North-Holland
Psychology
8.5
14 (1993) 85-119
February
6, 1992; accepted
August
31, 1992
Questionnaires
were distributed
to groups of people with either no debt, mild debt, or
serious debt to a public utility company.
Serious debtors were found to differ from the
Non-debtor
group on economic, sociological,
and psychological
variables: economic resources,
economic need, social support, attitude forming variables and attitudes all made independent
contributions
to the prediction of group membership
and the extent of self-reported
debt. Mild
debtors were generally
intermediate
between
Non-debtors
and Serious debtors.
Debt was
strongly correlated
with economic factors. Many results indicated that debt is a consequence
of
adverse family economic conditions:
Serious debtors were of lower socioeconomic
class, had
lower incomes, were less likely to own their own homes (and much less likely to own them
outright), had more children and were more likely to be single parents. They were also younger.
Social and psychological
factors were also found to be related to debt: Serious debtors were less
likely to claim Nonconformist,
Agnostic or Atheist religious views, and they had slightly more
permissive attitudes towards debt, although no group showed a general tendency to approval of
debt. They knew more other people who were in debt, and they were less likely to think that
their friends or relations would disapprove if they knew they were in debt. Multivariate
analyses
showed that economic, social and psychological
variables all had independent
correlation
with
debt. These results suggest that debt is stronly influenced
by adverse economic circumstances,
but that social and psychological
factors are also important.
The conditions for the development
of a self-sustaining
culture of debt do exist.
Introduction
Compared with many other
been relatively little explored
Correspondence
too:S.E.G. Lea, University of Exeter, Dept. of Psychology, Washington
Singer
Laboratories,
Exeter EX4 4QG, UK.
* We would like to thank Welsh Water plc for their financial support for this research,
and
particularly
Noel Hufton, Jeffrey Phillips and Wayne Rees for their interest and practical help.
We are also most grateful to Nicola Crichton and John Hinde for statistical advice, to Cathy
Walker and Emma Davies for background
research and much discussion, and to an anonymous
reviewer for much helpful advice. An earlier version of this paper was presented
to the joint
meeting of the International
Association
for Research in Economic Psychology and the Society
for the Advancement
of Socio-Economics,
Stockholm, June 1991.
0167-4870/93/$06.00
0 1993 - Elsevier
Science
Publishers
86
debt
87
88
debtors can be found through court records (e.g. Sullivan et al. 1989)
or advice services (e.g. Cameron and Golby in press). Debt is both less
frequent and less visible. It is also quite likely not to be acknowledged.
Like many other behaviours which are either illegal or not fully
socially acceptable, it is difficult to research by conventional surveys
(Robins 1963). 0 ne of the few studies to have looked at this kind of
ordinary debt is a study by Lunt and Livingstone (1991), and,
significantly, this looks at the explanations people give of other
peoples debts, not at individuals own experience of debt.
The present study looks at both crisis debt and at the more elusive
category of non-crisis (but non-agreed) debt. It took advantage of an
offer of co-operation from a major regional utility company, Welsh
Water plc. The company is the monopoly supplier of water and
sewerage services to approximately 1 million households in Wales
(and in part of one neighbouring English county). They were thus able
to supply us with random samples of up-to-date names and addresses
of householders.
Far more important, they were able to classify
customers according to their credit status with the company, from
those with no water debt to those with major debt problems. This
meant that (with appropriate measures to ensure confidentiality), we
could mail to relatively large numbers of people who were known to
be in debt to different degrees; and we could test the truthfulness of
peoples answers in a sensitive financial area.
The general aim of the study was thus to expand our knowledge of
the natural history of debt. From the previous literature, it was clear
that we could expect a large number of variables, some economic,
some sociological, and some psychological, to be correlated with debt.
