BEYOND REASONABLE DEBT
A background report on the indebtedness of New Zealand families

Part 2: Families’ characteristics and circumstances

Part 2 focuses on characteristics, circumstances and environmental factors that are likely to influence and help us explain different families’ decisions about savings and debt and their outcomes.

This part is divided into the following sections, each outlining relevant theory and overseas and New Zealand evidence:

• age and cohort effects
• relationships, children and transitions
• wealth and home ownership
• income, education and employment
• ethnicity and region
• economic and social climate and policy.

In reality, these factors do interact. We attempt to illustrate this as we outline available evidence in each of the sections. Kempson, McKay, and Willitts (2004), for instance, found that the rate of arrears was strongly related to the number of predisposing factors reported by a household, with a big jump among those with four and five predisposing factors (this finding is discussed later in this part). Causality is also difficult to determine. Kempson et el found that in most cases debt was as likely to come before as it was after other issues.

Parts 2 and 3 of this report adopt a multidisciplinary approach to exploring the variables that influence families’ savings and debt decisions and outcomes, as recommended by Livingstone and Lunt:

• Explanations for personal debt must be interdisciplinary, drawing on a range of social science disciplines. For example, economics is concerned with the effects of income and with life cycle models; demography emphasises the importance of life events; sociology considers the debtor in the context of social groups and social norms; social psychology recognises the importance of people’s social knowledge, locus of control, attitudes and values. While many factors influencing personal borrowing have been proposed, no clear conceptual model which integrates these has yet emerged and empirical studies tend to examine the role of a few factors in each study (1992, p 114).

Age and cohort effects

The standard theory of saving is captured by the life-cycle model, illustrated in Figure 8.

The life-cycle model predicts that a person’s net lifetime savings should be zero: people save during their working life the amount they intend to spend (or ‘dissave’) in their retirement. The aggregate saving rate (the collective savings of New Zealanders) should therefore also be zero if there is no real population or income growth (Coleman, 2006). By this token, borrowing is rational and predictable (if the means exist) provided people can meet their lifetime consumption needs. This process can also be thought of as consumption smoothing, as both saving and borrowing may be used to even out varying incomes or varying needs over the life course (Modigliani, in Eltis, Scott & Wolfe, 1970). Saving provides present resources drawn from past income, and borrowing provides present resources drawn from future income.

The model assumes that most individuals, or in this case families, go through predictable stages at predictable times. The life cycle model can therefore be thought of as capturing a series of age effects.[6] Normal patterns of human capital development and working life entail people having earnings streams that rise with age and then decline, so the theory of consumption smoothing implies a period of borrowing, followed by saving, followed by dissaving (as in Figure 8).

Figure 8: Life-cycle model


Of course, people borrowing before they have savings raises questions about the effect of this on lifetime wealth accumulation. People may accumulate less wealth over their working life (compare ‘a’ with ‘c’ in Figure 8), or they may have more income in the first place (compare ‘a’ with ‘b’ in Figure 8). If real incomes are rising over time, increasing use of debt by subsequent generations may indeed be rational and predictable. Substantial longitudinal data, however, would be required to test the presence of any cohort effect.[7]

Different cohorts exhibit different age and earnings profiles because of such factors as changing returns to education and skills, higher labour-force participation by women and later family-formation. There may also be growing dispersion. Thus, one could expect that saving rates and the age profiles of saving could vary significantly across cohorts because of these different age and earning profiles (and other factors relating to expectations and credit markets). There is some evidence that household saving rates do vary significantly across cohorts. In particular, those born between 1920 and 1939 have considerably higher saving rates than subsequent cohorts (Scobie & Gibson, 2003).

It would be interesting to test whether the following cohorts differ in their savings rates:

• ‘Generation X’ (born between 1965 and about 1987, although various end-dates are used). This cohort lived in a time of increased divorce, availability of oral contraception, more women in the workplace and student loan availability (from 1992 in New Zealand).
• ‘Generation Y’ (born between about 1988 and 2008). This cohort is considered to be peer-oriented (because of high rates of separation among their parents and less support from parents and grandparents); it faces higher educational costs than previous generations, and tends to be highly educated, ambitious and brand-conscious.

