Journal of Economic Psychology 26 (2005) 642663

www.elsevier.com/locate/joep

Debt and distress: Evaluating

the psychological cost of credit
Sarah Brown *, Karl Taylor, Stephen Wheatley Price
Department of Economics, University of Leicester, University Road, Leicester LEI 7RH, UK
Received 25 April 2004; received in revised form 9 December 2004; accepted 26 January 2005
Available online 4 May 2005

Abstract
In this paper we explore the association between debt and psychological well-being
amongst heads of households using the British Household Panel Survey. Our principle nding
is that those household heads who have outstanding (non-mortgage) credit, and who have
higher amounts of such debt, are signicantly less likely to report complete psychological
well-being. The average increase in the psychological distress is greater when outstanding
credit is measured at the individual, as opposed to household, level. No such signicant association is found in the case of mortgage debt. Our results highlight the psychological cost associated with the consumer credit culture in Britain.

2005 Elsevier B.V. All rights reserved.
JEL classication: G11; 131
PsycINFO classication: 3900; 3920
Keywords: Debt; Credit; Psychological well-being; Ordered probit models

Corresponding author. Tel.: +44 116 2522827; fax: +44 116 2522908.
E-mail address: sb118@le.ac.uk (S. Brown).

0167-4870/$ - see front matter
2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.joep.2005.01.002

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

643

My other piece of advice, Coppereld, said Mr. Micawber, you know.

Annual income twenty pounds, annual expenditure nineteen and six, result
happiness. Annual income twenty pounds, annual expenditure twenty pounds
ought and six, result misery.
(David Coppereld, Chap. 12, by Charles Dickens)

1. Introduction
There was a consumer credit explosion in the UK between 1994 and 2004. This
accompanied the sustained economic boom during this period and followed on from
the gradual relaxation of credit constraints in the late 1980s and early 1990s. The increased availability of unsecured credit is clear from the massive rise in the number
of dierent credit cards available (over 1300 in the UK, in 2000) and the broadening
of the range of nancial institutions oering unsecured loans. From once being primarily the preserve of the major banks, loans can also now be readily obtained from
building societies, UK and overseas-based nance companies and, even, supermarkets. In addition, the advent of telephone and internet-banking, and the availability
of credit at the point of purchase, has increased the accessibility of consumer credit
and the speed with which loan contracts are made.
Fig. 1 illustrates the dramatic escalation in the total value of outstanding consumer credit in the UK, between 1982 and 2002 (measured in 1995 pounds sterling
and excluding mortgage debt). Less than 1% of this change can be explained by the
5% growth in the size of the UK population during this period. Most of the increase
has arisen from the rise in the value of loans arranged directly (e.g. personal loans) or
indirectly (e.g. hire purchase agreements) with nancial institutions (the Other category). A growing proportion of outstanding consumer credit has been obtained
through the use of credit cards. As a percentage of GDP, the amount of unsecured
borrowing accumulated by individuals and households doubled, between 1993 and
2002, to 16%. By the end of 2004 the total amount of outstanding (non-mortgage)
credit was over 185 billion (in current prices), an average of more than 4800 for
every adult of working age in the UK.
Monetary policymakers have become concerned about the extent of personal
indebtedness, its sustainability and impact on aggregate economic performance
(e.g. Bank of England, 2004, pp. 910). However, there is also considerable concern
from social welfare lobbyists, amongst others, about the associated increase in the
number of individuals and families with problematic levels of personal debt. For
example, members of the National Association of Citizens Advice Bureau in the
UK dealt with approximately one million new personal debt enquiries during 2002
(NACAB, 2003). Over two-thirds of these contacts were associated with consumer
credit arising from bank loans, credit and store cards (whose interest rates are typically two or three times those of the banks), catalogue debts and hire purchase
agreements. Many of their clients were also in arrears with regard to housing rent,
council tax and utilities bills. Additionally, they report a 47% increase in the number

644

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

Fig. 1. Outstanding consumer credit (1995 prices). Notes: The data were obtained from the UK
Governments National Statistics Time Series Data website at http://www.statistics.gov.uk/.

of new contacts in this area over the period 19972002 (and a 25% increase in the
number of personal insolvencies, to 30,587). A quarter of these clients reported anxiety, depression and stress problems that resulted in them seeking medical treatment
(NACAB, 2003).
In this paper we examine the extent to which having outstanding credit inuences
the psychological well-being of household heads in the population as a whole, using
data from the 1995 and 2000 waves of the British Household Panel Survey. Our main
hypothesis is that debt may be associated with increased levels of psychological distress, a relationship which is most likely to hold amongst the principal nancial decision-makers in a household. Furthermore, we anticipate that unsecured debt is likely
to have a greater inuence on psychological well-being than secured debt. It is therefore crucial that we can, at least broadly, dierentiate between these two types of
debt in our empirical analyses. It is also critical that we distinguish between nancial
liabilities and nancial assets, rather than aggregate them together into an overall
measure of net wealth, allowing us to explore whether their associations with psychological well-being might be asymmetric.

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

645

It is important to clarify, at this point, that we use the terms borrowing, credit,
debt and indebtedness interchangeably in this paper. However, as Webley and Nyhus (2001, p. 426) point out these terms have distinct meanings in the psychology literature (and elsewhere). Specically, whilst borrowing is planned and intended and
may involve the granting of credit, it is possible to have a debt problem without
ever having borrowed money. In contrast Debt is unplanned and unintended
and may be . . . a stage (for some) en route to default and bankruptcy.
The principal empirical nding in this paper is that those heads of households
who have outstanding credit, and who have higher amounts of such debt, are significantly less likely to report complete psychological well-being. The average increase
in the psychological distress associated with this form of indebtedness is greater
when outstanding credit is measured at the individual, as opposed to household, level and exceeds that from mortgage debt. The paper is organised as follows. In Section 2 we review the contribution of previous studies, from both the economics and
psychology literatures, to our understanding of psychological well-being and indebtedness. In Section 3, we introduce our data source, dene the measures we employ
and describe our sample. Our empirical methodology is explained in Section 4 and
the estimates from our statistical models are discussed in Section 5. We summarise
our ndings and present our conclusions in Section 6.