An important question, however, is how far they make independent
contributions. The data of Livingstone and Lunt (1992) suggest that
the contributions of attitudes and economic variables are largely
independent, but they dealt largely with credit rather than debt, and
their measures of credit/debt depended on self-report. We wanted to
see whether their results would be reproduced with a sample of
including more people in more serious financial difficulties, and using
objective measures of indebtedness. We classified the variables we
considered as follows:
- Economic resources: Income, employment status, home ownership;
- Economic need: Household size and constitution;
89
These classes of variables form a rough hierarchy from the economic to the psychological. From an economists point of view, even
the most economic of them are fairly soft, since they were obtained
by self-report, while from the psychological point of view even the
most psychological of them are fairly superficial. However, since
debt is an economic behaviour, the simplest and in a sense the most
conservative hypothesis is that it will have economic causes, and any
correlation with other variables will arise only as a result of mutual
dependence. We therefore wished to test whether each successive
class of variables had any independent effect after the previous class
had been taken into account.
Method
Sample
90
codes. We selected
100 labels for each code/county
combination,
largely at random but rejecting (i) addresses outside the county area,
presumably
those of absentee landlords, and (ii) any where the address was printed in an eccentric form which might have identified
Welsh Water as the source of the labels, since we were anxious to
avoid biasing the answers to questions about debts to public utilities.
Different coloured questionnaires
were sent to addresses bearing each
code. For one code/county
combination,
only 40 labels were provided. In all, therefore,
2640 questionnaires
were mailed out. This
represents
about 0.25% of the households
in the Welsh Water area.
After all questionnaires
had been mailed out, surplus labels were
shredded, and Welsh Water then informed us which credit categories
corresponded
to each code. As an added measure to ensure confidentiality, none of those handling the questionnaires
had any substantial
current connection with Wales, and in fact none of us recognised any
of the names or addresses used.
Questionnaire
The questionnaire
consisted of a 4-page A4 leaflet. It contained:
(a) A question about the respondents
current financial position, to
be answered on a 5-point rating scale, followed by questions about
current debts to ten likely creditors,
to be answered on a 4-point
rating scale from None to Over &500; to allow a check on the
sampling procedure,
debt to The water company was included in this
list. The ten creditors
are shown in Figure lb. Note that moneylender, one of the creditors used, in everyday British English implies
a specialist (and often not very respectable)
trader, not a bank or
other general financial business.
(b) Questions about how urgent it would be to repay a debt to the
same ten creditors, and what the respondent
had in fact done when
last asked for repayment by them; plus three questions which between
them gave the respondents
order of priority for repayment to the four
public utilities (gas, electricity, water and telephone).
(cl Questions about whether other people had told the respondent
they were in debt, about the reaction they would expect from others
who knew the respondent
was in debt, and about people or organizations they had spoken to about any current debt.
(d) Twelve questions
exploring
attitudes
to debt, spending and
saving in general, intended to serve as a pilot towards the develop-
91
Results
Return rates
92
of consumer
debt
debtor categories respectively. After allowing for questionnaires returned as undeliverable by the post office, the return rate was 16%
overall, and 23%, 14% and 11% in each credit category. There were
no substantial differences in return rate between county areas. Adjusting for the fact that the two debt groups are minorities within the
population, these return rates would correspond to an approximately
21% return from a random sample of households. The proportion of
male respondents was 56%. Four respondents wrote separately, requesting information about the results of the research. One telephoned to complain about being contacted; one returned a defaced
questionnaire, and ten returned completely blank forms, presumably
also a form of protest.
Many questionnaires lacked answers to some questions. In the
analyses that follow, all available data for each question have been
used, even if the questionnaire was incomplete in other respects.
There was no obvious multimodality in the distribution of return
lags. Accordingly, those questionnaires that were returned after more
than 11 days were classified as late, and used as a model for
non-returns (cf. Oppenheim 1966). They constituted 38% of the total
usable returns, including 37% of the Non-debtor group, 43% of the
Mild debtors, and 33% of the Serious debtors.
Manipulation checks
93
Subjective
financial
position
50%
40%
30%
20%
10%
0%
Very
Just
difficult
coping
Could
Subjective
m
No debt
financial
@8
Mild debt
Self-reported
and credit
m
% with debts
Fig.