The impact of increasing life expectancy, policies like KiwiSaver[8] and changing patterns of work on the saving and debt patterns of different cohorts would also be interesting to explore.

Overseas evidence

Age appears to be an important predictor of both debt use and debt problems, with families headed by younger adults being more likely to use debt, have long-term debt and have difficulty managing debt.

In the United Kingdom, those in their 20s and 30s are more likely to have debt problems than other age groups: almost 40 percent of those who find debt a ‘heavy burden’ are aged between 25 and 34 (Tudela & Young, 2004). This age group is also particularly susceptible to long-term debt, which is consistent with acquiring major assets such as houses (Balmer, Pleasance, Buck, & Walker, 2005).

According to Kempson (2002), age is one of five key factors increasing the risk of arrears[9] in the United Kingdom, the others being family, income, use of consumer credit and priority given to paying bills:

• The relationship of age to debt problems may be a consequence of better access and more liberal attitudes to using credit, as well as higher rates of setting up new homes and having children among younger respondents, both of which are major causes of debt problems (Kempson, 2002).

Less clear, however, is the relationship between age and the amount of debt borrowed. This may simply represent an offsetting income effect – as people age their income and capacity to borrow increase. This is discussed in a later section.

Using United Kingdom survey data, Livingstone and Lunt have shown that the demographic variables of age and number of children are found to be determinants of indebtedness but not of the amount of debt (Livingstone & Lunt, 1992). This is consistent with their earlier work on determinants of saving: demographic variables (including age, sex and number of children) explained 11 percent of the variance in amount of total savings, but almost none of the variance in the amount of regular savings (Lunt & Livingstone, 1991).

This finding in relation to debt is echoed by Del-Río and Young (2005) who have observed that the age of the borrower is the main determinant of the decision to participate in the unsecured debt market (with 20–30-year-olds most likely to borrow unsecured debt) but that age seems to be less important in determining the amount of unsecured borrowing.

Trend data from Australia show that increasing owner-occupier housing debt is being driven by the 55–64 age group, who have lower debt-servicing and debt-asset (gearing) ratios than younger households (Reserve Bank of Australia, 2003). It is not entirely clear, however, whether this increase simply represents an increase in the amount borrowed. While there is evidence that the population holding mortgages has not increased, aggregate data could also be hiding a shift from young to older mortgagees due to the rising costs of buying a house.

There does not appear to be any longitudinal analysis of cohort effects in the overseas literature.

New Zealand evidence

SoFIE wave 2 data, illustrated in Figures 9 and 10, demonstrate that a life-cycle relationship does exist between age and total debt: on average, New Zealanders become slightly more reliant on debt as they move through their 20s. This plateaus though their 30s, 40s and early 50s, then falls noticeably from their late 50s into retirement. This relationship exists for both single and couple families, although a greater proportion of couples have debt than singles.

This relationship, however, is most apparent with mortgage debt and bank and credit card debt. Student loan debt and to a lesser extent ‘other’ debt, on the other hand, exhibit a negative relationship with age. This is to be expected in the case of student loan debt, as most students are likely to be young. The relationship with ‘other’ debt, however, may reflect greater reliance on non-mainstream (and unsecured) forms of credit for young people who have less income and asset security, especially those in a couple family.

Figure 9: Proportion of single families with different types of debt by age


Source: Statistics New Zealand SoFIE data, wave 2 2003/04

Figure 10: Proportion of couple families with different types of debt by age


Source: Statistics New Zealand SoFIE data, wave 2 2003/04

Preliminary examination of HES trend data, illustrated in Figure 11 below, suggests that older age groups are becoming more indebted. The following figure shows mortgage repayments (including interest and principal) against age of the household ‘reference person’ for those who reported such expenditure. Three points stand out: the broadly inverted U-shape is consistent with the life-cycle model; there is a general upward drift over time, consistent with an increase in household mortgage debt (this may include non-housing debt secured against property); and there appears to be an increase amongst older age groups.

