Content uploaded by Ida Aghdas Mirzaie
Author content
All content in this area was uploaded by Ida Aghdas Mirzaie on Dec 03, 2017
Content may be subject to copyright.
1
CONSUMER DEBT STRESS, CHANGES IN HOUSEHOLD DEBT,
AND THE GREAT RECESSION
Working Paper, Ohio State University Economics Department, 2013
Lucia F. Dunn and Ida A. Mirzaie*
Abstract
Although household debt fell more than income during the Great Recession, we find consumer
debt stress rose. This research tracks household debt to income for eight different types of debt
using unique data from a monthly national U.S. household survey covering the period 2006
through 2012. The rise in stress resulted partly from the shift out of collateralized debt into non-
collateralized debt, which involves more stressful collections practices. Women and Hispanics
experienced higher levels of stress. The impact of a collection agency encounter on debt stress is
found to be approximately 50 percent greater on women than on comparable men. (JEL Codes:
D12, D18)
KEY WORDS: Consumer Debt Stress, Debt Composition, Great Recession, Gender, Collection,
Discrimination
*Dunn (Corresponding Author): Professor, Department of Economics, Ohio State University, 1945 N. High St.,
Columbus, OH 43210, 1-614-292-8071, (dunn.4@osu.edu); Mirzaie: Senior Lecturer, Department of Economics,
1945 N. High St., Ohio State University, Columbus, OH 43210; 614-292-6110 (mirzaie.1@osu.edu).
2
I. INTRODUCTION
In the late 20th century, consumer debt began to rise to unprecedented levels among U.S.
households. Accompanying this, there was an upsurge in reports coming from financial and
family counselors, health care professionals, employers, and others pointing to a wide range of
health, family, and job problems for Americans arising from this increase in debt. After the
onset of the Great Recession and financial crisis, both the income and debt of U.S. households
fell. Our data show that debt actually fell faster than income for households during this period,
with the household debt to income ratio decreasing by 6 percent between 2008 and 2009 and
continuing to drop steadily there thereafter. Between 2008 and 2012 the ratio went from an
average value of 1.20 to 1.02 for a total decline of 15 percent.1 This fall in debt to income ratios
might have been expected to relieve some of the pressure on debt stress. However, the impact
was not straightforward, as the events of the recession brought about many changes related to the
household debt situation, including changes in assets, lending practices by banks, and the
composition of debt for households, moving them out of collateralized (or secured) debts and
toward non-collateralized debt types. For example, while mortgage debt fell, student loan debt
increased. We will examine how eight different debt types changed and the impact of this on
stress, representing these debts both in terms of debt size and monthly payment, both in ratio to
income.
We use data from the Consumer Finance Monthly Survey (CFM), a national survey of
household finances taken on a monthly basis. This survey contains unique measures that directly
quantify stress from debt, as well as unusually detailed data on separate types of debt. We find
1 Source: Consumer Finance Monthly survey. Aggregate figures show a similar decline for household debt to GDP.
3
that stress associated with debt increased by over 50 percent in mid-2009 compared to its 2006
level and had not returned to that earlier level as of the end of 2012.
Several factors were involved in the increase in debt stress, including the aforementioned
change in the composition of debt, moving households toward non-collateralized debt. Our
analysis shows that: (a) collateralized debts are less stressful for consumers than non-
collateralized debts, due in part to differences in collection practices that are generally more
stressful for non-collateralized debts; (b) declines in assets and reduced access to liquidity
provided by credit cards increase debt stress; and (c) women and Hispanics have been more
severely affected by the debt stress phenomenon than men and other ethnic groups. This last
point may also be connected to differences in collection practices and penalties, which top the
list of complaints2 to the Federal Trade Commission and which some reports suggest are tougher
for those viewed as vulnerable. We examine the effect of having debts sent to a collection
agency and find that the impact is approximately 50 percent greater on women than comparable
men.
The increase in debt in the U.S. at the end of the last century was partly due to the
introduction of new debt instruments by banks (i.e., credit cards, helocs, etc.) that made it
possible for a greater range of households to acquire debt. Some of the increase was due to
government policies, as in the case of student loans. The events of the recent recession brought
about significant regulatory reform for certain types of household debt. These events were also
instrumental in creating the Consumer Financial Protection Bureau. As the U.S. economy moves
into a new era of oversight of consumer debt markets, an understanding of the impact of debt in
its many forms for different segments of the population will become increasingly
2 U.S. Debt Collection Agencies: An Industry Analysis, April 1, 2012.
4
important. The data used here allow us to determine changes in household debt and their impact
on consumer wellbeing at a more detailed level than previously seen in the literature. Overall,
our analysis shows that the recession caused an upward structural shift in debt stress (after
controlling for socioeconomic factors including income) with which households are still
struggling.
In what follows, we first discuss background and previous research connecting debt to
general wellbeing. We then give details of the new survey and how the concept of debt stress is
measured for the present research. We present our empirical analysis of the determinants of debt
stress along with a discussion of the actual changes in debt composition for households during
the recession. We probe other factors that are found to be significant for debt stress. In
particular, we examine stress by gender and collection agency encounters.
II. BACKGROUND AND PREVIOUS RESEARCH
As household debt began to rise with the spreading use of credit cards, various research
efforts to quantify and analyze mental health impacts linked to debt were undertaken. A
significant effort has been carried out by researchers in the UK to understand the relationship
between debt and psychological health. This includes work by Lea, Webley, and Levine 1993);
Lea, Webley, and Walker (1995); Brown, Taylor, and Price (2005); and Bridges and Disney
(2010). These studies generally find that measures of psychological well-being are significantly
lower for those with debt as opposed to those with little or no debt. Efforts to investigate the
psychological health of debtors have also been carried out for Italian and Australian households
by Malgarini et al (2008) and Worthington (2003) respectively. In the U.S., Berger, Collins, and
Cuesta (2013) have used data from the National Survey of Families and Households to find a
relationship between household debt and depressive symptoms as measured by a standard
5
depression scale. Hurd and Rohwedder (2010) present information on the increase in financial
distress as defined by unemployment and mortgage problems between 2008 and 2010 in Rand’s
American Life Panel survey.
The earlier researchers who have established a link between indebtedness and
psychological well-being generally use stress measures that are not specifically tied to debt in the
survey instrument. One of the first efforts to generate data to connect stress directly to debt in a
survey instrument came in the late 1990s with the work of Lavrakas, et al. (2000) and Drentea
and Lavrakas (2000). These measures will be used in this work, as they identify stress
specifically arising from debt. Movements in the Michigan Index of Consumer Sentiment have
recently been linked to these debt stress measures by Ekici and Ozbeklik (2013).