2. Literature review
2.1. Psychological well-being
The investigation of the factors aecting human well-being is central to the discipline of psychology (see Kahnemann, Diener, & Schwarz, 1999, for a detailed review
of this literature). It is recognised that the best method to gain information about a
persons perspective on their life or work is to ask them directly. Economists have
traditionally been more reluctant to use self-reported subjective measures of wellbeing (Bertrand & Mullainthan, 2001) due to concerns about the interpretation of
such variables, the validity of inter-personal comparisons and the diculty of modelling such psychological outcomes (Jahoda, 1982, 1988).
Whilst life satisfaction measures are now more widely used by economists, in the
UK literature to date (e.g. Clark & Oswald, 1994; Shields & Wheatley Price, forthcoming) individual well-being measures have been mainly based on the General
Health Questionnaire (see Appendix B for details). The ordered ranking of responses, known as the GHQ12 score (Goldberg, 1972), is widely recognised as a reliable measure of psychological well-being (Argyle, 1989). From the perspective of
economic research, these psychological well-being (or, less precisely, happiness)
measures provide directly observable proxies for individuals well-being or utility.
Recent years have seen a large number of economic studies using such variables
(see Frey & Stutzer, 2002; Oswald, 1997).
An enormous literature, throughout the social sciences, has focussed attention on
the associations between individual well-being and economic outcomes. Economists

646

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

have been mainly interested in the eects of unemployment and income. The observation that being unemployed generally leads to a signicant deterioration in an individuals well-being has become a stylised fact, validated across countries, time
periods and data sources (e.g. Clark & Oswald, 1994). Importantly, the causal direction from unemployment to lower levels of self-reported well-being has been convincingly demonstrated using longitudinal and panel data (e.g. Clark, 2003). This
arises, not primarily due to the fall in income (Clark & Oswald, 2002), but mainly
due to the loss of the psychological benets from work, such as social recognition,
self-respect and the opportunities for social interaction (Darity & Goldsmith, 1996;
Jahoda, 1982; Lane, 1991). Crucially, unobservable individual heterogeneity, which
potentially could explain variations in such measures (see Kahnemann et al., 1999),
is relatively unimportant in these studies (Clark & Oswald, 2002).
Economic theory assumes a strong positive inuence of income on individual
well-being, but consistent empirical evidence of this is lacking which is supportive
of Lanes (1991) contrasting view. For example Campbell, Converse, and Rodgers
(1976) and Easterlin (1974, 1995) found that income is a poor predictor of many
measures of individual well-being, whilst Clark and Oswald (1994) found no robust
association between income and the GHQ12 score in the UK. In contrast, most
European-based empirical studies nd a small positive eect on self-reported life satisfaction (e.g. Frey & Stutzer, 2000; Winkelmann & Winkelmann, 1998). Perhaps the
most convincing evidence that higher income levels may lead to signicant improvements in individual well-being is provided by Frijters, Haisken-DeNew, and Shields
(2004), who follow the income and life satisfaction levels of East Germans after
reunication. An alternative hypothesis, that relative rather than absolute income
matters, has considerable empirical support (e.g. Clark & Oswald, 1996; Van Praag
& Frijters, 1999). Headey and Wooden (2004) and Headey, Muels, and Wooden
(2004) argue that net wealth and non-durable consumption expenditures have
greater positive impacts on life satisfaction than income.
A number of other consistent correlates of individual well-being are reported. For
instance, a U-shaped association with age has been found for many countries (e.g.
Clark, 2003) whereas marital dissolution and ill-health have signicant adverse
eects on individual well-being (e.g. Kahnemann et al., 1999; Shields & Wheatley
Price, forthcoming). No consistent empirical support for the associations between
gender, educational attainment or the presence of children and measures of individual well-being has been found.
2.2. Personal debt: Causes
The economic psychology literature represents one area where there has been signicant interest in the determinants of personal debt. Livingstone and Lunt (1992)
investigated the determinants of the level of debt and repayments across individuals
and found that attitudinal factors, such as whether individuals are pro or anti debt,
were key correlates. Davies and Lea (1995) analysed student attitudes towards debt
and found that as students increased borrowing levels, in order to nance investments in human capital, their attitudes became more tolerant towards credit and

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

647

debt. Lea, Webley, and Levine (1993) also found that debt levels are strongly associated with attitudinal factors and warned that the adaptation of these attitudes to
higher debt levels, in combination with the increased availability of credit could lead
to a self-sustaining culture of indebtedness (p. 118). They also demonstrated the
importance of economic circumstances in determining debt outcomes and found that
those with a tendency to have one form of debt were more likely to have many other
forms of debt as well.
Research by economists into personal debt issues is surprisingly scarce, especially
in the UK. Godwin (1997) has explored the dynamics of US household credit use
and found considerable mobility in debt status during the 1980s. More recently,
Crook (2001) has shown that income, home ownership and family size all impact
positively upon the level of debt in US households, whilst expectations of future
interest rate changes appear to have no inuence. Cox and Jappelli (1993) have estimated that on average, desired debt levels are 75% higher than actual levels amongst
US households, highlighting the role of credit constraints (Jappelli, 1990). Whilst
some of the latent demand for credit may be met from private transfers (Cox & Jappelli, 1990), Gross and Souleles (2002) observed that debt levels rise in response to
increases in credit (card) limits.
One intriguing puzzle is the apparent targeting of a specic credit utilisation rate
by credit card holders (Gross & Souleles, 2002) thus failing to eliminate costly debt
using available liquid assets. Bertaut and Haliassos (2002) have proposed an accountant-shopper model to explain such debt-revolvers and provide corroborating evidence from the 1995 and 1998 US Surveys of Consumer Finance. They argue that
consumption decisions are dissociated from portfolio allocations within the household. The accountant, who is in charge of household nancial decision-making, attempts to control consumption expenditure by the shopper, through holding credit
card balances as a xed proportion of their limit. Hence a certain level of debt is tolerated in order to prevent additional spending, despite high levels of liquid savings
we address this potential asymmetry in the empirical results.
In one of the few pieces of economic research on debt using UK data, Bridges and
Disney (2004) nd that dierences in the incidence of credit and default among lowincome households, are inuenced by labour market status, age, access to social
security benets and household composition. More recently, Brown, Garino, Taylor,
and Wheatley Price (2005) have shown that individuals and households with more
optimistic nancial expectations incur more debt in the UK.
2.3. Personal debt: Consequences
A key issue is whether psychological attributes are determinants of observed debt
outcomes or whether they are the result of being indebted. In addition to the studies
noted above there is considerable evidence of a strong statistical association between
nancial distress and severe psychological problems in the general population
(Weich & Lewis, 1998), including depression amongst British Civil servants (Marmot, Ry, Bumpass, Shipley, & Marks, 1997). There are also several studies noting
that indebted students are more likely to be exhibiting symptoms of psychological

648

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

distress (e.g. Roberts, Golding, & Towell, 1998; Stradling, 2001, Chap. 5). Roberts
et al. (1998) argue that this may be because psychologically distressed individuals
are more likely to get into monetary problems. Alternatively, the nancial strain
and worry of being in debt may lead to a decline in psychological well-being.
Webley and Nyhus (2001) argue that, in many of the cross-sectional psychological
studies cited above, the causality is unclear. Their analyses of panel data from the
Netherlands nd some evidence of causality from debt to psychological outcomes.
Further support for this contention is found in longitudinal studies, such as Marmot
et al. (1997) and Stradling (2001, Chap. 5). A nal concern is that if the credit is obtained in order to nance the purchase of consumer durables this might lead to an
increase in psychological well-being. Hence it is important to control for the potential utility gains from the presence of such goods in empirical models of psychological well-being.

