Bulletin – April 2004 The Composition and Distribution of Household Assets and Liabilities: Evidence from the 2002 HILDA Survey
Australian households' balance sheets have changed considerably over the past decade. The macroeconomic consequences of these changes depend in part on their distributional effects, but until now there has been little information about the distribution of assets and debts across households experiencing different circumstances or with different demographic characteristics.
Previously, primary sources provided information for only a subset of assets and liabilities, mainly related to owner-occupied housing. Estimates of the distribution of households' assets and liabilities, such as those by the National Centre for Social and Economic Modelling and by the ABS, were largely derived from income flows, usually combining different sources. By contrast, the 2002 wave of the Household, Income and Labour Dynamics in Australia (HILDA) Survey includes the first comprehensive survey of Australian households' assets, debts and savings behaviour from one source, allowing a disaggregated view of the entire household balance sheet.
The answers given by the 7245 households that responded to the HILDA Survey provide a detailed picture of the composition of households' financial and non-financial assets and liabilities. Because the data come from a single source, the financing of different assets, such as owner-occupied or investor housing, can be compared across different types of households.
This article analyses summary statistics of the assets and liabilities held by households across a number of dimensions, such as age or income groups. The principal summary statistics used are the median and the mean value of asset and debt holdings. The mean describes an average value, while the median is more representative of a ‘typical’ household since it describes the household in the middle of the distribution. For the detailed asset and liability holdings we report the percentage of various groups that have a certain asset or liability, and the median position of those households that have the asset or liability.
The Distribution of Net Wealth
The average household in the HILDA Survey had net wealth of $404,300 in 2002. The average net wealth of households in the wealthiest 20 per cent of the population, or ‘quintile’, was about three times as much as that in the second wealthiest quintile and about 15 times that in the second poorest quintile. Since the distribution of net wealth is skewed, the median, or ‘typical’, household had net wealth below the mean, of $218,500. Around 4 per cent of households reported negative net wealth.
Higher-income households typically also had higher wealth, which is not surprising given that wealthier people have more assets on which they earn income, but is presumably also due to the fact that people with higher incomes have a higher propensity to save. However, the positive relationship between income and net wealth was not exact: for example 56 per cent of households in the highest income quintile were not in the wealthiest quintile, and 7 per cent of households in the lowest income quintile were reported to be in the wealthiest quintile.
The distribution of net wealth across all households was more unequal than that of income – a result that has also been found in other countries. The household at the 80th percentile of the income distribution earned, before tax, about 4½ times more than the household at the 20th percentile. By comparison, the household at the 80th percentile of the wealth distribution was about 20 times wealthier than the household at the 20th percentile. This may sound like a large difference, but it is small compared to that found in the US. According to the Federal Reserve Survey of Consumer Finances in 2001 the equivalent ratio for the US was more than 50. This difference is partly due to the fact that the share of households with zero or negative net wealth is considerably higher in the US than in Australia.
The difference in net wealth across households may partly reflect life-cycle factors. Households towards the end of their working life have accumulated savings for their retirement, while many young families take out debt in order to finance education or the purchase of a home. Consistent with this, households where the reference person was aged between 55 and 64 years were typically richer than those younger or older. Moreover, the median household aged 75 years and older had still more than four times the net wealth of a household aged 35 years and younger. The difference occurs partly because older households had less debt, but is also because their assets were higher, providing resources to draw on during retirement or to bequeath to the next generation.
Across the household sector as a whole, non-financial assets account for the bulk of total assets (Graph 1). Households in the HILDA Survey held 68 per cent of assets as non-financial assets, slightly more than the 63 per cent share suggested by aggregate data. This difference could be due to different treatments of business assets and consumer durables. The share of non-financial assets in total assets peaked at 78 per cent for the middle wealth quintile and was around 65 per cent for both the wealthiest and the poorest wealth quintile. In contrast, debt as a share of assets fell steadily from 82 per cent in the poorest quintile to 9 per cent in the wealthiest quintile.
The data in the HILDA Survey showed that renters had considerably lower net wealth than home owners. This presumably reflects the fact that renters were on average younger and had lower incomes than home owners. Not surprisingly, renters on average held a higher share of their assets in financial assets than did home owners.