Physical health impacts resulting from debt stress have also been studied. Lavrakas and
Tompson (2009) find that stress from debt is related to health conditions such as ulcers, heart
attacks, and migraine headaches. Using the Survey of Consumer Finances, Lyons and Yilmazer
(2005) find that poor health contributes to financial strain, and the reverse causality holds for
lower income groups. Currie and Tekin (2011) find that mortgage foreclosures are associated
with increased hospital visits for a broad array of mental and physical health conditions. Further
studies linking credit card debt in particular to negative health outcomes include Drentea (2000)
for anxiety; Spinella, Yang, and Lester (2004) in the area of neuroscience for dysfunction
associated with credit card debt; and Nelson, et al. (2007) for unhealthy conditions among
college students with credit card debt.
In the financial counseling/personal finance literature, work on these issues has been
done by Bagwell (2001) for financial distress of credit counseling clients; Garman and
colleagues for a financial wellbeing scale (Prawitz, et al. 2006); and Bagwell and Kim (2003),
6
Kim et al. (2006) and Dew (2007) for problems including job absenteeism and marital
dissatisfaction linked to debt.
Finally, although collection agency encounters give rise to many complaints and are
obviously problematic for households with debt, relatively little research has been done by
economists in this area. A notable exception is the work of Finkelstein, et al. (2012), which finds
that declines in collection encounters are one outcome, together with improvements in mental
health, for the subjects enrolled in the Oregon Health Insurance Experiment.
With the collapse of the subprime mortgage market and ensuing financial crisis, the sharp
increases in household debt in the U.S. were brought to a halt and reversed. Some of the decline
in debt holding was lender-initiated as banks slashed credit and proceeded to close various loan
accounts. According to credit counselors, some of the decline was consumer-initiated, as
households proactively tried to draw down their debt and rein in borrowing. Decreases in debt
aggregates mask changes in the composition of debt holdings which are central to our findings.
The present study is, to our knowledge, the first to identify relative stress associated with a range
of particular debt types, allowing us to understand the changes in debt stress throughout the
recession and the consequences of this for different groups and markets in the society.
III. The Survey
A. Data
The Consumer Finance Monthly (CFM) survey began in 2005.3 It is a national-level,
monthly household telephone survey that uses random digit dialing techniques. Its data are
3 The Consumer Finance Monthly is administered by the Center for Human Resource Research at the Ohio State
University.
7
weighted using information from the Current Population Survey to reflect national
socioeconomic characteristics.
The survey asks a wide range of questions pertaining to household finances, including all
categories of debt, income, and assets. It thus provides a comprehensive look at critical aspects
of household financial condition on an ongoing monthly basis. It contains information on
several debt types that are difficult to obtain through aggregate data sources, such as payday
loans and loans from family and friends. The questions on debt stress were added to the survey
in 2006, and through December 2012, a total of 19,936 completed cases have been obtained on
the debt stress measures.
B. Measuring Debt Stress
The set of CFM survey questions designed to examine consumer stress from debt are
presented in Appendix A. These questions were developed and tested to directly quantify stress
from debt in a four-item scale.4 The questions refer to the individual respondent’s stress from all
consumer debt sources that they (and spouse/partner) owe. Using standard five-point response
categories, the survey questions elicit: (a) frequency of worry over debt; (b) amount of stress
from debt; (c) extent of expected problems from debt over the next five years; and (d) concern
about the inability to ever pay off debt. The responses are aggregated into a debt stress score
which then forms the dependent variable in our econometric analysis See Appendix A for
details.
4 The four components, which form a score, have been studied in reliability tests and found to have a high level of
internal consistency as well as good face validity. See Appendix A for details.
8
C. Historic Path of the Consumer Debt Stress Scores
For ease of graphical interpretation, we have aggregated the debt stress scores into a
monthly index based to the period January 2006, so that the index has a value of 100 in that
period. Subsequent movements can then be tracked relative to that base value. Computational
detail for the index is presented in Appendix A.
Figure 1 presents a graph of the aggregate consumer debt stress scores over time. We see
there that the path of debt stress remained fairly low in the early years of the survey, with values
hovering around the January 2006 base value of 100 throughout most of 2006 and the first part
of 2007. Consumer debt stress was at its historic low point of 88 in June 2007. The collapse of
the subprime mortgage market brought a turnaround of consumer debt stress in the third quarter
of 2007 and a subsequent upward trend. In September 2007, consumer debt stress in the nation
had increased to a value of 105. Thereafter debt stress fluctuated, but on an upward trajectory, as
shown in Figure 2, as the economy slid into recession. Another discernible up-tick in consumer
debt stress for U.S. households was recorded in the early fall of 2008 amid the turmoil in
financial markets brought about by the bankruptcy of Lehman Brothers and a general tightening
of credit conditions. Debt stress as measured in this survey reached its historic peak in July 2009
at a value of 155.5 This means that consumers were experiencing 55 percent more debt stress in
July 2009 than they were in January 2006. It is not surprising that stress from debt would
increase in a declining economy when incomes are falling. However, household debt dropped at
a considerably faster pace than household income as both consumers and banks took steps that
reduced the debt levels of households. One might have expected these changes to lower debt
stress in the population, but they did not for reasons explored below. As the economy began to
5 This roughly corresponds to trough of the Great Recession according to the NBER.
9
recover, Table 1 shows that debt stress trended downward. By 2012, the household debt to
income ratio had dropped to 1.02 from its 2008 peak of 1.20. Nevertheless, consumer debt stress
had still not returned to its level before the economic crisis.
Figure 1
The Path of Overall Consumer Debt Stress
IV. Determinants of Debt Stress
To investigate factors that give rise to differences in debt stress across the population, we
examine the debt stress scores in a regression model where we include socioeconomic and
demographic factors as well as debt-related variables. A two-sided tobit model is used since the
values of our debt stress scores by definition are constrained to be between zero and four.
To determine the impact of different types of debt, we consider eight types of debt as
explanatory variables: mortgage debt; home equity loans and lines of credit; credit card debt;
student loans; installment debt (auto, appliance, furniture); bank, insurance, and brokerage loans;
10
payday loans; and loans from family, friends and other sources. Each of the debt types examined
naturally includes a variety of loans with different characteristics. While here we represent the
different debts by their size to income, we have also carried out an investigation using monthly
payment to monthly income to represent each debt.6 See Appendix D for this discussion.
We control for additional financial factors in the household that may affect stress. One is
access to liquidity, which we enter as the amount of unused credit card credit limit for a
household in ratio to household income. The log of household assets is also included as an
explanatory variable. Increases in both assets and unused credit card line should presumably act
to reduce credit constraints and stress on a household.
Relevant socioeconomic and demographic variables are entered, of which gender and
ethnic background will be of particular interest in the discussion below. There is not a direct
measure of respondent unemployment in the survey, but we add year fixed effects for the
different years of the survey to capture and control for the macroeconomic environment, as well
as income variables as noted above. Definitions and descriptive statistics for the variables used
in the analysis are given in Appendix C.