1. Validation
No debt
About
be better
right
Very
healthy
position
a
Serious
debt
debt
group
a
Serious
debt
to
group related
debt.
to subjective
(a)
94
of consumer
debt
corresponding
figures for Mild and Serious debtors were 36%
16%. The creditors
to whom most respondents
were in debt
credit card companies (37% acknowledging
some debt, 11% for
X500); fewest acknowledged
debt to a close friend (5%, 1% for
fSOO>.
and
were
over
over
Repayment
Figures 2 and 3 show responses to questions about how urgent it
would be to repay a debt to ten possible creditors. Here there are
rather few differences
between credit groups. We also asked what
people had done on the last few occasions when they were asked to
repay money by the same ten creditors. For the water company, the
modal responses of the Non-debtor
group was I paid up immediately
(43%), while for the debtors it was I waited for the final demand
(Mild debtors, 54%; Serious debtors 66%). Scoring I paid up immediately as 4 and I didnt repay it as 1, the mean immediacy ratings for
repaying the other nine creditors were 3.0, 2.7 and 2.4.
Figure 4 shows the rankings of the urgency of repaying the four
public utilities in the three credit rating groups. These rankings
showed fair consistency (Kendalls W coefficient of concordance
= 0.32
for the sample as a whole, and 0.32, 0.36 and 0.27 for Non-debtor,
Mild debt and Serious debt groups). For all three credit status groups,
most urgency was attached to repaying an overdue electricity bill, and
least to a telephone
bill; gas and water bills were intermediate,
with
slightly more urgency attached to the gas bill.
First order differences between credit rating groups
Unless there is a specific comment
to the contrary,
all group
differences
described here reached at least the 0.05 significance level,
using chi-squared
tests. The differences
meeting this criterion
are
listed in Table 2, which also reports the actual significance
levels
reached.
Economic resources
We have already reported the differences
the subjective financial strain they reported.
95
Urgency
of
Welsh
repaying
different
creditors
Electricity
Water
Urgency
I
No debt
!Tm
British
Mortgage
or rent
(1)
board
Ol repayment
Mdd
debt
srnova
dab,
Telecom
TV rental
707.,
Fig. 2. Distributions
of reported urgencies
(1): Utilities,
rating
groups
specific measures of economic resources showed similar trends. Family income was asked for directly on the questionnaire (in the broad
categories used in the figure); both debt groups reported lower income than the Non-debtors. Occupational status was deduced from
96
LJrgency
of
Welsh
repaying
Fig. 3. Urgencies
creditors
(2)
Water
Fami1.y
Credit
different
card
Close
Money
friend
lender
groups
(2):
Priorities
Electricity
Fig. 4. Distributions
for
repayment
British
board
of priorities
97
for repayment
of a hypothetical
three credit-rating
groups.
Gas
in the
socioeconomic class for all three credit categories were bimodal, both
debtor groups clearly tended to lower occupational classes than the
Non-debtors.
Housing conditions also showed marked variation (Figure 6a). The
three main categories represented were outright ownership, ownership through a mortgage, and renting from the council (private renting
is probably underrepresented
in the sample since in many cases
landlords would be responsible for water charges). Among the Nondebtors, the modal category was outright ownership (50%), among
Mild debtors it was ownership via mortgage (53%), while among
Serious debtors it was council tenancy.