Differences in home values also explain geographical differences in net wealth. The typical household in NSW and Victoria had a higher net worth than that in the other large Australian states. This was mainly due to the higher values of homes owned by households that live in these states, with the median households' holdings of debt and financial assets more similar across states.
Almost all households reported that they owned some type of financial asset, with 97 per cent having an account with a positive balance at a bank or other intermediary. However, the median amount in these accounts was lower than for any other financial asset type, at around $5,000. More than half of all financial assets were held in superannuation, with a median value of $35,000 for the 76 per cent of households that held any assets in superannuation.
Superannuation and life insurance accounted for 55 per cent of all households' financial assets – similar to the share implied by aggregate data – with the relative importance slightly lower for only the wealthiest quintile (Graph 2). Deposits constituted 27 per cent of financial assets for the poorest households, but this share fell steadily to 12 per cent for the wealthiest quintile. Their place was taken by equity investments and trusts which accounted for 31 per cent of financial assets for the wealthiest quintile but only 6 per cent for the poorest quintile.
Wealthier households tended to spread their larger asset portfolios over a wider range of assets than less wealthy households. Households in the wealthiest quintile were more likely to own equity investments – 78 per cent of this group compared with 41 per cent for households in general – and on average held a greater share of their assets in this asset type. Nearly 90 per cent of households in this group held assets in superannuation. Although a much smaller fraction of households in this group had investments such as bonds or certificates of deposits, those who did typically invested around $45,000 in these assets.
It might be expected that households with shorter investment horizons prefer less risky assets with more predictable short-run returns. Not surprisingly, therefore, the HILDA Survey showed that older households were more likely than young households to hold low-risk bonds and deposits, and were less likely to hold riskier equity investments.
Not surprisingly, the most valuable non-financial asset for most households was their own home, with a median reported value of $250,000 for the 68 per cent of households in the HILDA Survey that were home owners. This median was almost the same for home owners both with and without debt outstanding on their home. The proportion of home owners and the value of the owner-occupied home increased with income, net wealth and age, except for households 75 years or older which were slightly less likely to own a home and tended to own homes valued below the median level.
On average, property assets accounted for 79 per cent of non-financial assets (Graph 3). This was lower than the share of 94 per cent based on aggregate data compiled by the RBA, largely because the HILDA Survey included business assets, which are not included in these other data for non-financial assets. The primary residence was by far the most valuable asset holding, especially for the households close to the centre of the net wealth distribution. Other residential properties accounted only for around 10 per cent in the middle quintiles, while they accounted for 20 per cent of non-financial assets for the wealthiest quintile. Other residential properties accounted also for around one-fifth of the non-financial assets in the poorest quintile, though in absolute terms the value of assets held by these households was very small. Moreover, only a few households in this group owned other residential properties, most of which were highly geared.
Business assets accounted for 14 per cent of non-financial assets and were relatively concentrated in the wealthiest quintile. Vehicles and collectibles accounted for 39 per cent of non-financial assets in the poorest quintile, and only for 5 per cent in the wealthiest quintile, but not surprisingly the average value of these assets held by the wealthiest was much more than that held by the poorest.
Around 17 per cent of households owned a residential property that was not their primary residence, with a median value of $200,000. As well as investment properties, this also includes holiday homes and second residences, with only 61 per cent of these households reporting that they received rental income. The share of households with investor properties is likely to be higher than the 10.3 per cent of households that both owned a second property and received rental income, since some investment properties might be vacant or rented out to a family member that does not pay rent.
Information from the HILDA Survey on the proportion of households with investment properties lines up reasonably well with data from other sources. Previous estimates had to rely on rental income as an indication of whether a person or household owns an investment property. In the HILDA Survey around 12 per cent of persons with taxable income reported in 2002 that they received rental income, comparable to estimates of around 12½ per cent obtained using 2000/01 ATO tax data. The difference between these figures and the estimate of the proportion of households that both owned a second property and received rental income, as cited above, suggests that a number of households received rental income, but did not own a second property. These could be households that received rental income from lodgers or from non-residential properties, although some responses might simply be reporting errors.