3. Data
3.1. Data
In this paper, we explore the empirical determinants of individual psychological
well-being in Great Britain, focusing on the inuence of outstanding debt. Our analysis is based on a sample of household heads drawn from the British Household Panel Survey (BHPS). Potential alternative UK datasets such as the Family Resources
Survey or the Family Expenditure Survey contain information on only one form of
debt each (mortgage repayments and personal loans, respectively) but include no
measure of psychological well-being. Uniquely, the BHPS contains information
on both the total outstanding amount of credit and individuals psychological
well-being.
The BHPS is a nationally representative random sample longitudinal survey, carried out by the Institute for Social and Economic Research, of every adult in more
than 5000 private households in Great Britain. The rst interviews were conducted
during the autumn of 1991 and annual re-interviews have taken place ever since.
Our sample consists of a balanced panel of 2193 heads of households, of working
age (1665), who responded to both the 1995 and 2000 waves of the BHPS. We focus
on household heads as they, typically, have nal responsibility for household nancial decision-making (the accountant role see Bertaut & Haliassos, 2002) and are
thus expected to bear the main psychological burden of the households nancial situation. A descriptive portrait of the main nancial features of our sample is provided
in Table A1, in Appendix A.
3.2. Key variable denitions
As with previous studies of individual psychological well-being in the UK we
examine the inverse caseness version of the GHQ12 score, which sums the binary
values to the responses from each question (1 indicating a high level of psychological

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

649

well-being and 0 signifying otherwise). This results in scores of 012, where higher
numbers indicate increased psychological well-being.
Our denition of outstanding debt is based on the question, asked only in the
1995 and 2000 waves of the BHPS, How much in total do you owe? The question
clearly relates to outstanding (non-mortgage) credit as details about mortgage debt
are asked in a separate question. This information is also used later as a point of
comparison. Similar self-reported measures of debt have been shown to be reliable
indicators of actual debt (Lea et al., 1993; Lea, Webley, & Walker, 1995). All nancial measures used are deated to 1995 pounds sterling. If the household head reports a non-zero value of debt they are coded as 1 in a dummy variable indicating
the individual has outstanding credit.
A left-censored measure of the natural logarithm of the individual level of outstanding credit is also constructed taking the value of zero for non-debtors and
the log of amount of outstanding (non-zero) debt otherwise. Similar household measures of debt are dened, based on summing the underlying individual responses
across all adult members within the household. The main limitations of these outstanding credit measures are that the contract starting date, expected duration of
the loan, applicable rate of interest and the type of creditor are not explored in
the BHPS.
The BHPS does record information on the amount saved each month, which we
use to create measures of annual savings at both the individual and household level.
Further questions on the amount held in investments, the receipt of a lump sum
windfall within the past year and the home owners subjective valuation of their
house provide the basis for our wealth controls. We also include controls for individual labour income or total household income in our empirical models. Finally, given
the importance of attitudinal factors in determining responses to psychological wellbeing questions and nancial expectations for debt levels we include dummy
variables indicating the perception of the household head of their nancial situation,
relative to one year previously, and their expectation of the direction of change over
the forthcoming year (Katona, 1975). These variables implicitly incorporate a synthesis of the respondents personal nancial outlook (e.g. income and labour market
position and prospects), their macroeconomic expectations (e.g. future interest and
taxation rates) and their general personality traits of optimism or pessimism.
3.3. Descriptive analysis
In our sample of household heads the mean value of the GHQ12 score is around
10, which is close to the maximum of 12, and this does not vary much across waves
(see Table A1). Reecting the overall growth in consumer credit in the UK (illustrated earlier in Fig. 1), the mean levels of individual and household outstanding
credit have increased by over 50%, between 1995 and 2000, in our sample. The individual outstanding credit levels, of those household heads in debt, have risen from an
average of 1957, in 1995, to 3192 in 2000. Mortgage debt also grew, by nearly 30%,
over this period. It is also interesting to note that household heads are, on average,
the predominant debtors within the household, individually responsible for over 85%

650

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

of total debt. However, the proportion of individuals and their households in debt,
and with a mortgage, has declined over these ve years (from 47% to 43% for individual level debt and from 56% to 51% for household level debt), perhaps reecting
their movement through the life-cycle.
Of particular relevance to this paper we nd that those household heads without
any debt have signicantly higher (t-statistic = 3.09, p-value < 0.001) mean levels of
psychological well-being (10.17) than those with some non-zero level of debt (9.87
not shown in Table A1). This nding holds for both individual and household measures of debt and across both waves of data. Later we explore whether this nding is
robust to controlling for potentially confounding factors.
Average labour income for these household heads grew by just over 4%, in real
terms, between 1995 and 2000 whilst household income grew by over 12%. Interestingly, individual annual savings grew by 20%, and investments declined by nearly
20%, during this period. Household savings grew more slowly. The value of windfalls
received more than doubled though the proportion of household heads receiving
them fell substantially and house values rose by over 60%. Importantly, these nancial variables are not highly correlated. Indeed these correlations are very low, typically around 0.1. In both 1995 and 2002 around 30% of household heads had
optimistic nancial outlooks, both retrospectively and prospectively, whilst across
these ve years a declining share had a negative opinion of their relative nancial
position.

4. Empirical methodology
4.1. Psychological well-being
Given our ordered dependent variable, the 012 ranking in the GHQ12 score,
the statistical model we employ is the standard ordered probit model (see, for
example, Greene, 2003, pp. 736740 for details) with constant thresholds. This
approach, as opposed to treating our dependent variable as continuous and tting
a linear model, is standard in the economics literature on psychological wellbeing. As noted by Fielding (1999), the linear model requires a number of restrictive assumptions, in particular, the assumption of cardinality, which is dicult to
accept in the present situation. Importantly, since we have two observations on
the same household heads, in 1995 and in 2000, we ensure the standard errors of
the estimated coecients are corrected both for the clustering of observations and
for heteroskedasticity.
Our primary interest is in estimating the association between debt and psychological distress in our sample of household heads. Firstly we explore whether the significant dierence in psychological well-being, between debtors and non-debtors, which
we observed in the raw data still exists once other potentially confounding factors
are accounted for. We estimate this, and all subsequent, models using both individual and household measures of debt (with individual level and household level nancial control variables, respectively). Secondly we examine the impact of the level of