Economic need
98
family
Annual
income
and
credit
group
30%
20%
10%
0%
under
L5000-L10000
L5000
L10000-Ll5000
Annual
m
No debt
Socio-economic
&8
over
L15000-L20000
family
L20000
income
Mild debt
class
Serious
and
debt
credit
group
30%
25%
20%
15%
10%
5%
0%
Fig. 5. Economic
No debt
resources:
distributions
k88$ Mild
debt
of reported
groups.
income
Serious
debt
and occupational
Housing
type
and
credit
99
group
60%
50%
40%
30%
20%
10%
0%
Owned
outright
Mortgage
Private
rent
Housing
m
No debt
Children
Council
Family/friends
type
Mild debt
in household
rent
Serious
debt
and credit
group
60%
40%
20%
0%
None
m
Fig. 6. Economic
m
Number
of
No debt
3 or more
children
Mild debt
in
household
/
Serious
debt
resources
and economic
needs: distributions
of (a) housing
numbers of children in household in each credit group.
status
and (b)
100
either marital status or the number of adults, though the two debtor
groups were somewhat less likely to be either married or living with a
partner, and more likely to be in households with one or three or
more adults rather than two. However, there were marked differences
in relation to the numbers of children in the household (Figure 6b).
The mode for all groups was none, but childless households accounted
for 75% of the Non-debtor group, 54% of the Mild debtors, and only
32% of the Serious debtor group, while those with 3 or more children
accounted for l%, 11% and 19% of the three groups respectively.
Although the number of adults in the household did not have a
significant effect as such, a variable constructed from it did. We coded
the 35 respondents who reported one adult and at least one child in
the household as single parents. They formed 5% of the Non-debtors,
6% of the Mild debtors, and 19% of the Serious debt group.
Social support for debt
We considered the following variables that might have been expected to contribute to the respondents attitude towards debt: gen-
Other
people
Are other
people
you
and
debt
know
in debt?
101
60%
50%
40%
30%
20%
10%
0%
Nearly all
Several
People
Others
you
know
attitudes
of LlOO for
if you
electricity,
had
water
debts
or gas
60%
50%
40%
30%
20%
10%
n4.
rhaapprore
Not
understand
Attitude
m
No debt
nderatand
but...
nmlc
it
norms,
of friends/family
Mild debt
Serious
debt
Fig. 7. Social support for debt in the three credit groups: distributions
of (a) numbers of other
people, known to the respondents
to be in debt and (b) expected attitudes of others if they knew
the respondent
was in debt.
102
but there were significant effects of age and reported religious affiliation (Figure 8). Although the Age effect is not completely straightforward, both debt groups were substantially younger than the Non-debtors. In all credit categories, the modal religious affiliation claimed
was Anglican, but in the Non-debtor
group, there were elevated
proportions
of Non-conformists
and Agnostics/Atheists.
Numbers of
Roman Catholic and Other responses were correspondingly
higher in
the debtor groups. Only one respondent
claimed to be Moslem and
none to be Hindu, suggesting that there was a low rate of return from
ethnic minority communities.
Attitudes to debt
Using the total sample, Cronbachs
(Y for the 12-item scale was
found to be 0.70, indicating an acceptable degree of scale homogeneity. Table 1 shows that all but two items (Big companies
can look
after themselves and A credit card is a ticket to careless spending)
had at least moderate
correlations
with the total of the remainder.
Although the (Y value could have been improved marginally by dropping these two items, in the interests of retaining a broad approach to
debt attitudes (and because a full psychometric
investigation is planned
for later in the research programme),
all items were retained. Since
the overall sample is not population-representative,
being overweighted with debtors, the reliability analysis was rerun on the Nondebtor group alone; the results were essentially unchanged.
Principal
components
analysis of the responses to the 12 items produced only
three components
with eigenvalues
greater than 1 (3.13, 1.91 and
1.06). The first component
could be identified as an approval of debt
measure, with all items loading appropriately
on it; no interpretation
of the other two major components
was possible. It was therefore
judged appropriate
to use the 12 items as a scale of attitudes towards
debt. Figure 9 shows the mean scale scores for the three respondent
groups, scoring Very strongly agree as 1 and Very strongly disagree
as 7. Clearly all groups tend to disagree with pro-debt statements, but
disagreement
is weaker in the debt groups.