Figure 11: Average annual mortgage repayments by age


Source: Statistics New Zealand Household Economic Survey data

As with Australian data, it is difficult to tell how much of this movement is due to interest rates, size of mortgages and number of mortgages. Increased interest rates explain some of the increase. House prices have risen (partly because of the higher purchasing power created by lower interest rates at the beginning of the period), which will also explain some of the increase (since more money is borrowed). While Australian and New Zealand data suggest the proportion of the population with mortgages has not increased, this does not rule out the possibility of movement between population groups or families with mortgages (such as a shift from young to old).

Longitudinal data are required for us to properly assess this and other cohort effects. Subsequent waves of SoFIE will provide some of this data.

Relationships, children and transitions

Family-formation is one of the key life stages captured by the life-cycle model. People have traditionally partnered and had children at the start of their working life, and this helps to explain the relatively high ratio of borrowing to saving at this life stage: incomes are low and costs are high.

Raising a child is an expensive exercise which puts real pressure on the family budget – although exactly how much is the subject of extensive literature on the costs of children with diverse definitions and methodologies (Poland & Seth-Purdie, 2005). All families are different and experience different pressures at different stages.

Increasingly, however, the ‘average’ family is forming later, having fewer children and is more likely to re-form or be a blended family (Statistics New Zealand, 2005).

While these factors are likely to have implications for families’ indebtedness, the correlation is arguably less transparent than before (in other words, we may not be able to observe a straightforward life-cycle relationship). The financial pressures of family-formation may not be as great when people have children later or have fewer children. Family-formation appears to be a more active decision or lifestyle choice; people may prefer to wait until they can afford to marry and have children.

Taking this approach, we would expect that simply having children would not be a strong indicator of indebtedness or over-indebtedness, but having children combined with being young and having a lower income or fewer assets could be a strong indicator of indebtedness or over-indebtedness.

We would also expect that family transitions, such as relationship breakdowns, would be positively correlated with indebtedness because of the impact of splitting finances and the greater costs of living separately.

Overseas evidence

There is mixed evidence as to the effect family size has on use of debt or indebtedness.

Livingstone and Lunt (1992) found evidence of a negative but insignificant relationship between use of debt and number of children.

• Contrary to the sociological literature (eg Hartropp et al 1987), those in debt did not have more children but in fact had fewer children than those not in debt (the number of dependent children did not discriminate the two groups significantly). Thus, greater family demands do not result more often in being in debt. Possibly those with more children are forced to adopt more conservative and fixed budgeting strategies because the economic demands on them are salient and constant, and so they more deliberately avoid debt, resisting the view of debt as part of modern budgeting strategy. In relation to the attribution items, those not in debt more often emphasised the pressure created by children’s demands for goods, while those in debt tended to emphasise either internal factors concerning loss of control or external and general factors connected with the credit system.

The mixed evidence may be due to confounding factors such as age, income and wealth. Large families are likely to be ‘older’ and have the means to support more children. It follows, then, that the relationship with indebtedness may vary depending on the representativeness of survey data of the age of the family and family incomes.

A positive correlation between family size and income may help explain Lindqvist’s finding that debt repayments were positively associated with family size and with owning one’s own home (Lindqvist, 1981). Debt repayments reflect what people pay back, rather than what they owe or whether they borrow in the first place.

In terms of over-indebtedness (or problem debt), the relationship is clearer, but is also affected by the relationship between family size and income.

Kempson et al (2004) found that, in the UK, larger families are more likely to have arrears, be out of work and receive social security benefits. They also found that a higher proportion of larger families than smaller families experience hardship, and comment that “it is perhaps unsurprising that larger families appear more likely to be in arrears”.