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

651

outstanding credit (using the log variables dened earlier with non-debtors assigned the value zero) on the psychological well-being of household heads. In these
empirical models, whose results are presented in Tables 1 and 2, we control for
monthly income, annual savings, investments, windfall payments received over the
previous year, outstanding mortgage loans and a subjective estimate of house value
(in the case of home owners). Since our data comprises information on the same individuals in both 1995 and 2000 we additionally include a dummy variable for the 2000
wave, and its interaction with the outstanding credit measure, to explore whether the
estimated associations have changed signicantly over this period.
Following previous studies of individual psychological well-being, particularly
those undertaken using data from the United Kingdom (e.g. Clark, 2003; Clark &
Oswald, 1994), we also include a number of standard explanatory variables in our
statistical models, namely age, gender, marital status, the number of dependent children and of adults in the household, ordinal indicators of self-reported health status,
labour market status, housing tenure, car ownership, educational attainment, ethnicity and region of residence. Following Taylor (2002) we include variables indicating
whether individuals expect their nancial situation to improve or worsen over the
following year as well as their assessments of their current nancial position, relative
to a year ago. Lastly we control extensively for recent purchases of consumer durable
goods using dummy variables to indicate the purchase of a colour television, videorecorder, freezer, washing machine, tumble dryer, dishwasher, microwave, computer
or CD-player, within the past year.
4.2. Endogenous measures of debt and savings
In the specications outlined above, we treat our measures of outstanding credit
level and annual savings as exogenous. However, it may be the case that the coecients on our measures of debt underestimate the true magnitude of the association
between outstanding credit levels and psychological well-being if there exist unobserved individual-specic factors which determine both the extent of indebtedness
and reported psychological distress. For instance, the presence of credit rationing
will not only lower the debt levels of the aected individuals but may also impact
upon their levels of psychological well-being. Therefore, in the estimates reported
in Table 3, we relax this assumption and explore the importance of predicted measures of the outstanding credit level and a predicted measure of savings. In each case
we a use Tobit specication to estimate predicted debt and savings, whilst controlling
for a range of potentially important determinants. The predicted debt and savings
measures, calculated at both the individual and household levels as appropriate, then
replace the exogenous variables in the instrumented ordered probit models reported
in Table 3.
The model specication and main control variables for these predicted models
build on the determinants of debt models estimated in Brown et al. (2005). Log debt
and log savings are assumed to be determined by the same set of personal characteristics and other control variables used in the psychological well-being models with
the following over-identifying covariates: in Models A & B the debt variable is

652

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

Table 1
Head of households psychological well-being and nancial behaviour
Covariates

Individual nancial
behaviour
S.E.

b
Aged
Aged
Aged
Aged

2534
3544
4554
5564

years
years
years
years

old
old
old
old

Household nancial
behaviour
M.E.

S.E.

M.E.

.0147
.0640
.1188
.3536**

.0944
.1008
.1039
.1154

.0059
.0255
.0473
.1392

.0251
.0738
.1211
.3377**

.0951
.1016
.1047
.1154

.0100
.0294
.0482
.1331

.1973**

.0475

.0785

.2110**

.0474

.0840

Log(individual labour income last month)

Log(total household income last month)

**

.0329

.0101

.0131

.0314

.0270

.0125

Individual has outstanding credit

Individual has outstanding credit in
2000 wave
Household has outstanding credit
Household has outstanding credit in
2000 wave

.1567**
.0546

.0492
.0689

.0625
.0218

.1184*
.0324

.0497
.0689

.0472
.0129

Individual saves money each year

Household saves money each year

.0928*

.0400

.0370

.1104**

.0402

.0440

Individual has investments

Individual received a lump sum windfall
Individual has an outstanding
mortgage loan
Log(value of house home owners only)

.0217
.0347
.0666

.0411
.0394
.0507

.0086
.0138
.0266

.0236
.0303
.0628

.0408
.0393
.0509

.0094
.0121
.0251

.0016

.0105

.0006

.0028

.0108

.0011

Believes nancial situation is better than

1 year ago
Believes nancial situation is worse than
1 year ago

.0130

.0440

.0052

.0047

.0437

.0018

.4088**

.0443

.1614

.4158**

.0442

.1641

Expects nancial situation to improve

in next year
Expects nancial situation to worsen
in next year

.0157

.0421

.0063

.0123

.0422

.0049

.1912**

.0613

.0760

.1875**

.0612

.0745

Observation from 2000 wave

.0127

.0478

.0051

.0324

.0689

.0056

Male

Log likelihood (constant only model)

Log likelihood
LR Model v2 (d.f. 60)
Sample size

7370.97
6932.60
849.13**

7370.97
6939.09
831.78**
4186

Notes:
1. Fitted values from ordered probit models with the inverse caseness version of the GHQ12 score as the
dependent variable. Balanced panel sample of heads of households present in both the 1995 and 2000
waves of the British Household Panel Study.
2. b is the coecient of the covariate included in the model and S.E. is the estimated standard error of the
reported. b, M.E. is the simulated marginal eect of the change in the probability of an average individual reporting complete psychological well-being (i.e. a GHQ12 score of 12), due to a change in the
explanatory variable.

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

653

Table 1 (continued)
3. Omitted categories: aged less than 25 years old, female, believes nancial situation is the same as 1 year
ago and expects nancial situation to remain the same in the next year.
4. * and ** indicate statistical signicance at the 5% and 1% levels, respectively.
5. Controls for marital status, ethnicity, self-reported general health status, educational qualication
level, labour market status, living in rented accommodation, car ownership, the number of children
in the household, the number of adults in the household, region of residence and the type of consumer
goods possessed are also included in each model. These results are not reported for the sake of brevity.
6. Eleven constant thresholds were also estimated.

instrumented by whether the individual has a credit card whereas the individuals social class is used in Models C & D; savings is instrumented, in Models B & D ; only,
by whether the individual is part of an employer pension scheme and whether the
individual contributes to a private pension plan but remains exogenous in Models
A & C. The dierent instruments used and the dierent combinations with exogenous and endogenous measures of savings provide some condence in the robustness
of these results, given the inherent diculties in nding valid instruments for both
debt and savings simultaneously.
4.3. Marginal eects
The coecient estimates from ordered probit models indicate the change in the
latent variable arising from a change in the respective correlates. However, it is
also useful to evaluate the impact of a change in each explanatory variable, on
the cumulative probability of psychological well-being (Fielding, 1999). Therefore
we simulate and report these marginal eects (M.E.) for each estimated model.
For the binary independent variables we report the change in the predicted probability of an otherwise average individual, reporting a GHQ12 score of 12 rather
than a lower score on either scale, when a particular characteristic holds compared
to when relevant base characteristic is present. These numbers show the separate
eect of each explanatory variable on an average individuals probability of having
the highest level of self-reported psychological well-being, compared to lower levels. This is to be the primary threshold of interest, with 49.70% and 52.88% of individuals reporting this category in 1995 and 2000 respectively. For the non-binary
independent variables, which have all been included in natural logarithmic form,
the marginal eect indicates the change in the probability of reporting a GHQ12
score of 12 arising from a 1% increase in the underlying continuous variable from
its mean value.