Multivariate analysis
Dependent variables
We attempted
to predict four measures of indebtedness:
the credit
rating group; whether or not the respondent
reported any debt; and
and
Age
credit
103
debt
group
35%,
20%
15%
10%
5%
0%
under
25
25-34
Age
No debt
Religious
65-74
55-64
75 and
over
group
Mild debt
affiliation
Serious
debt
and credit
group
60%
50%
40%
30%
20%
10%
0%
Roman
Catholic
Anglican
Non-conformistAgnostic/Atheist
Religious
m
Fig. 8. Attitude-forming
No debt
variables:
m
distributions
Other
affiliation
Mild debt
Serious
debt
affiliation
104
Table 1
Items used in the debt attitudes
scale, together
with scoring
correlations.
Item
Scoring
Item-total
correlation
a if item
deleted
+
_
0.29
0.40
0.69
0.67
+
+
_
-
0.51
0.26
0.43
0.30
0.48
0.07
0.17
0.45
0.66
0.69
0.67
0.69
0.67
0.72
0.71
0.66
0.35
0.45
0.68
0.67
+
_
+
+
_
+
for those who did report some debt, two measures of its severity,
scored as follows. The first measure emphasised the breadth of
indebtedness, by scoring each creditor from 1 (no reported debt) to 4
(debts of over &SO0reported), and totalling these scores across the 10
creditors. The second measure emphasised the depth of indebtedness,
by replacing these ordinal scores by midpoints of the answer categories (0, f50, &300 and flOO0 respectively) before summing. The two
measures are referred to below as the breadth and depth measures
of self-reported debt. Because many respondents failed to answer one
or two questions, the effective sample sizes for the analysis of creditrating group is reduced to 332; it was substantially lower for the
self-reported debt analysis, with only 253 usable responses. Of these,
143 reported some debt, so this was the sample size for the third
analysis. Prediction of credit-rating group was carried out by ordered
logit analysis, prediction of the presence or absence of self-reported
debt by logit analysis, and prediction of level of self-reported debt by
multiple regression.
The ten potential creditors listed on the questionnaire were a
somewhat heterogeneous group, so the responses to these items of
those respondents acknowledging some debt were subjected to reliability and principal components analysis, before and after converting
Attitudes
1 =strongly
Agreement
toward
disagree,
with pro-debt
,
No debt
Fig. 9. Attitudes
debt
7=strongly
agree
statements
Mild debt
Credit
105
group
I
/
Serious
debt
with a pro-debt
attitude on the
variables
106
Table 2
Summary of differences
between
least the 0.05 level are shown.
of consumer debt
categories.
Only differences
significant
Variable
Tendency
groups
Income
Socioeconomic
status
Household
financial
position
Housing
Lower
Lower status
18.02
33.53
8
14
More difficult
100.14
0.00005
More council
tenants, fewer
own houses
outright
More
95.56
0.00005
66.20
0.00005
More
More
16.75
68.74
2
6
0.0005
0.00005
Less disapproving
49.24
0.00005
Younger
Fewer nonconformists,
atheists and
agnostics
Less unfavourable
65.80
23.14
12
8
0.00005
0.01
62.40
Children in
household
Single parents
Number of
friends/family
who are in debt
Attitude of
friends/family
to respondents
debt
Age
Religious
affiliation
Attitudes towards
debt
Debts to other
creditors
of debtor
x2
14.93
at at
(Y
df
2,377
2
0.05
0.01
0.0005
0.00005
multivariate
analyses. In some cases categories
were combined
because group sizes would otherwise have been too small. The final list
of variables used is given in Table 4. They include 7 ordinal or
dichotomous
variables, and three variables (housing status, age, and
religious affiliation),
represented
by 3, 4 and 2 dummy variables
respectively,
making a total of 16 regressors.
It is to be expected that data from a study of the present sort will
show intercorrelation
of the independent
variables.
It is therefore
important to test whether there is sufficient multicollinearity
between
107
108
Table 3
Results of multivariate
analyses: (a) Hierarchical
analysis. Each row shows the effect of adding a
group of variables
to the predictor
set while retaining
all those given above. Entries under
success give the proportion
of individuals correctly categorized
after adding the new group of
variables for the first two analyses, and the R*-adjusted
value for the third. x2 and F statistics
test the significance of the improvement
in predictiveness
produced by adding them.