V. Empirical Results
Our main empirical results for the determinants of debt stress are presented in Table 1
below. As noted earlier, the individual debt stress scores (ranging from 0 to 4) form the
dependent variable in this tobit analysis.
6 Interest rates may also be considered a debt characteristic, but their effect is contained in the size of the payment.
11
Table 1
Determinants of Consumer Debt Stress Score
Variable
Coefficient
Std error
Intercept
2.55***
(0.14)
Debt-to-income ratios:
Mortgage
0.28***
(0.010)
Heloc (home equity loan/line)
0.35***
(0.039)
Credit card
1.02***
(0.08)
Student loan
0.45***
(0.046)
Installment (auto, appliance, furn.)
0.35***
(0.053)
Bank, insurance, brokerage
0.41***
(0.08)
Payday loan
1.91***
(0.72)
Family and friends, other loans
0.45***
(0.101)
Log assets
–0.106***
(0.0102)
Available credit card line/Income
–0.015***
(0.0035)
Gender (Male = 1)
–0.23 ***
(0.0302)
Marital status (Married = 1)
0.15***
(0.034)
Children present
0.17***
(0.037)
Age
–0.021***
(0.001)
Education (years)
–0.0081
(0.0058)
African American
0.059
(0.063)
Hispanic
0.21**
(0.072)
Other ethnic/racial groups
–0.073
(0.064)
Year 2007 (omitted year: 2006)
0.058
(0.056)
Year 2008
0.25***
(0.052)
Year 2009
0.45***
(0.060)
Year 2010
0.33***
(0.054)
Year 2011
0.27***
(0.053)
Year 2012
0.36***
(0.056)
*** Significant at 1 percent level; ** Significant at 5 percent level; n = 8,723
12
Results by Debt Type
The coefficients of the debt amount-to-income ratios are found to have a statistically
significant impact (at better than the 1% level) for each of the eight types of debt.7 There is,
however, considerable variation in the magnitudes of these coefficients, and Table 2 which
follows presents the rank ordering of the debt types in descending order according to their impact
on the debt stress score.
Table 2
Rank of Debt Type by Descending Impact on Debt Stress
1. Payday Loan
2. Credit Card Debt
3. Student Loan
4. Loan from Family & Friends, Other
5. Bank Loan (including Insurance, Brokerage)
6. Installment Loan (Auto, Appliance, Furniture.)
7. Heloc
8. Mortgage
Collateralized vs. Non-Collateralized. Among the characteristics of the different debt types, one
salient identifiable feature that distinguishes the higher-ranked from the lower-ranker debt types
as seen in Table 2 above (and in other results we discuss below) is whether are not the debts are
collateralized. The three most stressful debt types, denoted as Category 1, are non-collateralized:
payday loan, credit card debt, and student loan. By contrast, the three least stressful debt types,
7 Marginal effects at the sample means differ slightly from the reported coefficients and are available from the
authors upon request.
13
denoted as Category 2, are collateralized: mortgage, heloc (home equity loan or line of credit),
and installment loans (auto, appliance, furniture).8
Pairwise comparisons of the coefficients of the debt-to-income ratios typically show
statistically significant differences between the debt types in Category 1 and the debt types in
Category 2. However, the coefficients of debt types within categories are typically not
significantly different. Thus, for example, the stress from credit card debt and student debt are
not significantly different from each other; the stress from mortgage and installment loans are not
significantly different from each other; but the stress from both mortgage debt and installment
loans are significantly different from the stress associated with credit card debt. It should be
noted, however, that payday loans are uniquely stressful and are significantly different from
every other type of loan considered. Appendix E presents the significance levels for Wald tests
of pairwise differences among a selection of debt types.9
The finding of greater stress from non-collateralized debts may be associated with the
fact that non-performing non-collateralized debts often involve stressful encounters with
collection agencies for an indefinite period of time.10 Non-collateralized debt can be difficult or
impossible to erase even with a declaration of bankruptcy, as in the case of student loans. By
contrast, most non-performing collateralized debts can be brought to termination through
repossession/foreclosure and forfeiture of the collateral, with a definite cessation of any
collection efforts. During the Great Recession, most mortgage foreclosures were handled “in-
8 Loans from family and friends are usually non-collateral based, and bank loans usually are collateralized, thus
following the same stress pattern. However, this level of detail is not available in the survey, and thus we will not
categorize these types of debt by collateralization.
9 A complete set of comparisons is available from the authors upon request.
10 Among a subgroup in our sample, we find that an encounter with a collections agency over credit card debt
increased stress by approximately one point on the scale from zero to four (see Appendix F). See Avery, Calem, and
Canner (2003) for an overview of credit and collections data.
14
house” by banks according to set guidelines and without the tactics typically seen with outside
collection agencies. Further discussion of the impact of collection agency encounters for our
sample is found below under gender effects.
Our findings on the high-stress nature of non-collateralized loans are in line with the
findings of at least two earlier studies. Brown, Taylor, and Price (2005) found that measures of
general psychological well-being of household heads are significantly lower for those with some
debt compared to those with no debt, and that the effect comes from unsecured non-mortgage
debt and was not significant for mortgage debt. Berger, Collins and Cuesta (2013) have found
the association between household debt and depression scale measures is driven by short-term,
unsecured debt with little evidence of depressive symptoms for mid- or long-term debt.
In addition to these findings on non-collateralized versus collateralized debt types, we
note that by far the greatest impact on debt stress by amount of debt to income comes from
payday loans, involving almost twice the stress impact as the debt-to-income ratio of the next-
highest-stress loan type. Payday loans are short-term loans—typically for no longer than two
weeks—that carry annual interest charges of several hundred percent, depending on the state
where the loan is issued (Skiba and Tobacman, 2009). Among institutional lenders, it is the type
of loan that carries by far the highest annual interest rate and, by some reports, is the fastest
growing segment of the consumer finance industry. Only one to two percent of the sample
report currently having a payday loan. Thus the mean amount of these debt holdings in the total
population is low; but, as we show in Table 4 below, for those having payday loans, the dollar
amount is substantial.
15
The Tightening of Credit: Changes in Household Debt and Debt Composition
Household debt positions began to change substantially in 2008 as the recession set in.
Some of these changes may have been the result of households deliberately tightening their belts
in the face of hard times. Some were bank-initiated through foreclosure and repossession, as
these measures accelerated sharply. Finally, many households simply found that their access to
credit/debt was being reduced or denied by financial institutions despite the Federal Reserve’s
substantial lowering of interest rates in an effort to combat the recession (Hilsenrath, 2012).