However, once income is adjusted for family size, Kempson et al found the link between the number of children and being in arrears (a measure of over-indebtednesss) is much weaker. They also note the complexity of interactions between age, family, income, use of consumer credit and priority given to paying bills. For example:

• Older people and couples without children had a low propensity for arrears, even if their income was low.
• Young people and couples with children were seldom in arrears if their income was high.
• Those at greatest risk were young people on low incomes and low-income families.
• The more children in a low-income household, the greater the risk of arrears.
• The rate of arrears was strongly related to the number of predisposing factors reported by a household, with a big jump in the level of risk among those with four and five predisposing factors.

Berthoud (1989, in Valins, 2004) has also found income to be a confounding variable when considering the impact of family structure on indebtedness: over-indebtedness tends to affect families which have both low incomes and children; among families without children, low income does not seem to make much difference. Having children and low income is a better predictor of problem debt than low income alone (Valins, 2004).

Relationship breakdown has also been found to be positively related to over-indebtedness. According to Balmer et al (2005), relationship breakdown (and other key variables such as ill health) is a significant predictor of debt problems. Kempson et al (2004) have also found that domestic violence and relationship breakdown problems more often occurred before debt problems, indicating the severe change in circumstances that can follow family breakdown. In a recent United Kingdom study, experience of domestic violence, personal injury, clinical negligence and relationship breakdown significantly increased the likelihood of debt problems (Balmer et al, 2005).

In the United Kingdom, a link between lone parenthood and debt has also been observed, with up to one in three single parents falling into arrears. Relationship breakdown or marital separation is considered the primary cause of these problems (Edwards, 2003, in Balmer et al, 2005). Single parents, followed by couples with children, had the highest rates of debt problems (Edwards, 2003, in Balmer et al, 2005).

New Zealand evidence

The Living Standards Report (Ministry of Social Development (MSD), 2006) found that families with dependent children have lower living standards than the overall population because more of these families are reliant on income-tested benefits. Families with market incomes have living standards that are similar to the overall population.

The research also found that:

• Sole-parent families have substantially lower living standards than two-parent families. This is largely because the majority of sole-parent families are reliant on benefits.
• Families with three or more children have lower living standards than families with one or two children.
• Living standards were lower among families with high numbers of doctor visits for child illnesses, and also among families that were restricted in their social and economic participation because of a child’s serious health condition.
• Living standards were lower among families where a parent had had a marriage break-up.

The role that indebtedness plays in these disparities has not been fully explored with LSS data to date, although good data on debt and financial strain have been captured. This will be explored as part of the multivariate analysis of the LSS dataset planned by the Families and Retirement Commissions for 2008/09.

In another New Zealand study using Summary Instalment Order[10] (SIO) data, subjects with more than three dependants at the time of application for an SIO had an estimated four-fold increase in bankruptcy risk compared to subjects without dependants in the first several months after application. Other risk factors for the same time interval included the size of the SIO instalment (Allen & Rose, 2004).

According to SoFIE wave 2 data, illustrated in Figure 12, the proportion of single and couple families with debt is higher if those families have children. The difference between having one and having two or more children, however, appears insignificant.

Couple families generally have a greater proportion of secured debt than single families. However, having children does not appear to disproportionately increase reliance on unsecured debt.

Figure 12: Parenting status of single and couple families with debt (proportions)


Source: Statistics New Zealand SoFIE data, wave 2 2003/04

Multivariate analysis of SoFIE data would allow us to examine whether age, income and wealth have confounding effects. Future SoFIE data could also shed more light on the effects of having fewer children and having them later, and of family transitions.

Wealth and home ownership

For most New Zealand families, borrowing – typically for home ownership – is an important mechanism for accumulating assets or building wealth. Debt secured by assets (even those not yet fully owned) is generally less risky than debt secured by disposable income (discussed in the next section).

Net worth measures the difference between a person’s or family’s assets and liabilities. Prudent financial management, as captured by the life-cycle model, suggests that net worth should always be greater than zero, should increase over one’s working life and should decrease over one’s retirement.