5. Empirical results
The parameter estimates, associated standard errors and marginal eects from
our ordered probit models of psychological well-being are presented in Tables 13.
Models are tted using both individual and household measures of the nancial variables. In the reported models it is clear that the null hypotheses, of the Likelihood

654

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

Table 2
Head of households psychological well-being individual and household levels of credit and income
Covariates

Individual credit and income

Household credit and income

S.E.

M.E.

.0158
.0654
.1159
.3489**

.0943
.1006
.1037
.1156

.0063
.0261
.0462
.1374

.0248
.0738
.1178
.3332**

Male

.2034**

.0476

.0810

Log(individual labour
income last month)
Log(total household
income last month)

.0336**

.0101

Log(individual level of
outstanding credit)
Log(individual outstanding
credit) in year 2000
Log(household level of
outstanding credit)
Log(household outstanding
credit) in year 2000

b
Aged
Aged
Aged
Aged

2534
3544
4554
5564

years
years
years
years

old
old
old
old

S.E.

M.E.

.0950
.1014
.1046
.1156

.0099
.0294
.0469
.1314

.2144**

.0474

.0854

.0134

.0324

.0272

.0129

.0231**

.0071

.0092

.0121

.0095

.0048

.0163*

.0070

.006:5

.0081

.0093

.0032

Log(individual amount
saved each year)
Log(household amount
saved each year)

.0126*

.0060

.0050

.0154**

.0059

.0062

Log(total amount held

in investments)
Log(amount received as a
lump sum windfall)
Log(value of outstanding
mortgage loans)
Log(value of house home
owners only)

.0017

.0055

.0007

.0016

.0054

.0006

.0018

.0065

.0007

.0006

.0065

.0002

.0077

.0049

.0031

.0075

.0049

.0030

.0006

.0106

.0003

.0040

.0105

.0007

Believes nancial situation

is better than 1 year ago
Believes nancial situation
is worse than 1 year ago

.0133

.0441

.0053

.0047

.0438

.0019

.4089**

.0444

.1615

.4147**

.0444

.1637

Expects nancial situation

to improve in next year
Expects nancial situation
to worsen in next year

.0172

.0422

.0069

.0134

.0423

.0053

.1925**

.0613

.0765

.1891**

.0613

.0752

.0021

.0463

.0008

.0021

.0504

.0008

Observation from 2000 wave

Log likelihood (constant only model)
Log likelihood
LR Model v2 (d.f. 60)
Sample size
Notes: As Table 1.

7370.97
6933.32
847.49**

7370.97
6939.88
831.45**
4186

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

655

Table 3
Parameter estimates using endogenous measures of outstanding credit and annual savings
Model

Log(outstanding credit)
S.E.

b
Individual credit and income
Estimates from Table 2 for comparison
Model
Model
Model
Model

A endogenous credit only

B endogenous credit and savings
C endogenous credit only
D endogenous credit and savings

Household credit and income

Estimates from Table 2 for comparison
Model
Model
Model
Model

A endogenous credit only

B endogenous credit and savings
C endogenous credit only
D endogenous credit and savings

Sample size

.0231**

.0071

Log(annual savings)

M.E.

S.E.

M.E.

.0092

.0126*

.0060

.0050

.0265
.0256*
.0448**
.0450**

.0123
.0122
.0159
.0153

.0106
.0102
.0179
.0180

.0140
.0113
.0134*
.0093

.0060
.0177
.0060
.0177

.0056
.0045
.0054
.0037

.0163*

.0070

.0065

.0154**

.0059

.0062

.0111
.0110
.0133
.0143

**

.0059
.0172
.0059
.0172

.0066
.0178
.0064
.0178

.0279
.0277*
.0334*
.0358*

.0150
.0150
.0171
.0170

.0165
.0446**
.0159**
.0447**

4186

Notes:
1. All covariates as in Table 2. The full results are not reported for the sake of brevity.
2. #, * and ** indicate statistical signicance at the 10%, 5% and 1% levels, respectively.
3. Model A replaces the relevant log(outstanding credit) variable with its predicted value estimated from
a debt equation with the following over-identifying covariate: whether the individual has a credit card.
The log(annual savings) variable is treated as exogenous.
4. Model B replaces the relevant log(outstanding credit) variable with its predicted value estimated from a
debt equation with the following over-identifying covariate: whether the individual has a credit card.
The log(annual savings) variable is also replaced by its predicted value from a savings equation with
the following over-identifying covariates: whether the individual is part of an employer pension scheme
and whether the individual contributes to a private pension plan.
5. Model C replaces the relevant log(outstanding credit) variable with its predicted value estimated from a
debt equation with the following over-identifying covariates: dummy variables for the individuals
social class (occupational status). The log(annual savings) variable is treated as exogenous.
6. Model D replaces the relevant log(outstanding credit) variable with its predicted value estimated from
a debt equation with the following over-identifying covariates: dummy variables for the individuals
social class (occupational status). The log(annual savings) variable is also replaced by its predicted
value from a savings equation with the following over-identifying covariates: whether the individual
is part of an employer pension scheme and whether the individual contributes to a private pension
plan.

Ratio test that all estimated coecients are equal to zero (Greene, 2003, pp. 390
391), are clearly rejected. Our simple tests for any statistical dierence in the size
of the association between psychological well-being and outstanding credit, or in
the level of psychological well-being, across the two waves (see the coecients on
the 2000 wave shift and interaction controls) are also clearly rejected in all estimated
models. This suggests that there has been no attenuation in the psychological impact
of debt (as Davies & Lea, 1995, found amongst students) over this period, despite the
dramatic increase of the levels of outstanding credit amongst these household heads.

