstanding has
been declining since the onset of the recession, as
households that had relied on easy access to credit
are paying off debt balances [St. Louis Fed 2010].
Ongoing concerns about job security, along with
reduced housing and financial asset values may
make consumers cautious about returning to past
levels of borrowing and spending. At the same
time, lenders are more restrictive in granting credit.
Consumers with a credit score below 600—an
indication of high default risk—are unlikely to
qualify for additional credit. Historically, 15 per-
cent of those with active credit accounts are in this
category. However, a recent report by FICO Inc.
shows that 25.5 percent of consumers are now at
that level [Li 2010]. Perhaps the thinking about
credit use will shift, or perhaps not. In the second
half of 2009 the savings rate began to decline from
its recession high. The evidence does not clearly
point toward a persistent shift in attitude toward
credit use.
Policies designed to protect the consumer from
unfair and abusive lending practices are prevalent.
For example, new regulations require credit card
issuers to make interest charges easier to under-
stand and due dates more transparent. Mortgages
with zero-down payments or amounts higher
than the home values are quickly disappearing.
Although these changes may serve to protect bor-
rowers, a more sustainable solution may lie, not in
more regulation, but rather in education—inform-
ing consumers about the consequences of borrow-
ing beyond an amount that their income and
wealth can bear. The study’s results support the
need for financial education that changes con-
sumers’ attitudes about credit use and increases
financial discipline. As households’ attitude toward
borrowing becomes more conservative and their
awareness of the benefits of saving increases, they
may disregard lenders’ enticing offers, reducing
debt burdens and default rates.
Acknowledgment
We are grateful to the reviewers whose suggestions
improved our paper.
REFERENCES
Aizcorbe, Ana, Arthur Kennickell, and Kevin Moore.
2003. “Recent Changes in U.S. Family Finances:
Evidence from the 1998 and 2001 Survey of Consumer
Finances.” Federal Reserve Bulletin, 89: 1–32.
Baek, Eunyoung, and Gong-Soog Hong. 2004. “Effects of
Family Life-Cycle Stages on Consumer Debt.” Journal
of Family and Economic Issues, 25(3): 359–85.
Bojanic, David C. 1992. “A Look at a Modernized Family
Life-Cycle and Overseas Travel.” Journal of Travel and
Tourism Marketing, 1(1): 61–78.
Browning, Martin, Angus Deaton, and Margaret Irish.
1985. “A Profitable Approach to Labor Supply and
Commodity Demands Over the Life-Cycle.” Econome-
trica, 53: 503–33.
Carroll, Christopher. 1992. “The Buffer Stock Theory of
Saving: Some Macroeconomic Evidence.” Brookings
Papers on Economic Activity, 2: 61–135.
Carroll, Christopher, and Lawrence Summers. 1991.
“Consumption Growth Parallels Income Growth: Some
New Evidence,” in National Saving and Economic Per-
formance, edited by B.D. Bernheim and J.B. Shoven.
Association for Financial Counseling and Planning
Education, 305–43.
Deaton, Angus. 1991. “Saving and Liquidity Constraints.”
Econometrica, 59(5): 1221–48.
Deaton, Angus. 1992. Understanding Consumption. Oxford
University Press Inc.
Federal Reserve Board, 2007. Survey of Consumer Finances
[Data File], www.federalreserve.gov/pubs/oss/oss2/2007/
scf2007data.html.
Federal Reserve Bank of St. Louis, 2010. http://research
.stlouisfed.org/fred2/ (accessed July 15, 2010).
Friedman, Milton. 1957. A Theory of the Consumption
Function. Princeton University Press.
Gongloff, Mark. 2008. “ “Wealth Effect” May be Near
Payback Time.” Wall Street Journal , (June 5): C1.
Kennickell, Arthur. 1991. “Imputation of the 1989 Survey of
Consumer Finances: Stochastic Relaxation and Multiple
Imputation,” Proceedings of the Section on Survey Re-
search Methods, American Statistical Association.
Laise, Eleanor. 2008. “Tapped Out: Pinched Consumers
Scramble for Cash.” Wall Street Journal, (June 2): A1.
Li, Shan. 2010. “Credit Scores Sink to New Lows.” Los Angeles
Times, (July 12) http://articles.latimes.com/2010/jul/12/
business/la-fi-credit-scores-20100712 (accessed July 15, 2010).
Modigliani, Franco, and Richard Brumberg. 1954. “Utility
Analysis and the Consumption Function: An Interpreta-
tion of the Cross-Section Data,” in PostKeynesian Eco-
nomics , edited by K. Kurihara. Rutgers Universtiy Press.
Rubin, Donald. 1987. Multiple Imputation for Nonresponse
in Surveys. John Wiley and Sons, Inc.
S&P/Case-Shiller Home Price Indices, 2009. http://www2
.standardandpoors.com/spf/pdf/index/csnational_value_
052619.xls (accessed June 12, 2009).
Diane K. Schooley and Debra Drecnik Worden
276