Owner-occupiers were substantially more likely than renters to own other residential property. Even so, 10 per cent of renters owned a residential property, despite the perceived advantages of first investing in home ownership. One reason could be that these renters were not able to afford to buy a property close to their preferred location. The median value of other residential properties owned by renters was $160,000, somewhat lower than those owned by home owners which had a median value of $200,000. Around 70 per cent of properties owned by renters were geared, that is, they had debt outstanding against them, compared with only 42 per cent of other properties owned by home owners. However, looking only at geared properties, the typical gearing ratio for other residential properties was similar whether it was owned by a renter or by a home owner.
Around one-third of all households had no debt. Indeed, for a number of categories of respondents the median household did not have any debt. This could reflect either demand or supply factors. Some categories of households may not want to incur debt, such as older households or retirees, while others may have found it difficult to borrow, such as households with low incomes or with low levels of formal education.
The median debt-to-asset ratio was 7.8 per cent across all households and 23.3 per cent across only those households with any debt. Because these figures are median debt-to-asset ratios of individual households, they are different from ratios calculated from aggregate household asset and debt data. The comparable figure, calculated by aggregating the HILDA asset and debt data and then taking the ratio, at 14½ per cent is close to previous aggregate estimates.
The composition of household debt across net wealth quintiles showed a pattern quite similar to that for non-financial assets (Graph 4). Just as the share of the primary residence in total non-financial assets was highest for the middle net wealth quintiles, so too was the share of home loans in total debt. For the second and third quintiles, home loans accounted for around three-quarters of total debt.
Around 30 per cent of all households had credit card debt, but the median value ($2,000) was only a fraction of residential secured debt. A further 40 per cent of households had credit cards, but had no outstanding balance on them. Most of these regularly paid off their credit card balances in full. In the first wealth quintile, personal debt accounted for 50 per cent of debt. This category includes credit card and HECS debt, but for these households it was dominated by personal loans. The share of personal debt fell to 15 per cent of all debt or less for the other, wealthier quintiles, being replaced by investment property loans and business debt. Business debt – similar to business assets – was almost exclusively concentrated in the wealthiest group.
While the proportion of households with debt was roughly the same across net wealth quintiles, for those with debt, the median amount owed increased with net wealth. However, gearing ratios decreased with net wealth, implying that the higher debt in the groups with higher net wealth was more than outweighed by higher assets.
Debt on property reported in the HILDA Survey included not only mortgages and home equity loans, but also loans obtained from friends or family members to purchase a property. Around 2 per cent of the total value of home loans outstanding were loans from friends or family, around 80 per cent were first mortgages and the remainder were home equity loans and second mortgages. The median gearing ratio for households that owned a property – either owner-occupied or other residential property – with debt outstanding was 39 per cent, comparable to the ratio of 43 per cent implied by aggregate data.
Half of all home owners had a loan outstanding on their primary residence, with a median value of $90,000. Data from the 2001 Census suggest that 40 per cent of home owners had a mortgage outstanding. This is lower than that reported in the HILDA Survey, possibly because respondents to the Census may not have included some housing-related debt, such as home equity loans or loans from friends or relatives.
The proportion of other residential properties that were geared was 47 per cent, similar to the proportion reported for owner-occupied properties. However, for those with debt the typical amount outstanding was higher than for owner-occupied homes. Since the median value of other residential properties was lower than for owner-occupied homes, the typical gearing ratio on other residential properties was also higher than for owner-occupiers, at 56 per cent. One explanation for this difference could be that home owners may choose to pay off their debt faster than owners of investment properties, who can deduct interest expenses from taxable income.
Data in this Appendix are extracted from the HILDA Survey 2002 (unconfidentialised data set), Melbourne Institute of Applied Economic and Social Research, which provides unit record responses from 7,245 households. All net wealth and income data in this article are based on actual responses, and imputed responses if some answers were missing. We also adjusted the weight of answers given by individual households in order to ensure representativeness for all Australian households. Household characteristics such as age, education and labour force status are based on the household reference person, which was chosen from the household members as the person with the highest income, highest salary, labour force status, age and respondent number, in this order.
In this article, we only recorded a household as having assets or debt of a particular type if they had a positive balance in it. Thus we excluded a number of households with bank accounts, credit cards or business assets or debt with zero balances from the group of households possessing these assets or debts.
We have not reported results which were based on less than 0.2 per cent of the sample, and denoted these with ‘–’ in the tables.