656

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

This argument is reinforced by the decline in the proportion of household heads with
debt over the two waves if it is assumed that those for whom debt caused the greatest
psychological distress were able to pay it o.
Given the focus of this paper we report and discuss only the estimated associations between the nancial variables and psychological well-being in our sample
of household heads. The general determinants, such as older individuals and males
having signicantly higher GHQ12 scores, are clearly in line with the ndings of
previous UK studies of representative samples of the entire adult population. Of
particular relevance is the inuence of income at both the individual and household levels. For household heads the level of their own labour income clearly
has a signicantly positive inuence on their reported levels of psychological
well-being.
However, the level of household income is not signicantly associated with household heads reported GHQ12 scores, despite an estimated coecient of only a slightly
smaller magnitude. We explored directly how the income of other household members aected our dependent variable by separately including such a variable (results
not reported). Its estimated coecient was positive, but close to zero, and clearly
insignicant. Hence the average impact, of the household heads own labour income
and the contribution from the rest of the household, gives us the reported overall
insignicant result. We also explored whether this result was sensitive to the denition of household income and found that estimates using household equivalence
measures (of income and other household nancial variables) gave qualitatively
equivalent results (not reported).
5.1. Psychological well-being, debtors and non-debtors
Household heads who have some outstanding credit, at either the individual level
or within their household, report signicantly lower levels of psychological wellbeing than those with no debt. As reported in Table 1 the presence of individual
(household) debt reduces the probability of scoring the maximum on the GHQ12
score by over 6% (nearly 5%). Interestingly household heads with secured debt, in
the form a mortgage loan, do not report signicantly dierent levels of psychological
distress conrming our contention that these dierent forms of debt may have distinct psychological aects.
Household heads who save themselves, or whose households save, on a regular
basis are found to be around 4% more likely to report complete psychological
well-being than non-savers. Interestingly the positive benet from being a saver is
outweighed by the negative eect of being in debt which, together with the dierential impact of outstanding credit and savings being much greater when measured at
the individual, rather than household, level, suggests a clear asymmetry in the way
these nancial behaviours inuence psychological well-being. However, other nancial assets appear to have little eect as those with more valuable houses, who have
investments or who have received a lump sum windfall within the past year are no
more likely to report higher levels of psychological well-being.

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

657

In common with Taylor (2002), we nd a strong statistical association between

individuals nancial expectations and their levels of psychological well-being. In
particular, those who have a pessimistic view of their relative nancial position report signicantly lower GHQ12 scores than otherwise equivalent household heads.
Interestingly the association with psychological well-being is more than twice as
strong for those who view their current nancial position as worse than one year previously (M.E.  0.16) as compared to those are pessimist about future nancial
conditions (M.E.  0.075).

5.2. Psychological well-being and the quantity of outstanding credit

In the estimates reported in Table 2 we have replaced the outstanding credit, savings, investments, windfalls and mortgage with their level measures (the log variables
dened earlier) in our estimated models. Since the signicance and magnitude of the
other estimated parameters are unaected we focus our discussion on the eect of
these quantity measures. We nd evidence of a negative statistical association between the levels of individual and household outstanding credit and household
heads GHQ12 scores, though the size of the marginal eect of the former is approximately 50% larger.
In order to appreciate the magnitude of these eects we consider how much additional monthly income would be required in order to oset the negative impact on
the probability of reporting complete psychological well-being, for an otherwise
average individual, of a 10% rise in the average level of outstanding credit. At the
individual level mean monthly labour income, over the whole sample, is 936.5.
The average level of outstanding credit, for those with some debt, is 2574.75 so
the average debt level over the whole sample is 1153.5 as only 45% of household
heads in the sample are in debt. A 10% increase in the level of outstanding credit
(i.e. an additional 115.35) would reduce the probability, of a household head with
otherwise mean characteristics, reporting a maximum GHQ12 score by 0.092
(10 0.0092). To eliminate this eect monthly labour income would have to rise
by 64.30, nearly 7% (.092/.0134). Similarly average annual savings, of the whole
sample (635.5), would need to increase by 116.93 or over 18% (.092/.005) in order
to maintain the average probability of complete psychological well-being. It is
important to emphasise that these are average eects over the whole sample of
household heads. Amongst those in debt the marginal increase in psychological
well-being would be larger and the corresponding osetting eects would need to
be much more substantial.
A clear limitation of these ndings is that we only observe our measures of outstanding credit in two time periods, 1995 and 2000, and so cannot rmly establish
that our parameter estimates are the causal eects of debt on psychological wellbeing. However, a number of recent UK studies have demonstrated that the causality runs from many of our covariates to self-reported GHQ12 scores (e.g. Clark,
2003). Furthermore, the studies by psychologists, which have incorporated a longitudinal element, support our contention that causality is primarily from increased

658

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

levels of debt to reduced levels of psychological well-being. Nevertheless, we

acknowledge that the possibility of reverse causality, where debt results from behaviour associated with psychological distress, in our sample.
The amount of regular (annual) savings is signicantly associated with increased
levels of psychological well-being, amongst heads of households in our sample (M.E.
of a 10% increase in annual savings .05). As with the dummy variable indicators,
neither the quantity of total investments, the size of lump sum windfalls nor homeowners house valuations have any signicant association with our self-reported psychological well-being measure. Our disaggregated ndings, for heads of households
GHQ12 scores, thus contrast with those of Headey and Wooden (2004) (using Australian data) and Headey et al. (2004) who nd that households net worth impacts
positively on adult life satisfaction scores in a number of countries, including Britain.
There are a number of possible explanations for the dierences in our ndings to
those of Headey et al. (2004), and the British data they use: Headey et al. (2004) focus upon a single cross section the year 2000 only; they measure life satisfaction (a
7-point scale) rather than the GHQ12 inverse caseness score; and nally their measure of net worth is dened as assets minus debt. By denition net worth is a linear
combination of variables which we nd insignicant (assets) and signicant (debt).
As such it is perhaps not surprising that Headey et al. (2004) nd that net worth
is signicant our detailed disaggregation of net worth suggests that debt may drive
this relationship.
5.3. Psychological well-being and endogenous measures of debt and savings
The results of tting our ordered probit models using predicted measures of outstanding credit levels and annual savings are presented in Table 3. In every case the
parameter estimate of the association between outstanding credit, at both the individual and household levels, increases in magnitude. This conrms our contention
that the exogenous debt parameter estimates should be treated as lower bounds of
the true eect. The coecients on the exogenous savings variables change little when
predicted measures of debt are used. However, when we attempt to simultaneously
control for the potential endogeneity of savings, as well as our debt measures, we
nd that the coecient on predicted savings becomes insignicant at the individual
level. In contrast the household level estimates retain their statistical signicance levels and increase in magnitude. Hence, our estimates of the impact of our debt measures on psychological distress are robust to whether savings is also instrumented or
not.