This relationship is also captured by the debt-asset ratio, or gearing ratio. This ratio compares a stock with a stock, so provides a reasonable measure of affordability. However, it can be difficult to accurately measure and may be slightly misleading. A ratio approaching 1 suggests you owe almost as much as you own: you have virtually no net worth. A ratio approaching 0 suggests you owe very little and have some net worth, but does not tell what that net worth actually is – a family that owes $5,000 but owns assets worth $50,000 appears equivalent to a family that owes $100,000 but owns assets worth $1,000,000.

Overseas evidence

In the United Kingdom, tenants rather than homeowners are more likely to experience debt problems and are approximately five times more likely to fall behind with rent payments than homeowners are to fall behind with mortgage payments (Balmer et al, 2005; Department for Work and Pensions (DWP) and Department for Trade and Industry (DTI) 2004; Valins, 2004). This may, of course, simply be capturing an age effect or the fact that renters have difficulty saving.

New Zealand evidence

According to aggregate Reserve Bank of New Zealand data, the ratio of the household sector’s financial liabilities to financial assets doubled from 40 percent in 1991 to 80 percent in 2006 (Reserve Bank of New Zealand, 2006). The ratio of financial liabilities and student loans to financial assets and housing, however, increased only marginally over the same period, from 15 percent to 20 percent. This is reasonably consistent with Statistics New Zealand’s analysis of SoFIE wave 2 data (2003/04), which produced a debt-asset ratio of 16.4 percent (Cheung, 2007). It will be possible to observe change once wave 4 of SoFIE (2005/06) is available.

According to Statistics New Zealand’s analysis of wave 2 of SoFIE, the debt-asset ratio increases between the 15–19 and 20–25 age groups, then declines rapidly between 25 and 35 and steadily thereafter over a person’s working life to be almost zero by retirement (Cheung, 2007). This is consistent with life-cycle theory: net worth increases as people age.

According to our analysis of wave 2 of SOFIE,[11] illustrated in Figure 13, the median debt-asset ratio starts off relatively high for the 18–24 age group, whom we would expect to have little in the way of asset security (the single families column is probably reasonably representative of this age group). The ratio then declines steadily as people age, almost disappearing by retirement.

Figure 13: Single and couple families’ median debt-asset ratios by age


Source: Statistics New Zealand SoFIE data, wave 2 2003/04

Tenure provides us with some insight into the type of assets families use to secure their debt. Tenure is particularly interesting in New Zealand because New Zealanders have a high propensity to accumulate wealth by buying a house. This is therefore likely to be the primary form of security for most New Zealanders. In wave 2, 33 percent of single families and 66 percent of couples ‘owned’ the home they lived in (with or without debt).

This is apparent with the scale of mortgage debt compared to non-mortgage debt, illustrated in Figure 14. The median amount of both mortgage and non-mortgage debt held by single families is also noticeably lower than that held by couple families, as would be expected given the age differences between these two groups.

Figure 14: Median amount of debt held by renters and owners


Source: Statistics New Zealand SoFIE data, wave 2 2003/04

Less expected is the comparability of mortgage debt between single renters and owners and between couple renters and owners. This suggests that some renters have housing assets, which are not represented by the house they live in. This may capture the anecdotal evidence of young singles and couples buying houses, but affording them by renting them out while living with parents or renting cheaper accommodation themselves. The proportion of these families is small, however. Less than 10 percent of renters are in this situation, compared with around 50 percent of home owners.

As expected, however, there is a significant difference between renters and owners in terms of non-mortgage debt. Single renters, in particular, own very little or have very little net worth. This almost certainly captures an age effect, as most renters are likely to be single and young.

Income, education and employment

The relationship between income and borrowing is not entirely straightforward. In theory, as your income increases you can afford to service more debt. However, incomes tend to increase with age, and the life-cycle model suggests that people’s propensity or need to borrow declines as they age and accumulate wealth. So which effect is greater? Is income positively related to indebtedness, but negatively to over-indebtedness?

Education and employment are also relevant to this question. Higher education is likely to mean more stable employment and higher income, as well as better financial literacy.