The events of the recession and credit tightening brought with it changes in both the level
of debt holdings and in the composition of those debt holdings for U.S. households. Tables 3
and 4 below present the changes in household debt using the CFM data between 2008 and 2011,
the major years encompassing the Great Recession and ensuing credit crunch.11 Although some
aspects of the recession began at different times in different parts of the country and in different
industrial settings, this time period should capture the transition to a highly credit-constrained
economy.
11 Declines in debt during the Great Recession have also been found in other data sources such as the Survey of
Consumer Finances. See for example, Emmonds and Noeth (2013), Fry (2013), and Bricker et al. (2012). The SCF
was taken in 2007 and 2010, and thus these studies do not correspond to the period covered in our research.
16
TABLE 3
Percentage of Households Holding Specific Debt Types
Debt Type
2008
2011
Percentage
change
Payday loan
1.1 %
1.4 %
+ 27 %
Credit card
34
25
– 26
Student loan
14
17
+ 21 **
Family and
friends, other
3.1
1.9
– 39 ***
Bank loan
4
3.2
– 20 *
Installment loan
31
27
– 13 ***
Heloc
15
10
– 33 ***
Mortgage
46
39
– 15 ***
Debt-All Types
71
66
– 7.0***
Source: Consumer Finance Monthly survey. The figures in these tables are aggregated
from observations taken in the particular year.
*** Significant at 1% level; ** Significant at 5 % level; * Significant at 10% level.
2008 n = 3,038; 2011 n = 2,693
From Table 3 we see that in 6 of the 8 specific categories of debt covered in this study,
the percentage of households holding that type of debt went down significantly between 2008
and 2011. As noted, debt stress rose even as the percentage of households with debt in many
categories was declining. Part of the explanation for this comes from an examination of changes
in the types of debt held and the level of debt holdings in high-stress categories, as shown in
Table 4 below. Declines in the levels of debt held came in the 4 lowest-stress debt types for the
population as a whole: mortgage, installment loan, heloc, and bank loan. Foreclosure and
17
repossession took a heavy toll on these debt instruments, and the generally tight credit climate
meant that it was harder for new borrowers to successfully qualify for these kinds of loans.
The percentage of the population holding student loans in our sample increased by 21
percent, and the mean amount of that particular type of debt in the general population increased
by 47 percent (see Table 4). The issue of student loan debt involves a number of factors, many
of which (for example, rise in enrollments and tuition, decline in state educational subsidies, etc.)
are beyond the score of this study. However, there is evidence which indicates a tradeoff exists
between student loan debt, mortgages, and credit card debt using data from the Survey of
Consumer Finances (Shand, 2013). Additionally, families with college students may have
substituted more student debt into their household debt holdings as access to other types of debt
(such as helocs) dried up during the recession. Thus the change in student debt may be
connected to broader changes in debt composition during this period which shifted households
toward more stressful debt types.
With regard to credit card debt, the percentage of the population with this type of debt did
not change at the 10 percent significance level in the general population, although Table 4 shows
that the mean level of credit card debt holdings in the total population did fall. However, Table 4
also shows that there was a significant increase of 18 percent in the amount of credit card debt
held among those consumers who did in fact carry this kind of debt.
The figures for payday loans increased in all categories presented here, but none of the
differences reach the 10 percent significance level due to the relatively small number of sample
members who held this type of debt (1–2% of the roughly 20,000 person sample).
To summarize the information in Tables 3–4, the major direction of change evident in
these data is away from the relatively low-stress collateralized types of debt and toward the more
18
high-stress non-collateralized debt types. The most recent U.S. recession is unique in that
consumers went into it holding record amounts of debt. The aftermath of events that set in after
2007, including the credit crunch, compounded the familiar recessionary problems, changing the
composition of consumer debt in the population and increasing stress from debt.
Access to Liquidity and Declining Assets
Available credit card credit line is found to be a significant factor, increasing debt stress
as credit line is used up. Financial advisors report that some households resort to their credit
cards as a supplement to income when they experience unemployment and other types of
financial distress. Sullivan (2008) has explored the link between the use of unsecured credit and
earnings shortfalls for low-asset households and finds that although households in the lowest
decile of the asset distribution do not respond in this way, this behavior is seen for the second
and third deciles. Credit from a credit card can serve as a form of precautionary savings (Brito
and Hartley, 1995) and a buffer for households during contractionary periods. During the Great
Recession, many households simply found that the access to liquidity provided by their credit
line on credit cards had been slashed as a tight credit situation developed. Issuance of new cards
was restricted, and introductory offers were down sharply. These factors, together with the
19
TABLE 4
Mean Debt Conditional on Holding Specific Debt Types and for Total Population
Conditional on holding specific debt type
Total population
Debt type
Mean debt
2008
Mean debt
2011
Percentage
change
Mean debt
2008
Mean debt
2011
Percentage
change
Payday loan
$ 1,220
(2,085)
$ 1,679
(2,437)
+ 38 %
$ 13
(249)
$ 23
(346)
+ 77 %
Credit card
7,022
(10,298)
8,305
(12,530)
+ 18 %**
2,550
(7,065)
2,124
(7,299)
– 17 %***
Student loan
19,102
(22,085)
23,581
(30,028)
+ 23 %*
2,724
(10,685)
4,013
(15,230)
+ 47 %***
Family and
friends, other
11,927
(14,802)
9,960
(15,863)
– 16 %
375
(3,348)
188
(2,569)
– 50 %***
Bank loan
28,057
(71,003)
10,566
(34,757)
– 62 %***
1,224
(15,900)
343
(6,535)
– 72 %***
Installment loan
13,270
(10,232)
12,042
(9,485)
– 9 %***
4,172
(8,419)
3,273
(7,290)
– 22 %***
Heloc
34,236
(29,973)
29,320
(26,801)
– 14 %***
4,990
(16,640)
2,933
(12,216)
– 41 %***
Mortgage
138,556
(129,146)
134,264
(105,149)
– 3 %
63,838
(11,599)
52,586
(92,873)
– 18 %***
Debt-All Types
112,741
(136,589)
100,077
(111,992)
– 11%***
79,630
(125,754)
66,245
(102,681)
– 17 %***
Source: Consumer Finance Monthly survey. The figures in these tables are aggregated from observations taken in the particular year.
Standard deviations in parentheses.
*** Significant at 1% level; ** Significant at 5 % level; * Significant at 10% level
20
rising credit card debt for those having such debt (as seen in Table 4) meant that access to the
liquidity provided by credit cards was dwindling for these consumers. This result, which
emphasizes the importance of credit card liquidity to consumers, is in line with the findings of
Gross and Souleles (2002), Cohen-Cole and Morse (2010), and Wang (2012).