Predictor group
(with increment
and residual (df)
Credit-rating
Self-reported
debt
Success
Success
x2
x2
Breadth
measure
Success
CR*-adj.)
Depth
measure
F
Success
(R2-adj.)
Economic
resources
58%
92.12 71%
45.56
8%
4.09 =
9%
4.64 b
(4,139)
Economic
need
60%
15.30 c 71%
6.13 a
9%
1.75
9%
0.53
(2,137)
Social support
debt
for
61%
10.29
74%
33.30 c 20%
(2,135)
Attitude forming
variables
60%
12.47
77%
17.81 b 23%
(6,129)
Attitudes
59%
6.56 a 79%
6.15 a 28%
10.48 c 15%
1.96
18%
8.54 b 21%
6.30 b
1.83
6.15 a
(1,128)
a p < 0.05;
h p < 0.01;
= p < 0.001.
109
Table 4
Results of the multivariate
analyses: (b) Overall analysis. Entries in the main body of the table
are regression
coefficients
for each predictors
independent
contribution
to the dependent
variable.
Positive coefficients
indicate an increasing
tendency
to indebtedness
for increasing
values of the predictor.
Dependent
Predictor
Credit
rating of
grow
Economic resources
Family income
Occupational
class
Housing (difference from mortgage)
Owned outright
Rented or living with family or
friends
Economic need
Children in household
Single parent
Social support for debt
Number of family or friends in debt
Attitudes of others to respondents
debt
Attitude forming variables
Age (differences
from 35-44):
under 25
25-34
45-54
55 and over
Religion (differences
from
Anglican):
Roman Catholic or other
Non-conformist,
agnostic or atheist
Attitudes
a p < 0.05;
cp
< 0.001.
-65
-11
- 0.532 a
- 0.456
- 0.735 a
- 0.225
- 0.532
- 395
0.987 b
0.509
- 1.532 =
-460
0.430 c
- 0.241
0.153
1.507
0.284
0.496
0.270 =
0.714 b
1.119 =
0.093
0.462 a
0.081
1.137
0.272
0.818 a
- 0.027
16.030
1.564 b
- 0.030
- 0.244
1.774
0.530
- 1.358
- 0.463
754 =
219
- 280
-43
0.428
- 0.281
0.099
- 0.034
0.039
- 0.681
- 146
-34
0.649 a
253
108.94
1.133 b
143
27.5%
4.62
df)
Depth
measure
-0.152
- 0.040
332
137.17 =
b p < 0.01;
Existence
selfreported
l-l.+
F;L
- 0.281 a
0.007
0.477 =
variable
63
7
250 b
-53
267 a
143
21.3%
3.39 =
110
dependent
variables are on differing scales, numerical comparisons
of
these coefficients
either across the rows or down the columns of the
table are not always meaningful.
However,
both the signs of the
coefficients
and comparisons
of ratios of coefficients
do have a direct
interpretation.
It can be seen that not all the variables in the groups
had significant effects on all dependent
variables, and some had no
significant effect on any variable. In the Economic Resource group,
occupational
class had no significant effect on any variable (indicating
that the effect seen in Figure 5b can be explained by the correlation of
class with such variables as income and housing status). In the Social
Support group, the number of other people respondents
knew who
were in debt had an independent
effect on all dependent
variables,
but respondents
estimates of others attitudes to their (the respondents) debt had no significant effects. In the Attitude-forming
variables group, Age showed some independent
effects but religious
affiliation did not. The only anomaly, relative to the first order effects
shown in the figures, occurs in the Economic Resources group, where
both measures of the extent of debt show a negative effect of renting
rather than owning a house on a mortgage.