Financial literacy is an indicator of how well people understand, amongst other things, the terms and conditions surrounding debt. There is some qualitative evidence that poor money management is a significant component of over-indebtedness (Valins, 2004), and other research shows that consumers who are financially knowledgeable are more likely to behave in financially responsible ways (Perry & Morris, 2005).

The debt-income ratio is typically reported as a measure of debt affordability. However, this measure has two significant shortcomings:

1. It compares a stock with a flow: the ‘debt servicing costs’-income ratio is a much more accurate measure of affordability.
2. The rule of thumb that one should not spend more than 30 percent of their disposable income on debt ignores some important relativities – surviving on 70 percent of a low income is not the same thing as surviving on 70 percent of a high income!

In terms of vulnerability, those who have unsecured debt and spend a considerable proportion of their income servicing debt are at more risk than those with assets or those who have a reasonable income buffer.

Those with higher incomes also have greater choice about the amount and cost of debt. The cost of borrowing is much higher for those with fewer means:

• Credit markets are segmented in nature, structured in particular ways so that income bears a close relationship to available credit options. The structure of these markets means that low income people often have to rely on second tier financial services, forcing them into high interest debt (Williams & O’Brien, 2003, p 17).

Del-Río and Young (2005) also note that secured debt is typically cheaper than unsecured debt and will therefore be used in preference by those with access to both types of debt.

Overseas evidence

Disposable income seems to be irrelevant to whether one gets into debt (presumably once a certain minimum income is obtained), but it is a moderate predictor of how far one gets into debt and an important predictor of how much one repays. Repayments are also predicted by the amount owed: the more one owes, the more one repays, provided one has the resources to do so (Livingstone & Lunt, 1992).

In the United Kingdom, low income has been found to be a reasonable predictor of debt problems (Webley & Nyhus, 2001, in Balmer et al, 2005). Arrears also tend to be higher for those on low incomes than for the extremely poor (Valins, 2004).

Del-Río and Young (2005) assess the key factors determining participation in and the amount borrowed from the unsecured debt market. They find that positive expectations of the individual’s future financial position are associated with a higher probability of participation in the unsecured debt market. Higher educational qualifications were also found to have the same association, suggesting that better qualifications make individuals more optimistic and more confident about their future income levels. Individuals with no educational qualifications were found to have a probability of debt that was 10 percentage points lower than that of qualified people. They also found that, for debt holders, the higher the educational qualification, the larger the amount of unsecured debt held. Borrowing for education, however, could likely have a significant influence on these findings. It is not clear whether this form of borrowing is included as unsecured debt, or whether income has been held constant in this study.

In the United Kingdom, those not in employment are more vulnerable to debt and twice as likely to be in arrears as those who are employed (DWP & DTI, 2004). More than a quarter of UK Citizens Advice Bureau clients also reported job loss as a major factor contributing to their debt problem (Edwards, 2003, in Balmer et al, 2005; Kempson, 2002).

According to Balmer et al (2005), being in receipt of benefits or suffering a long-term illness or disability is considered the strongest predictor of debt problems.

New Zealand evidence

Living on a low income for a long period was found to be a major cause of indebtedness in some recent New Zealand case studies, and increasing income is considered the only way out (Williams & O’Brien, 2003).

According to SoFIE wave 2 data, illustrated in Figure 15,[12] median mortgage servicing ratios clearly decline over the life cycle and are generally higher for families with lower incomes. In other words, families with lower incomes allocate more of their income to servicing a mortgage, but all families allocate less of their income to mortgage repayments as they age. Of course, incomes tend to rise with age, so even if the proportion of income is decreasing, the actual amount repaid may still be increasing.

Figure 15: Median mortgage servicing costs to income ratios for single and couple families with mortgages by age and income quintile

Source: Statistics New Zealand SoFIE data, wave 2 2003/04

Ethnicity

Financial decision-making appears to be influenced by many personal factors. Since ethnic groups can have shared attitudes and beliefs, it is possible that ethnicity is one of the influencing factors.