The negative coefficient on assets suggests a similar phenomenon in that assets also
provide another source of potential access to liquidity (for example, through helocs) and thus act
as a buffer for consumers. At the bottom of the Great Recession, asset values among the sample
had dropped, as house prices and stock market values declined sharply and many consumers lost
the equity in their homes through foreclosure. For our sample during the recession, mean assets
dropped from $455,744 in 2008 to $384,366 in 2011, a decline of 16 percent; and median assets
went from $239,350 in 2008 to $192,000 in 2011, a 20 percent decline. Similar patterns of
declines in assets are also documented in the Survey of Consumer Finances in the period from
2007-2010 (Bricker et al., 2012). These changes contributed to the rise in debt stress observed in
our study.
Results by Gender
One of the most persistent differences in our results occurs between the genders. In
Figure 3 below, we have plotted separately the monthly debt stress scores of men and women in
our sample across time, again in indexed form based to January 2006. The plot is for the period
starting in January 2006 and running through December 2012. While the plots of both men and
women show the same general pattern that was exhibited in the plot of overall debt stress, the
level of debt stress for women throughout this period is persistently highly than that of men.
21
Thus in general, while the debt stress of both genders moved in similar ways with economic
conditions, the response is found to be greater for women. Note that women carry somewhat less
debt than men, but they have considerably less income so that their debt-to-income ratio is higher
(see Appendix E). Our results were obtained controlling for the ratio of debt-to-income in each
debt category. Thus for given debt to income ratios, women exhibit more stress than men.
Furthermore, the gender difference is not explained by differences in the composition of debt, as
men hold more debt than women in every category except payday loans.
While one could speculate that the observed gender difference may result from the fact
that women express their feelings and emotions more freely than men, previous research which
adjusts for these factors concludes that the differences in distress expressed by women are indeed
genuine (Mirowsky and Ross, 1995). Some research suggests that women may actually
understate their negative feelings to conform to feminine display rules which require the
suppression of negative emotions (Simpson and Stroh, 2004).
Collections. The result on gender may be related to the well-documented finding that women
have higher risk aversion than men, especially in financial matters. (See Croson and Gneezy
(2009); Eckel and Grossman (2008); Jianakoplos and Bernasek (1998)). A higher aversion to
risk would naturally tend to make women more stressed over debt. Both the higher risk aversion
and debt stress which women express in surveys may be connected to historic problems they
have encountered in financial markets. One of these involves collections encounters. Although
access to credit is regulated and overseen by banking authorities with the intent of insuring equal
access by gender, ethnic group, etc., collection practices are more difficult to police and vary
widely by institution. Federal regulations outlining permissible collection procedures can be
22
found in the Fair Debt Collection Practices Act, the Consumer Credit Protection Act, and the
Fair Credit Reporting Act. These laws attempt to set guidelines about number of contacts per
FIGURE 3
Debt Stress by Gender
day, divulging information to neighbors, the use of threats, and other techniques that collection
agents have been known to use. State Laws governing the collections process vary, and accounts
of dubious collections practices are widespread.12 Little systematic research has been done on
the specifics of collections practices, but investigative reporters have presented evidence that
some collections specialists use more heavy-handed tactics with debtors who are more
vulnerable, including women (Silver-Greenberg, 2011; Chan 2006).13
Collections. The CFM survey provides some empirical evidence on the collections issue. A
subsample of our respondents who had reported being at least 60 days late on a credit card
payment in the last six months were additionally asked if any of the accounts had been sent to a
12 U.S. Debt Collection Agencies: An Industry Analysis, April 1, 2102.
23
collection agency during that period. A probit analysis of having or not having one’s debt sent to
a collection agency revealed that women were not more likely than men in the same
circumstances to have their delinquent credit card accounts sent to a collections agency.
However, examining debt stress scores using an interaction term between “gender” and “having
a debt sent to a collections agency” in a tobit analysis we find that the impact of a collections
encounter on women is approximately 50 percent higher than it is for comparable men. These
fits are presented in Appendix F.
Debt Stress Results for Other Socioeconomic Variables
Stress is higher among married consumers and higher when children under the age of 18
are present in the household. Various specifications were tried interacting marital status and
gender with presence of children. The critical factor in all cases was found to be presence of
children.14 Contrary to some popular sentiment about the stresses of single female-headed
households, the married parent was as stressed as the single parent across both genders. The
most significant stressful family structure variable for both genders is the presence of children.
Debt stress was found to decrease with advancing age. Previous work using the CFM
survey data has found younger cohorts taking on non-collateralized debt (i.e., credit card) at
higher rates and paying off at lower rates than older cohorts (Jiang and Dunn, 2013). Thus
younger people face a debt horizon with more uncertainty than older cohorts, and this may
explain some of their greater debt stress. Furthermore, the persistent high level of
unemployment in the economy in recent years has affected the young more severely and made
debt more problematic for them. Young people have also incurred significant amounts of
13 Research on collection agents, who historically have been mostly male, shows that they are deliberately recruited
for harsh personality characteristics (Sutton, 1991).
14 Xiao and Rao (2011) find the presence of children to be important in explaining consumer debt delinquency.
24
student loan debt, and these debts have been found to interfere with first home buying as well as
marriage and family formation (Shand, 2008). These factors as well have contributed to debt
stress for young people.
The coefficient on years of education was negative, as might be expected. However,
education did not reach significance at the 10 percent level, although education has been linked
to greater financial literary and more favorable borrowing terms (Huston, 2012). Among the
different ethnic/racial groups considered, only Hispanics show significantly more debt stress than
whites. This finding again raises questions about possible differences in collections/penalty
practices for this group, since as noted, some observers have raised concerns that collection
agencies are more likely to use “hard ball” tactics with those viewed as vulnerable. The
coefficient for African Americans is positive but insignificant. The “other” category consisted
primarily of Asians and shows a negative but insignificant coefficient. The smaller sample size
for the subgroup probed about collection agencies prevents us from effective testing for the
impact of collection agency encounters by ethnic/racial background.
Debt stress moved in step with general conditions in the economy as seen from the
coefficients of the year fixed effects, which should capture macro environment impacts not
directly controlled. However, the year fixed effects show that the upward shift in debt stress had
still not returned to its pre-recession levels as of the end of 2012 despite other signs of economic
recovery.
VI. SUMMARY AND CONCLUSIONS
Debt poses a problem for many American households in not only in economic terms but
in terms other measures of wellbeing. This research has examined changes in debt, its
composition, and the accompanying debt stress during the Great Recession. It uses data from a
25
nationwide household survey, the Consumer Finance Monthly, with questions which specifically
identify the stress impact of debt on households from 2006 through 2012. The measures
developed from these questions show that despite household debt falling faster than income
during the recession, debt stress in the general population increased, reaching record levels in
2009.