6. Conclusions
In this paper we have explored the impact of debt on the self-reported psychological well-being of household heads in Great Britain. Our ordered probit estimates are
based on a balanced panel sample from the 1995 and 2000 waves of the British
Household Panel Surveys. The evidence conrms our main hypothesis, that debt

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

659

is associated with increased levels of psychological distress. We also nd that

unsecured debt, as measured by outstanding (non-mortgage) credit, has a greater
negative inuence on psychological well-being than secured (mortgage) debt, for
whom no signicant statistical relationship is found. Furthermore, we have
shown that the psychological eects of being in debt and being a regular saver are
not only opposing but asymmetric at the individual level. Our estimated marginal
eect of having outstanding credit is nearly double that of being a regular saver.
Finally, we nd no positive psychological benet from investments, windfalls or
house values justifying our disaggregated approach to controlling for assets and
liabilities.
Simple simulations have revealed that plausible proportionate changes in outstanding credit levels are associated with a non-trivial decrease in the probability
of reporting the highest level of-psychological well-being. For an otherwise average
individual a 10% increase in the level of individual outstanding credit would need a
7% increase in monthly income, or a 18% increase in annual savings, to oset the
negative impact on their psychological well-being. Additionally, we have presented
some econometric evidence, which suggests that our estimates of the size of the exogenous outstanding credit eects are downwardly biased. We conclude that there may
be a substantive psychological cost associated with consumer credit culture in Britain. Future government policy perhaps ought not to just focus on the potential macroeconomic consequence of the rising levels of consumer indebtedness in the UK but
also take consider the more general welfare eects of increased psychological distress
amongst debtors.

Acknowledgements
We are indebted to the editor, Simon Kemp, two anonymous referees and Mike
Shields for their constructive comments on earlier drafts, which have helped us
greatly improve this paper. We are also grateful to the Data Archive at the University of Essex, for supplying the 1995 and 2000 waves of the British Household Panel
Surveys, and to the principal investigator (Institute for Social and Economic
Research, University of Essex), the data collectors (NOP Market Research Ltd
and the Oce for National Statistics) and sponsors (Economic and Social Research
Council, Health Education Authority, Oce for National Statistics and Eurostat).
This paper was written whilst Stephen Wheatley Price was on Study Leave
from the University of Leicester, whose nancial support he gratefully acknowledges. All views expressed and any remaining errors are the authors joint
responsibility.

Appendix A
See Table A1.

660

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

Table A1
Descriptive statistics for heads of householdsa
Variable

1995

2000

Mean or
proportionb

Standard
deviation

Mean or
proportionb

Standard
deviation

10.00

3.00

10.07

3.05

.089
.267
.275
.259
.110

.285
.443
.447
.438
.313

.017
.191
.297
.267
.228

.128
.393
.457
.442
.420

Malec

.525

.499

.525

.499

Log(individual labour income last month)

916.8

894.9

955.5

1033.2

Log(total household income last month)

1791.7

1263.7

2021.0

1358.1

Individual has outstanding credit

Individual amount of outstanding creditc
Household has outstanding creditb
Household amount of outstanding creditc

.469
1957.1
.564
2309.1

.499
3631.3
.496
3938.6

.427
3192.4
.511
3645.1

.495
4500.8
.500
5478.9

Individual saves money each yearb

Individual amount saved each yearc
Household saves money each yearb
Household amount saved each yearc

.404
1412.0
.513
1755.4

.491
2094.6
.500
2351.4

.417
1695.5
.521
1973.1

.493
2648.0
.500
2757.4

Individual has investmentsb

Individual amount held in investmentsc
Individual received a lump sum windfallb
Individual amount received as a lump sum windfallc
Individual has an outstanding mortgage loanb
Value of outstanding mortgage loanc
Individual lives in own homeb
Value of housec

.310
12078
.452
2362.8
.603
39351
.756
74048

.463
42482
.498
9438.6
.489
39864
.430
48561

.334
9789.2
.262
5383.3
.562
50626
.797
119887

.472
23449
.440
17971
.496
50034
.402
91157

Believes nancial situation is better

than 1 year agob
Believes nancial situation is worse
than 1 year agob

.293

.455

.314

.464

.316

.465

.239

.426

Expects nancial situation to

improve in next yearb
Expects nancial situation to
worsen in next yearb

.318

.465

.290

.454

.125

.330

.081

.273

GHQ12 score
Aged
Aged
Aged
Aged
Aged

1625
2534
3544
4554
5564

years
years
years
years
years

old
oldb
oldb
oldb
oldb

Sample size
a

2193

2193

Balanced panel sample of heads of households present in both the 1995 and 2000 waves of the British
Household Panel Study.
b
The proportion of the sample is reported, not the mean, where indicated.
c
All means reported are for those who have non-zero values of each nancial variable. The means for
the whole sample can be obtained by multiplying by the proportion who have non-zero values of the
relevant variable.

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

661

Appendix B. The relevant British Household Panel Survey questions

These are self-completion questions which respondents undertake in the presence
of an interviewer.
In this paper we use the inverse of the GHQ12 caseness score (see Goldberg &
Williams, 1988, pp. 1112, for a detailed discussion). Therefore we assign a score
of 1 to a response indicating a high level of psychological well-being (i.e. the rst
two categories) and a score of 0 otherwise.
The General Health Questionnaire 12 Score Questions
Instruction: We should like to know how your health has been in general over the
past few weeks.
Please answer ALL questions (indicating) which (choice of answer given in brackets below each question) you think most applies to you.
HAVE YOU RECENTLY:
1. been able to concentrate on whatever youre doing?
(better than usual; same as usual; less than usual; much less than usual)
2. lost much sleep over worry?
(not at all; no more than usual; rather more than usual; much more than usual)
3. felt that you are playing a useful part in things?
(more so than usual; same as usual; less so than usual; much less than usual)
4. felt capable of making decisions about things?
(more so than usual; same as usual; less so than usual; much less than usual)
5. felt constantly under strain?
(not at all; no more than usual; rather more than usual; much more than usual)
6. felt you couldnt overcome your diculties?
(not at all; no more than usual; rather more than usual; much more than usual)
7. been able to enjoy your normal day-to-day activities?
(more so than usual; same as usual; less so than usual; much less than usual)
8. been able to face up to your problems?
(more so than usual; same as usual; less so than usual; much less than usual)
9. been feeling unhappy and depressed?
(not at all; no more than usual; rather more than usual; much more than usual)
10. been losing condence in yourself?
(not at all; no more than usual; rather more than usual; much more than usual)
11. been thinking of yourself as a worthless person?
(not at all; no more than usual; rather more than usual; much more than usual)
12. being feeling reasonably happy; all things considered?
(more so than usual; same as usual; less so than usual; much less than usual).
References
Argyle, M. (1989). The psychology of happiness. London: Routledge.
Bank of England. (2004). Ination report, February. Available from http://www.bankofengland.co.uk.