The relative effectiveness
of the predictors (indicated by the significant levels they reach) were slightly different for the different dependent variables. For credit-rating
group, the variables that emerged as
having the largest independent
effect were the number of children in
the household and the housing variables. For the existence of self-reported debt, the number of family and friends in debt and age group
(with young respondents
being most likely to report debt) were most
important. For the extent of self-reported
debt, number of family and
friends in debt and (especially) attitudes were most significant. Table
4 also shows some interesting reversals. House renters are significantly
more likely to be in one of the debt groups, and (non-significantly)
more likely to self-report
debts, but if they do report debts, the
amount is significantly less than for mortgage holders. Those in the
45-54 age group are significantly more likely to be in the debt groups,
but (non-significantly)
less likely to self-report
debt, and if they do
report any, report less than the modal (35-44) age group.
As a check on the possible distortions introduced
by non-returns,
the factor of late return was introduced
into all four multivariate
analyses after obtaining the results reported
in Table 4. It had no
significant effect on any dependent
variable, and the general pattern
of help by respondents
in the Serious
111
debt group.
Source of help
Percentage
of the group
reporting using source
Person/institution
to which money is owed
Friend
Family
Bank or building society manager
Unemployment/Social
security office
Citizens Advice Bureau
Local council
Other
33%
32%
28%
26%
16%
14%
13%
3%
112
Serious debt group. The only exception is that those not reporting
debt claim that their family or friends would be less disapproving of
their (the respondents) debt if they knew of it.
Discussion
Representativeness of the sample
113
and seriously in debt. First, when asked directly about debt to the
water company, almost none of the Non-debtor group, a third of the
Mild debt group, and a substantial majority of the Serious debt
groups, acknowledged some debt. Bearing in mind the lag between
selection of addresses and return of questionnaires, the anomalies are
not a major cause for concern, though they may indicate some
tendency to self presentation. Second, on every measure that could
reflect financial strain, the three groups differed in the expected way.
There was, indeed, nothing at all special about water charge debt.
Respondents whom we knew to owe money to the water company also
Table 6
Significant
any form
company.
then there
differences within the Serious debt group according as to whether or not they sought
of help, and whether
or not they reported
that they had any debt to the water
Note that if there is no entry in the X2 column for one of the two partitions
group,
was no significant difference for that grouping for the variable concerned.
Variable
Subjective financial
position
Overdue payments on
mortgage or rent
Debt to water company
Debt to electricity
board
Debt to British Gas
Debt to British
Telecom
Action on last credit
card debt
Number of family
friends in debt
Reactions of
friends/family
to
respondents
debt
Help sought from
local council
Marital status
Housing
status
Income
a p < 0.05;
bp < 0.01.
Responses of non-help
users and/or nonacknowledgers
of
debt:
X2
(df)
For selfreport of
water debt
Less difficult
11.98 b (3)
14.43 b
Less
9.19 a (3)
Less
Less
9.46 a (3)
10.68 a (3)
Less
Less
10.17 a (3)
6.17 a (2)
Repaid
sooner
Fewer
11.68 = (3)
10.44 a (3)
Less disapproving
8.19 = (3)
Less
3.92 a (1)
8.97 a (3)
10.63 a (4)
10.52 a (4)
9.91 a (4)
12.41 a (4)
114
acknowledged
debts to other creditors,
and those whose water debt
was more serious acknowledged
more additional debts. Public utility
companies differ from some of the other creditors we asked about, in
that they do not set out to lend people money, and people do not ask
their permission before borrowing from them (i.e. paying their bills
late or not at all). Furthermore
they have a well-structured
system of
reminders,
court proceedings,
and ultimately disconnection,
the first
stages of which, at least, are well known to their customers. However,
none of these differences
seemed to produce behaviour that was very
different from that shown towards other creditors.
Debt and financial strain
The overwhelming
impression
from the results is that debt is
primarily a problem of poverty, and in particular
of family poverty.
Compared with the Non-debtors,
the Serious debt group have lower
household income, are of lower socioeconomic
class, are less likely to
own their homes (and much less likely to own them free of a
mortgage), and have more children. Both these results agree with data
obtained by Berthoud and Kempson (1990) concerning problem debts.