Several factors are found to be involved in this. One was the change in the composition
of aggregate debt holding as households shifted away from collateralized debts and towards non-
collateralized debts. Examining the impact of eight different types of household debt on stress
both in terms of debt size to income and monthly payment to income, our analysis shows that
collateralized debts are less stressful for consumers than non-collateralized debts. Between 2008
and 2011, the percentage of households holding three major types of collateralized debt—
mortgage, heloc, and installment (auto, appliance, furniture)—fell significantly, while the
percentage holding student loans (non-collateralized) rose. In addition, the magnitude of debt
conditional on holding specific debt types rose significantly for credit card debt as well as for
student loans and fell significantly for collateralized installment loans and helocs.
The finding of greater stress for non-collateralized debts is tied in part to differences in
collection practices. The collection procedures for non-performing collateralized debts can
involve stressful encounters with collection agencies for an unpredictable period of time into the
future. By contrast, collateralized debts can usually be brought to termination through
repossession/foreclosure and forfeiture of the collateral, at which point any collection efforts
cease.
Two other factors found to cause increases in debt stress were the decline in assets and
reduced access to liquidity provided by credit cards during the period of our study. The year
26
fixed effects show a significant upward shift in debt stress beginning in 2008 which had not
returned to its pre-recession levels as of the end of 2012 despite other signs of economic
recovery.
Finally, our research has found a persistent difference in debt stress between the genders,
with women showing higher stress than men. This is true even after controlling for debt-to-
income ratios, assets, and other socioeconomic variables and is also related to collections
practices. While men and women in our sample are found to have the same probability of having
a credit card debt sent to a collections agency, the impact of a collection agency encounter on the
stress of women is approximately 50 percent greater than that of comparable men. Among
different racial/ethnic groups, debt stress is also found to be higher for Hispanics. Further
research is needed to determine if women and Hispanics are likely to be subject to more
aggressive collections practices and/or penalties than comparable males and non-Hispanics.
REFERENCES
Avery, Robert, Paul Calem, and Glenn Canner, ‘‘An Overview of Consumer Data and Credit
Reporting,’’ Federal Reserve Bulletin, 89 (2003), pp. 47–78.
Bagwell, D.C. and J. Kim. “Financial Stress, health Status, and Absenteeism in Credit counseling
Clients.” Journal of Consumer Education, 2003, Vol. 20, pp. 50-58.
Berger, Lawrence, J. Michael Collins, and Laura Cuesta. “Household Debt and Adult
Depressive Symptoms.” SSRN: http://ssrn.com/abstract=2200927, January 2013.
Bricker, Jesse, Arthur Kennickell, Devin Moore, and John Sabelhaus. "Changes in U.S. Family
Finances from 2007 to 2010: Evidence from the Survey of Consumer Finances." Federal
Reserve Bulletin, 2012, Vol. 98, pp. 1-80.
Bridges, Sarah and Richard Disney. “Debt and Depression.” Journal of Health Economics, May
2010, Vol. 29, No. 3, pp. 388-401.
Brito, L.D. and Peter R. Hartley. Consumer Rationality and Credit Cards. The Journal of
Political Economy, 1995, Vol. 103, No. 2, pp. 400-433.
27
Brown, Sarah, Karl Taylor, and Stephen Price. “Debt and distress: Evaluation the Psychological
Cost of Credit.” Journal of Economic Psychology, October 2005, Vol. 26, No. 5, pp. 642-663.
Chan, Sewell. "An Outcry Rises as Debt Collectors Play Rough." The New York Times. July 5,
2006. http://www.nytimes.com/2006/07/05/nyregion/05credit.html.
Cohen-Cole, Ethan and Jonathan Morse. May 2010. “Your House or Your Credit Card, Which
Would You Choose? Personal Delinquency Tradeoffs and Precautionary Liquidity Motives.
Working paper, SSRN: http://ssrn.com/abstract=1411291.
Croson, Rachel and Uri Gneezy. “Gender Differences in Preferences.” Journal of Economic
Literature, June, 2009, Vol. 47, No. 2, pp. 448-474.
Currie, Janet and Erdal Tekin. “Is the Foreclosure Crisis Making Us Sick?” NBER Working
Paper No. 17310, August 2011.
Dew, Jeffrey. “Two Sides of the Same Coin? The Differing Roles of Assets and Consumer Debt
in Marriage.” Journal and Family and Economic Issues, March 2007, Vol. 28, No. 1, pg 89-104.
Drentea, Patricia. “Age, Debt, and Anxiety.” Journal of Health and Social Behavior. December
2000, Vol. 41, No. 4, pp. 437-450.
Drentea, Patricia, and Paul J. Lavrakas. “The Association among Health, Race, and Debt.”
Social Science and Medicine. February 2000, Vol. 50, No. 4, pp. 517-529.
Eckel, Catherine C. and Philip J. Grossman. “Men, Women and Risk Aversion: Experimental
Evidence.” In Handbook of Experimental Economics Results, Vol. 1, Ed. Charles Plott and
Vernon Smith. New York: Elsevier, 2008, pp. 1061-1073.
Ekici, Tufan and Serkan Ozbeklik. “Financial Distress and Consumer Confidence: Are There
Any Animal Spirits?” Working Paper, METU NCC, 2013.
Emmons , William R. and Bryan J. Noeth. “Why Did Young Families Lose So Much Wealth
during the Crisi: The Role of Homeownership.” Federal Reserve Bank of St. Louis Review,
January/February 2013, Vol. 95, No.1, pp. 1-26.
Finkelstein, Amy, Sarah Taubman, Bill Wright, Mira Bernstein, Jonathan Gruber, Joseph P.
Newhouse, Heidi Allen, and Katherine Baicker. “The Oregon health Insurance Experiment:
Evidence from the First Year.” The Quarterly Journal of Economics, 2012, Vol.127, No.3, pp.
1057-1106.
Fry, Richard. Young Adults after the Recession: Fewer Homes, Fewer Cars, Less Debt.
http://www.pewsocialtrends.org/2013/02/21/young-adults-after-the-recession-fewer.
Gross, David B. and Nicholas S. Souleles. “Do Liquidity Constraints and Interest Rates Mater
for Consumer Behavior? Evidence from Credit Card Data.” The Quarterly Journal of
Economics, 2002, Vol. 117, No.1, pp. 149-185.
28
Hilsenrath, Jon. “Fed Wrestles with How Best to Bridge U.S. Credit Divide.” The Wall Street
Journal, Vol. CCLIX, No.142, June 19, 2012, page A1.
Hurd, Michael and Susann Rohwedder. 2010. Effects of the Financial Crisis and Great Recession
on American Households.” NBER Working Paper 16407. Cambridge Mass.: National Bureau of
Economic Research.
Huston, Sandra. “Financial Literacy and the Cost of Borrowing.” International Journal of
Consumer Studies, 2012, Vol 36, pp 566-572.