662

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

Bertaut, C., & Haliassos, M. (2002). Debt revolvers for self-control. Mimeo., University of Cyprus.
Bertrand, M., & Mullainthan, S. (2001). Do people mean what they say? Implications for subjective survey
data. American Economic Review, Papers and Proceedings, 91, 6772.
Bridges, S., & Disney, R. (2004). Use of credit and debt among low-income families in the UK: An
empirical analysis. Fiscal Studies, 25, 125.
Brown, S., Garino, G., Taylor, K., & Wheatley Price, S. (2005). Debt and nancial expectations: An
individual and household analysis. Economic Inquiry, 43, 100120.
Campbell, A., Converse, P., & Rodgers, W. (1976). The quality of American life. New York: Russell Sage.
Clark, A. (2003). Unemployment as a social norm: Psychological evidence from panel data. Journal of
Labor Economics, 21, 323351.
Clark, A., & Oswald, A. (1994). Unhappiness and unemployment. Economic Journal, 104, 648659.
Clark, A., & Oswald, A. (1996). Satisfaction and comparison income. Journal of Public Economics, 61,
359381.
Clark, A., & Oswald, A. (2002). Well-being in panels. Mimeo., DELTA, CRNS, France.
Cox, D., & Jappelli, T. (1990). Credit rationing and private transfers: Evidence from survey data. The
Review of Economics and Statistics, 72, 445454.
Cox, D., & Jappelli, T. (1993). The eect of borrowing constraints on consumer liabilities. Journal of
Money, Credit and Banking, 25, 197213.
Crook, J. (2001). The demand for household debt in the USA: Evidence from the 1995 survey of consumer
nance. Applied Financial Economics, 11, 8391.
Darity, W., Jr., & Goldsmith, A. (1996). Social psychology, unemployment and macroeconomics. Journal
of Economic Perspectives, 10, 121140.
Davies, E., & Lea, S. (1995). Student attitudes towards student debt. Journal of Economic Psychology, 16,
663679.
Easterlin, R. (1974). Does economic growth improve the human lot? Some empirical evidence. In P. David
& M. Reder (Eds.), Nations and households in economic growth. New York, London: Academic Press.
Easterlin, R. (1995). Will raising the incomes of all increase the happiness of all? Journal of Economic
Behaviour and Organisation, 27, 3548.
Fielding, A. (1999). Why use arbitrary points scores? Ordered categories in models of educational
progress. Journal of the Royal Statistical Society, Series A, 162, 303328.
Frijters, P., Haisken-DeNew, J., & Shields, M. (2004). Money does matter! Evidence from increasing real
incomes and life satisfaction in East Germany following reunication. American Economic Review, 94,
730740.
Frey, B., & Stutzer, A. (2000). Happiness, economy and institutions. Economic Journal, 110, 918938.
Frey, B., & Stutzer, A. (2002). What can economists learn from happiness research? Journal of Economic
Literature, XL, 402435.
Godwin, D. (1997). Dynamics of households income debt and attitudes towards credit. The Journal of
Consumer Aairs, 31, 303325.
Goldberg, D. (1972). The detection of psychiatric illness by questionnaire. Oxford: Oxford University Press.
Goldberg, D., & Williams, P. (1988). Users guide to the general health questionnaire. NFER-Nelson.
Greene, W. H. (2003). Econometric analysis (5th ed.). Prentice-Hall.
Gross, D., & Souleles, N. (2002). Do liquidity constraints and interest rates matter for consumer
behaviour? Evidence from credit card data. The Quarterly Journal of Economics, 117, 149
185.
Headey, B., & Wooden, M. (2004). The eects of wealth and income on subjective well-being and ill-being.
Institute for the Study of Labor (IZA) Discussion Paper, no. 1032, IZA, Bonn.
Headey, B., Muels, R., & Wooden, M., (2004). Money doesnt buy happiness... or does it? Are
consideration based in the combined eects of wealth, income and consumption. Institute for the Study
of Labor (IZA) Discussion Paper, no. 1218, IZA, Bonn.
Jahoda, M. (1982). Employment and unemployment: A social psychological approach. Cambridge:
Cambridge University Press.
Jahoda, M. (1988). Economic recession and mental health: Some conceptual issues. Journal of Social
Issues, 44, 1323.

S. Brown et al. / Journal of Economic Psychology 26 (2005) 642663

663

Jappelli, T. (1990). Who is credit constrained in the US economy? The Quarterly Journal of Economics,
105, 219234.
Kahnemann, D., Diener, E., & Schwarz, N. (1999). Well-being: The foundations of hedonic psychology.
New York: Russell Sage Foundation.
Katona, G. (1975). Psychological economics. New York: Elsevier.
Lane, R. E. (1991). The market experience. Cambridge: Cambridge University Press.
Lea, S., Webley, P., & Levine, R. (1993). The economic psychology of consumer debt. Journal of Economic
Psychology, 14, 85119.
Lea, S., Webley, P., & Walker, C. (1995). Psychological factors in consumer debt: Money management,
economic socialization, and credit use. Journal of Economic Psychology, 16, 681701.
Livingstone, S., & Lunt, P. (1992). Predicting personal debt and debt repayment: Psychological, social and
economic determinants. Journal of Economic Psychology, 13, 111134.
Marmot, M., Ry, C., Bumpass, L., Shipley, M., & Marks, N. (1997). Social inequalities in health: Next
questions and converging evidence. Social Science and Medicine, 44(6), 901910.
National Association of Citizens Advice Bureau. (2003). In Too deep: CAB clients experience of debt.
Available from http://www.nacab.org.uk.
Oswald, A. J. (1997). Happiness and economic performance. Economic Journal, 107, 18151831.
Roberts, R., Golding, J., & Towell, T. (1998). Student nance and mental health. The Psychologist
(October), 489491.
Shields, M. A., & Wheatley Price, S. (forthcoming). Exploring the economic and social determinants of
psychological health and perceived social support in England. Journal of the Royal Statistical Society
Series A.
Stradling, S. (2001). The psychological eects of student debt. In A. Scott, A. Lewis, & S. Lea (Eds.),
Student debt: The causes and consequences of undergraduate borrowing in the UK (pp. 5975). Bristol:
The Policy Press.
Taylor, M. (2002). Tell me why I dont like Mondays: Investigating day of the week eects on job
satisfaction and psychological well-being. Institute for Social and Economic Research Working Paper,
no. 2002-22, University of Essex.
Van Praag, B., & Frijters, P. (1999). The measurement of welfare and well-being: The Leyden approach. In
D. Kahneman, E. Diener, & N. Schwarz (Eds.), Well-being: The foundations of hedonic psychology.
New York: Russell Sage Foundation.
Webley, P., & Nyhus, E. K. (2001). Lifecycle and dispositional routes into problem debt. British Journal of
Psychology, 92, 423446.
Weich, S., & Lewis, G. (1998). Poverty, unemployment and common mental disorders: Population-based
cohort study. British Medical Journal, 314, 591595.
Winkelmann, L., & Winkelmann, R. (1998). Why are the unemployed so unhappy? Evidence from panel
data. Economica, 65, 117.