It is particularly
striking that only 1% of the Non-debtor
group had
three or more children in their households. With the exception of the
effects of class, all these results are confirmed as independent
of one
another by the multivariate
analyses. There were also trends, though
they fall short of significance, for debtors to be less likely to be living
in two-adult households built round stable relationships.
On first-order
analysis, the debtor groups emerge as younger (again a finding agreeing with Berthoud and Kempsons, 1990, analysis), so presumably their
children are younger; it should be noted, however, that the multivariate analysis of credit-rating
group suggests a subsidiary
peak of
indebtedness
in the 45-54 age group. In all these respects, the Mild
debtors are intermediate
between the Non-debtors
and the Serious
debtors, which adds credibility to the results. The results here are
consistent with the classic cross-sectional
econometric
results on the
saving ratio (e.g. Bean 19461, which show that as income increases, so
does the proportion
of it that is saved; since running up debt is simply
a form of negative saving, it is in no way surprising that it is more
often done by those with low incomes.
There is considerable
evidence,
then, that debt is an unsought
consequence
of a households financial difficulties. Those difficulties
115
116
psychology of consumer
debt
young and to have low incomes. However, Cameron and Golby were
looking at crisis debtors, and Berthoud and Kempson at people who
were using credit rather than running up debts. Those with high
income can safely use larger credit facilities, and if things go wrong
are likely to be left with larger unmanageable debts. But our data
show that, as one would expect, everyday debt is nonetheless associated with low income.
Debt attitudes and a debtor culture?
But if economic factors play an important part in debt, social and
psychological factors are also relevant. Age effects were at least partly
independent of both economic and attitude variables, and in so far as
that is true, they are presumably also mediated by psychological
processes. The religious backgrounds of the groups were somewhat
different, and it is hard to avoid seeing the remnants of the Victorian
Non-conformist conscience in the higher numbers of this group among
the Non-debtors. The same could be said of the agnostic/atheist
group, also likely to represent a thought-out rather than a casually
acquired position. It should be noted, however, that none of the
religious affiliation effects emerged as independently significant in the
multivariate analysis (though the same trends were evident), so these
differences may just reflect the different characteristic affiliations of
those who are, for other reasons, more and less likely to be in debt.
There was clearly quite a lot of social support for debt among the
debtor groups. They were much more likely to know others who were
in debt, and less likely to think that others would disapprove of their
debts; 33% of Serious debtors thought that their friends would think it
normal to owe at least flO0 to a public utility company. While this is a
striking result, it is not a surprising one. The general trend of the
results shows that debt is primarily a consequence of poverty. The
social worlds of those who have little money are likely to contain many
other people who are experiencing the same difficulties and therefore
both understand and sympathise with them. In this relatively brief
questionnaire, we could make no attempt to determine whether the
friends and relations we referred to were reference groups for the
respondents in the technical sense, but in the absence of indications to
the contrary it is reasonable to suppose that they were.
117
118
References
Bean, L.H., 1946. Relation of disposable income and the business cycle to expenditures.
Review
of Economics and Statistics 28, 199-207.
Berthoud,
R. and E. Kempson,
1990. First findings from the PSI survey: Credit and debt in
Britain. London: Policy Studies Institute.
Berthoud,
R. and E. Kempson, 1992. Credit and debt: The PSI report. London: Policy Studies
Institute.
Cameron,
S. and D. Golby, in press. An economic
analysis of personal
debt. Bulletin of
Economic Research.
Corkish, J., in preparation.
An overview of consumer credit and an analysis of factors causing
unmanageable
debt in 25-34 year old consumers.
Unpublished
Ph.D. dissertation,
University
of Wales.
Earl, P.E., 1991. The complementarity
of economic applications
of cognitive dissonance
theory
and personal construct psychology. In S.E.G. Lea, P. Webley and B.M. Young (Eds.), New
Directions in Economic Psychology. Cheltenham:
Elgar.
119