Jianakoplos, Nancy Ammon and Alexandra Bernasek. “Are Women More Risk Averse?”
Economic Inquiry, October 1998, Vol. 36, No. 4, pp 620-630.
Jiang, Sarah and Lucia Dunn. “New Evidence on Credit Card Borrowing and Repayment
Patterns.” Economic Inquiry, January 2013, Vol. 55, No. 1, pp. 394-407.
Kim, Jinhee, Sorhaindo Benoit and E. Thomas Garman. “Relationship between Financial Stress
and Wrokplace Absenteeism of Credit Counsiling Clients.” Journal of Family and Economic
Issues, Fall 2006, Vol. 27, No.3, pp. 458-478.
Lavrakas, Paul J., and Trevor N. Tompson. “New Research on the Measurement of Debt Stress
and Related Health Problems.” American Association for Public Opinion Research, May 2009.
Lavrakas, Paul J., Lucia Dunn, Jeff,Stec, and T.H. Kim. “ A Debt Stress Index for Measuring
the Stress Associated with One’s Total Debt.” The American Statistical Association
Proceedings, 2000, pp. 78-81.
Lea, Stephen, Paul Webley and R.M. Levine. “The Economic Psychology of Consumer Debt.”
Journal of Economic Psychology. March 1993, Vol. 14, No. 1, pp. 85-119.
Lea, Stephen, Paul Webley, and Catherine Walker. “Psychological Factors in Consumer Debt:
Money Management, Economic Socialization, and Credit Use.” Journal of Economic
Psychology. 1995, Vol. 16, pp. 681-701.
Lyons, Angela, and Tansel Yilmazer. “Health and Financial Strain: Evidence from the Survey
of Consumer Finances. Southern Economic Journal. April 2005, Vol. 71, No. 4, pp. 873-890.
Manning, Robert D. Credit Card Nation: The Consequences of American’s Addition to Credit.
New York: Basic Books. 2000.
Malgarini, Marco, Maurizio Pugno, and Gennaro Zezza. “Life Satisfaction in Italy: Evidence
from the ISAE Consumers Survey.” Rivista di Politica Economica, September-October, 2008.
Vol. 98 (9-10), pp. 325-56.
29
Mirowsky, John and Catherine Ross. “Sex Differences in Distress: Real or Artifact?”
American Sociological Review, June 1995, Vol. 60, No. 3, pp. 449-468.
Nelson, Melissa, Katherine Lust, and Ed Ehlinger. “Credit Card Debt, Stress, and Key Health
Risk Behaviors Amond College Students.” American Journal of Health Promotion.
July/August 2008, Vol 22. No. 6, pp. 400-407.
Prawitz, Aimee D., E. Thomas Garman, Benoit Sorhaindo, Barbara O’Neill, Jinhee Kim, and
Patricia Drentea. “InCharge Financial Distress/Financial Well-Being Scale: Development,
Administration, and Score Interpretation.” Financial Counseling and Planning, (2006), Vol. 17,
No. 1, pp. 34-50.
Silver-Greenberg, Jessica. “For the Families of Some Debtors, Death Offers No Respite.” The
Wall Street Journal, December 3, 2011, Vol. 258, No. 131, pp A1 and ff.
Simpson, Patricia A. and Linda K. Stroh. “Gender Differences: Emotional Expression and
Feelings of Personal Inauthenticity.” Journal of Applied Psychology, 2004, Vol. 89, No.4, pp.
715-721.
Sutton, R.I. “Maintaining Norms about Expressed Emotions: The Case of Bill Collectors.”
Administrative Science Quarterly, 1991. Vol.36, pp 245-268.
Skiba, Paige and Jeremy Tobacman. “Do Payday Loans Cause Bankruptcy?” Vanderbilt Law and
Economics Research Paper, Nov. 9, 2009 No. 11-13.
Shand, Jennifer. “The Influence of Credit Constraints, Educational and Personal Debt on
Homeownership among Young Households.” Virginia Polytechnic Institute and State University
Working Paper, 2013.
Spinella, Marcello, BijouYang and David Lester “Prefrontal System Dysfunction and Credit Card
Debt.” International Journal of Neuroscience, 2004, Vol. 114, No. 10 , pp. 1323-1332 .
Sullivan, James X. “Borrowing During Unemployment.” The Journal of Human Resources,
2008, Vol. XLIII, No.2, pp. 383 – 411.
Wang, Hui. “Strategic Consumer Default: Mortgage Versus Consumer Debt.” Ohio State
University Doctoral Dissertation, 2012.
Worthington, Andrew. 2003. “Debt as a Source of Financial Stress in Australian Households.”
School of Economics and Finance Working Papers Series 164, 2003, Queensland University of
Technology.
U.S. Debt Collection Agencies: An Industry Analysis.
http://www.marketresearch.com/Marketdata-Enterprises-Inc-v416/Debt-Collection-
Agencies-6884999/ April 1, 2012.
30
Xiao, Jing and Rui Rao. “Consumer Debt Delinquency Over Life Cycle Stages.” Networks
Financial Institute Working Paper No. 2011-WP-18, August 2011.
APPENDIX A
The Survey
The Consumer Finance Monthly is a national telephone survey based on Random Digit
Dialing techniques which began in February 2005 and is on-going. It is specifically aimed at
collecting household financial information and covers all consumer debt instruments as well as
complete savings/asset/income information and a wide range of demographic variables. About
200–300 new cases are added each month. The responses of the sample are weighted to more
accurately reflect the socioeconomic make-up of the general population.
Survey Questions
The wording of the survey questions is as follows.
1. Overall, how often do you worry about the total amount you (and your spouse/partner) owe in
overall debt? Would you say you worry (a) all of the time; (b) most of the time; (c) some of the
time; (d) hardly ever; or (e) not at all?
2. How much stress does the total debt you (and your spouse/partner) are carrying cause to you?
It is (a) a great deal of stress; (b) quite a bit; (c) some stress; (d) not very much; or (e) no stress at
all?
3. Now, thinking ahead over the next five years, how much of a problem, if any, will the total
debt you (and your spouse/partner) have taken on be for you? Will it be (a) an extreme problem;
(b) a large problem; (c) medium problem; (d) small problem; or (e) no problem at all?
4. How concerned are you that you (and your spouse/partner) will never be able to pay off these
debts? Are you (a) very much concerned; (b) quite concerned; (c) somewhat concerned; (d) not
very concerned; or (e) not at all concerned?
31
Construction of Debt Stress Scores and Index
The debt stress scores used in our analysis are derived by aggregating the responses to the
four stress item questions for individual sample members to form a respondent score. 15 This is
done by assigning numbers based on the 5-point scale from zero (meaning no debt stress) to four
(the highest level of debt stress) to each response. We then average these values across the four
stress item categories. The resulting scores form the dependent variable in our econometric
analysis.