New Zealand evidence

There is some evidence that ethnicity and culture influence debt behaviour in New Zealand. According to Williams and O’Brien (2003) there are cultural pressures for Pacific people to borrow money to support extended family and churches. The Ministry of Consumer Affairs, however, found that while this is a factor for Pacific people, this is not the primary reason they borrow (Ministry of Consumer Affairs, 2006).[13]

Research for the Centre for Housing Research Aotearoa New Zealand (CHRANZ) and Ministry of Pacific Island Affairs (MPIA) asked Pacific families about cultural barriers to fulfilling their housing aspirations:

• Pacific families are expected to contribute to family events, community initiatives and to the church both in New Zealand and the Pacific nations. Many agreed that financial obligations to family and community limited their ability to save. While most accepted this as an integral part of Pacific life, some would like to see the practice modified so families are able to give priority to their housing needs (Koloto, Duncan, de Raad, Wang, & Gray, 2007, p 61).

According to wave 2 SoFIE data, the population of single adults is made up of 75 percent Europeans, 13 percent Māori, four percent Pacific, six percent Asian and two percent other. The population of couples is made up of 75 percent both Europeans, 13 percent both Māori or Pacific, six percent both other, nine percent Māori-European and three percent other mixed. The proportion of each population group with debt and the median amount of debt for each population group is broadly comparable. Māori and Pacific single and couple families have slightly lower median amounts of unsecured debt than other families.

The debt-asset ratios for each population group, however, vary considerably, as illustrated in Figure 16.[14] For single families, while mortgage debt-asset ratios are broadly comparable (all less than 50 percent), single Māori, Pacific and other families have high non-mortgage debt-asset ratios. The median single Pacific family, in particular, appears to hold twice as much non-mortgage debt as assets. This suggests they have very low or negative net worth, which is supported by previous analysis of SoFIE data and HSS data. These analyses, however, suggest that age is a significant confounding effect. For couple families, all ethnic groups carry a greater proportion of secured debt to assets than unsecured debt. Couple families reporting as ‘both Māori/Pacific’ and ‘other ethnicities’ carry more secured and unsecured debt relative to their assets than other ethnicities.

Figure 16: Median debt-asset ratios for single and couple families by ethnicity

Source: Statistics New Zealand SoFIE data, wave 2 2003/04

Economic and social climate and policy

People tend to save more in periods of economic hardship (possibly to anticipate a period of unemployment) and less in a period of prosperity. Katona (1975) links both saving and borrowing in the form of instalment buying to his concept of consumer sentiment – that is, how people feel about the economy and personal economic decisions, which is itself linked to the current economic performance of society. If times are bad, incomes decrease, people feel pessimistic and so they are motivated to build up their resources and to purchase durable goods before prices rise. Thus, people are more inclined to save, to buy on instalments or with credit and to make unusual cash expenditures. When the economy improves, such behaviours decrease as incomes increase and confidence and optimism rise (Livingstone & Lunt, 1992).

This may help explain why observed net lifetime savings may not always tend to zero (as the basic life-cycle model predicts) or why aggregate savings are positive when there is no real population or income growth. In other words, it may explain why people might save more or less than is necessary for their retirement.

The existence of a cohort savings effect, mentioned earlier, suggests that in addition to the age effect discussed above, a range of social and economic factors can influence saving rates (and what people save for, when they save and how much they save). Influences include the state of the labour market, welfare and tax policies and changing family structure. Scobie and Gibson (2003) go so far as to conclude: “social welfare policies do seem to matter to the amount people are prepared to save” (p 26). This conclusion is tempered with recognition that individuals themselves (or cohorts) can in fact influence those social policies (Scobie & Gibson, 2003; Thomson, 1996).

Compared to 30 years ago, New Zealanders are now wealthier, have increased access to credit and have more secure government services. New Zealand has experienced a sustained period of economic growth, witnessed considerable deregulation of the credit industry and seen the advent of the Superfund and KiwiSaver.

Summary

Part 2 focused on various characteristics and circumstances, as well as environmental factors, that are likely to influence and therefore help us explain differences in families’ decisions about savings and debt, and in outcomes.

