Household wealth in Australia: its components, distribution and correlates.

Wealth and its distribution are important in understanding modern societies. Wealth is an obvious indicator of position in the social structure, and almost certainly a superior indicator to income-based measures (Podder and Kakwani, 1976), but is not usually incorporated in existing concepts or measures of social location (Adair, 2001; Sorensen, 2000). It is almost certainly closely associated with financial security, poverty and consumption behaviour, and is an important dimension of social inequality. Indeed, research has consistently demonstrated that the distribution of wealth is far more unequal than the distributions of earnings or income Ouster and Smith, 1997; Keister, 2003b; Rodriguez et al., 2002). Furthermore, there are substantial inequalities in wealth that relate to gender, race and ethnicity (Keister, 2000; Straight, 2001; Warren and Britton, 2003; Warren et al., 2001). These inequalities are likely to have consequences for a range of social outcomes including well-being, marital formation and dissolution, retirement and health (Dahl et al., 2003; Huie et al., 2003; Mullis, 1992; White and Rogers, 2000). Finally, wealth may be important in the reproduction of social inequality; it may contribute substantially to children's educational attainment (Conley, 2001; Orr, 2003) or be used directly to provide employment and business opportunities for the next generation (Robinson, 1984).

Data on the wealth of Australian households, however, are limited, and as a result, relatively little research had been conducted on wealth in Australia. While aggregate statistics on wealth holdings have been compiled for some time and are now regularly reported as part of the National Accounts, recent estimates of the distribution of wealth across households have generally been imputed from survey data on income flows. Wealth data collected directly from households are rare. In 2002, however, the Department of Family and Community Services, in association with the Reserve Bank of Australia, funded the inclusion of a wealth module in Wave 2 of the Household, Income and Labour Dynamics in Australia (HILDA) Survey.

This article uses these data to describe the components, distribution and correlates of wealth in Australia. More specifically, this article presents information on: the assets, debts and net worth (wealth) of Australian households; the contribution of housing and other property, business and farm assets, investments, bank accounts, vehicles and other assets to total assets; and the proportions of debt attributable to housing, businesses and farms, the Higher Education Contribution Scheme (HECS), credit cards and other forms of debt. This article also provides estimates of the amount of realizable wealth of Australian households; that is, wealth not held in the form of home equity or superannuation. Finally, and like previous Australian studies of household wealth, this article presents information on the correlations between household wealth and a range of demographic and socio-economic characteristics. The analysis presented here, however, both considers a wider range of correlates of wealth and estimates the size of these relationships with wealth after taking into account its associations with other correlates.

Previous research

Many readers might be surprised to learn that the first serious attempt at collecting wealth data from Australian households occurred as long ago as 1915. Moreover, this collection was based on data collected not from a population sample but from a census--the 1915 War Census conducted by the Commonwealth Bureau of Census and Statistics (see Soltow, 1972). Since that time there has been only one other significant attempt at collecting wealth data directly from Australian households--a university-based survey conducted over the period 1966-8 (see Podder and Kakwani, 1976). The relatively small sample size--just 2757 households--combined with high rates of non-response, however, has meant that relatively few researchers have attached much weight to the results from this survey. (1)

Since that time we are unaware of any other survey-based study that has, with the exception of housing, attempted directly to measure the wealth holdings of individual Australian households. Instead, the dominant approach has been to estimate the size of different wealth components based on the size of the income flows generated by different assets. (2) Dilnot (1990), for example, used information from the 1986 Income Distribution Survey (IDS) conducted by the Australian Bureau of Statistics (ABS) to derive estimates for households of investment income. Combined with estimates of net housing wealth collected directly from households in the IDS, he was then able to derive an estimate of the distribution of household wealth. This approach has been extended by researchers at the National Centre for Social and Economic Modelling (NATSEM) who have provided more recent estimates of the distribution of wealth using both a more complete list of assets (notably the addition of superannuation) and more sophisticated methods for deriving the rate of return to different assets (see Baekgaard, 1998a, 1998b; Kelly, 2001). Most recently, the ABS, using a variety of data sources, has developed household-level estimates for wealth for the period 1994-2000 which seemingly cover all forms of assets and debts held by households, rather than just a subset (ABS, 2002; Northwood et al., 2002). There are thus two groups of recent studies on the distribution of wealth in Australia; the NATSEM studies covering data collected in 1986, 1993 and 1998, and the ABS 1994-2000 series. Estimates within each series are comparable, but there are significant definitional and measurement differences between the two groups of studies.

The main conclusions from these previous studies are threefold. First, the major component of household wealth is housing, with previous research indicating that approximately 40 to 55 percent of either wealth or assets is held in the form of property. The significance of housing in total wealth, however, does appear to be declining. Kelly (2001), for example, estimated that owner-occupied housing constituted 43 percent of wealth in 1998, compared to 49 percent in 1986. The ABS studies, on the other hand, suggest a smaller decline. Dwellings--both owner-occupied and rental investments--were estimated to account for 48 percent of total assets in 1994 and 46 percent in 2000 (Northwood et al., 2002: 27). Other major components of wealth are superannuation, businesses and farms, bank deposits, and shares and other investments. According to Kelly (2001), 22 percent of total household wealth in 1998 was held in superannuation funds, 12 percent in business and farms, 9 percent in interest-bearing deposits, 8 percent in shares and other investments, and 6 percent in rental properties. The most significant change since 1986 was the increase in the proportion of wealth held in superannuation, which increased from 14 to 22 percent.

Second, the distribution of wealth in Australia is very unequal. The richest 5 percent own about 30 percent of wealth (Baekgaard, 1998b; Kelly, 2001), while the top decile's share is around 45 percent (Baekgaard, 1998b; Kelly, 2001; Northwood et al., 2002: 39). In contrast, the bottom three deciles have no wealth at all. Indeed, the bottom decile often shows negative wealth--that is their debts exceed their assets (Baekgaard, 1998a; Northwood et al., 2002:122). The Gini coefficient (3)--a standard measure of income inequality--is around 0.64 for wealth (Kelly, 2001: 16) compared to between 0.30 and 0.45 for income. (4)

While inequality may be substantial, both the NATSEM and ABS studies suggest no increase in wealth inequality since the mid-1980s. Further, wealth inequality has probably declined since 1915, when the wealthiest 5 percent of households owned two-thirds of total wealth and the wealthiest 20 percent of households owned 90 percent (Kelly, 2001: 2; Soltow, 1972). The Gini coefficient in 1915, for example, was 0.86, which compares to about 0.64 in 1996 and 1998 (Kelly, 2001: 2; Soltow, 1972). It is important to note, however, that unlike the 1915 data, the more recent estimates come from sample surveys, and in most cases the sample sizes are insufficient to adequately represent the most wealthy (see Juster and Smith, 1997). In other words, the true Gini is likely to be greater than that calculated from most sample surveys.

Third, the level of household wealth is associated with various demographic and household characteristics. Arguably, the strongest relationships found have been with age. This no doubt reflects life-cycle processes wherein households accumulate wealth as individuals and couples enter the housing market and subsequently increase their home equity. Furthermore, the value of superannuation and other investments generally increases with time. Thus net worth has been found to be relatively low among 15- to 24-year-olds, but increases substantially with each successive age cohort before peaking in the 55- to 64-year-old cohort, after which it declines (Baekgaard, 1998a: 28; Northwood et al., 2002: 30). This latter finding reflects the impact of retirement--retirees are both less able to accumulate wealth and more likely to use their assets to fund consumption. Baekgaard (1998a) also presented evidence to indicate that the association between wealth and …