Debt Stress in indexed form is obtained by averaging the debt scores across the n
individuals in the sample in a given month. These monthly average scores are rescaled so that
the Debt Stress Index takes the value 100 in the base period, January 2006 (the initial period of
available data). The computation of the index can thus be represented as follows:
4,3,2,1,
4
1iiiii XXXXDS
n
ii
DS
n
DSI
1
1
946.0
100
where Xi,j is the reported stress by individual i for the category j. The divisor 0.946 represents
the raw average score in the base period. To reduce sampling variation, we have used a three-
month rolling average of the index in the time-series plots shown in the text above. (The
regression analysis uses the individual scores
i
DS
.)
15 The four components of the score have been found to have a high level of internal consistency and strong validity.
For an item analysis of the questions, see Lavrakas, et al., (2000).
32
APPENDIX C
Variable Definitions and Summary Statistics
Variable
Variable definition
Mean
(S.D.)
Mortgage Debt/Income
Mortgage debt to total household
income
0.78
(1.39)
Heloc Debt/Income
Home equity loan/line debt to total
household income
0.061
(0.34)
Credit Card Debt/ Income
Credit card debt to total household
income
0.058
(0.29)
Student Loan Debt/Income
Student loan debt to total household
income
0.076
(0.36)
Installment Debt/ Income
Auto, appliances, furniture, and other
debt to total household income
0.080
(0.31)
Bank Loan Debt/ Income
Loans from banks, insurance policies
or brokers to total household income
0.014
(0.16)
Payday Loan Debt/ Income
Payday loans to total household
income
0.00081
(0.018)
Family/Friends, Other Debt/
Income
Loans from family, friends, and
similar to total household income
0.0078
(0.14)
Log(Assets)
Total assets (physical & financial)
11.86
(1.92)
Available Credit Card
Credit to Income
(Total credit limit – total balance)
/Total income
.65
(16.55)
Gender
Male = 1; Female = 0
0.43
(0.50)
Marital Status
Married = 1; Non-married = 0
0.59
(0.49)
Children Present
Presence of children < 18 years = 1;
no children = 0
0.4006
(0.49)
Education Years
Respondent’s years of schooling
14.62
(2.78)
Age
Age of respondent in years
50.43
(16.24)
African American
1 if respondent is African American;
0 otherwise
0.074
(0.26)
Hispanic
1 if respondent is Hispanic;
0 Otherwise
0.052
(0.23)
Other ethnic/racial groups
1 if respondent is other; 0 otherwise
0.062
(0.24)
*Financial variables refer to respondent and spouse/partner where applicable. Income is annual.
n = 19,936
33
APPENDIX D
Debt Types Represented by Monthly Payment to Monthly Income
Using monthly payment to represent each debt type in our tobit fit, we again find that all
debts have a statistically significant impact on stress. While the results in this case are not
identical, they show a similar trend in stress with regards to collateralization, with most non-
collateralized debts having higher stress than collateralized debts. The rank by descending order
of impact on stress is: family and friends, student loans, payday loans16, heloc, mortgage, bank
loan, installment loan, credit card. The two noteworthy differences using monthly payments
involve credit cards and loans from family and friends. Credit cards debt, where minimum
required monthly payment to monthly income is used, falls to the bottom of the ranking. This
underscores the limited nature of the constraint placed on consumers by the typically low
minimum required payment on credit cards. The circumstance of low minimum required
payments has been an issue between banks and regulators and was addressed in the Bankruptcy
Reform Act of 2005, whose success has been controversial. Loans from family and friends
moves up in the rankings. It is not uncommon for respondents with debts from family and
friends to have no regular monthly payments. Thus the requirement of a regular payment for
these loans may be a contentious factor for such respondents, giving rise to greater stress. The
limitations posed by these factors, as well as the fact that the payday loan position may have
been affected by their not involving the same kind of regular monthly payments as other debts,
suggests that debt amount to income may be a better variable to represent the debt types than
monthly payments. The impact of other explanatory
16 These are typically 2-week loans, and this has been adjusted for in the fits.
34
variables is qualitatively similar in both fits and will not be reported separately. Full results are
available from the authors upon request.
APPENDIX E
Significance Levels for Pairwise Difference in Debt Stress by Debt Type
(Wald Test)
Mortgage Debt /Income
Heloc Debt/Income
Installment
Debt/Income
CC Debt/Income
<.0001
<.0001
<.0001
Student LoanDebt/Income
0.0004
0.1038
0.1632
Payday Loan/Income
0.0237
0.0303
0.0304
APPENDIX F
Gender Effects
Means and Medians of Variables by Gender, 2006–2012
Variables
All
Female
Male
Average debt stress score
1.05
(1.09)
1.16
(1.14)
0.91
(1.00)
Annual income – mean
70,272
(95,689)
63,672
(84,204)
78,855
(108,227)
Annual income – median
50,000
44,880
56,400
Assets – mean
456,294
(908,853)
375,995
(755,068)
557,262
(1,062,762)
Assets – median
225,000
186,000
281,000
Total mean debt
(for those with debt)
99,893
(120,983)
95,152
(113,801)
106,096
(129,520)
Mean debt to income
(for those with debt)
1.53
(1.65)
1.56
(1.66)
1.48
(1.63)
n = 11,301 (females); 8,635 (males)
35
Probit Fit of Being Sent to a Collections Agency
Variable
Coefficient
Std error
Intercept
1.58
0.25***
Gender (male = 1)
–0.081
0.077
Log Assets
–0.24
0.022***
Total Debt/Income
0.083
0.020***
Year 2007 (2006 omitted)
0.13
0.100
Year 2008
0.20
0.11
Year 2009
0.54
0.14***
Year 2010
0.0030
0.168
Year 2011
–0.13
0.18
Year 2012
–0.23
0.19
*** Significant at 1 percent level; n = 1,842 (subsample delinquent at least 60days
on credit card debt).
Tobit Results -- Effect of Collections Agency Encounter on Debt Stress
Variable
Coefficient
Std error
Intercept
2.01
0.28***
Gender (male = 1)
–0.29
0.081***
Collection Agency Encounter
1.16
0.13***
Gender x Collection Encounter
–0.39
0.21*
Log Assets
–0.14
0.023***
Total Debt/Income
0.37
0.021***
Year 2007 (2006 omitted)
0.14
0.11
Year 2008
0.47
0.11***
Year 2009
0.80
0.14***
Year 2010
0.84
0.14***
Year 2011
0.84
0.15***
Year 2012
1.00
0.15***
*** Significant at 1 percent level; * Significant at 10 percent level; n = 1,429 (subsample
delinquent at least 60days on credit card debt).