Overseas and New Zealand cross-sectional evidence based on point-in-time cross-sectional data suggests that there is a life-cycle relationship between age and indebtedness in terms of mortgages, credit cards and bank loans.

This relationship only appears with respect to debt usage or participation in the debt market, however. The relationship with the amount of debt borrowed is weak or ambiguous. This is likely to be because of the offsetting effect income has on the amount people are able to borrow as they age.

Trend analysis of recent Australian and New Zealand data, however, raises some questions about both of these relationships in terms of mortgage debt, but further data and analysis are required. Mortgage debt of older people has noticeably increased in the past decade. It is not clear whether this is because they are borrowing more, or because more older people are entering (or re-entering) the mortgage market.

Age seems to be correlated with problem debt. Younger people are more likely to get into difficulty when using debt. There may also be an income effect that is not being accounted for here.

The evidence is inconclusive as to the impact family size or number of children has on indebtedness, but this is likely to be due to such confounding factors as age, income and wealth: large families are more likely to be older and have the means to support more children. There is more evidence of a positive relationship between family size and over-indebtedness (or problem debt). This relationship is weakened, however, once income is taken into account. There is also evidence of a positive, causal relationship between relationship breakdown and over-indebtedness.

According to international evidence, income is irrelevant in determining who gets into debt, but a reasonable indicator of the amount of debt taken on (which SoFIE data appear to support) and a strong indicator of problem debt.

Having an optimistic view of one’s future financial position and having higher qualifications are associated with increased use of unsecured debt, although income level and borrowing for education may have confounding effects.

Unemployment, low (rather than extremely low) income, benefit receipt and long-term illness or disability are all positively associated with over-indebtedness.

There is little evidence linking ethnicity with indebtedness or over-indebtedness. This relationship, particularly with Pacific cultural factors, would be worth exploring further if the effects of significant confounding factors such as age and income can be held constant.

In 2008/09 the Families and Retirement Commissions are planning to undertake multivariate analysis of the LSS dataset to examine the interrelationships between circumstances and indebtedness (and over-indebtedness).

Footnotes

[6]

See Definitions. [Return to reference]

[7]

See Definitions. [Return to reference]

[8]

There is some evidence that the economic theory of “Ricardian equivalence” holds; namely, that individuals exactly offset government spending and vice versa. (Some caution with the data is required, however. It is also possible that an apparent increase in debt is in fact an increased use of trusts: people are still saving to bequest, but this is not being captured by data.) One implication is that people will not save if they know the Government will provide for them in retirement – behaviour that is consistent with the life-cycle theory. A further implication of this could be that people may even borrow against this source of future income. For example, from 1958 until the 1980s, family benefits could be capitalised and paid in advance to help parents own their own homes, make alterations to existing homes to accommodate larger families and, under certain conditions, to pay off or reduce mortgages. [Return to reference]

[9]

Being in arrears is not only a form of debt, but is also a signal that families may be having difficulty managing their finances. [Return to reference]

[10]

A consumer debt repayment plan administered by a court as a possible alterrnative to bankruptcy. The SIO allows the debtor to repay their debts in regular instalments without the threat of further legal action while the order is in force. [Return to reference]

[11]

Note that these data were generated by the Treasury on 30 May 2008, removing 17 debt-servicing ‘outliers’ who had debt-servicing to income ratios greater than five. This dataset was not used to reproduce other tables in this report because it did not appear to affect proportions. [Return to reference]

[12]

Note that these data were also generated by the Treasury on 30 May 2008, removing 17 debt-servicing ‘outliers’ who had debt-servicing to income ratios greater than five. [Return to reference]

[13]

The Ministry of Consumer Affairs reports that ‘at-risk’ borrowers primarily borrow for essential items, followed by large items and social or cultural obligations, and that reciprocity is central to Pacific cultures (p 105). [Return to reference]

[14]

Note that these figures have not been updated with 30 May data removing outliers. Some of the higher ratios are therefore likely to be exaggerated. [Return to reference]