Christopher Sheil and Frank Stilwell
For the first time in more than half a century, it is clear that the richest 10% of Australian households now own more than half the nation’s private wealth. Wealth inequality has grown significantly. Over the four years from 2012, the proportion of the nation’s wealth owned by the Top 10% increased from 48.1 to 50.2 per cent. Most of the increase went to the Top 1%, whose share of total wealth rose from 14.2 to 16.2 per cent.
The increase in the wealth-share owned by the Top 10% is about equal to the total existing wealth of the poorest 40% of Australia’s households, whose share has remained virtually stable at 2.8 per cent. In other words, most of the increase in the wealth-share owned by top decile has come at the expense of the proportion owned by a broad middle-class. The wealth-share owned by the three deciles below the Top 10% (i.e. measuring from the top, households from the 90th percentile to the 60th percentile) has fallen from 37.8 to 36.7 per cent, whereas the middle quintile’s share (from the 60th to the 40th percentile) has fallen from 11.3 to 10.4 per cent.
These are the main findings of the Evatt Foundation’s inequality project, which began monitoring the distribution of Australia’s wealth following the election of the Abbott government in 2013. The research is unique in that it utilises OECD and ABS data to arrive at the most robust possible estimates of the wealth of the Top 10%. The estimates should also be regarded as conservative.
Over the four years that reliable data are available to 2016, the pattern is clear. Wealth inequality in Australia is evolving along two fault lines. The bottom 40% of Australian households have practically no share of the rising total. Meanwhile, the middle 50% of households have a declining share relative to the Top 10%, and particularly relative to the Top 1%. The pattern is illustrated in the below ‘Figure: The concentration of wealth in Australia since 2012’.
The Evatt Foundation’s interest in economic inequality has focused on the wealth of the Top 10% of households. Until recently, information on this Top 10% has been a missing statistical link in Australia’s official inequality data.
By international standards, Australia has high quality data on income inequality, including high quality measures of the incomes of the Top 10%. But income is different from wealth. Income is a flow, whether in the form of wages, profits, dividends, rents or interest payments, whereas wealth is a stock, whether held as financial assets or physical assets. By comparison with the distribution of Australia’s incomes, our knowledge of the distribution of the nation’s wealth is crude and sparse. The principal sources are a biennial Survey of Income and Housing (SIH) conducted by the Australian Bureau of Statistics (ABS), and the Australian System of National Accounts (ASNA), which estimates household wealth from many sources as an integral part of its function as a systematic summary of the economy.
This information is invaluable but has shortcomings by contemporary international standards. In particular, the SIH does not publish the wealth-shares of the Top 10%, Top 5% or Top 1%, which have become the centre of attention in the worldwide debate over inequality and are now standard features of the official statistics in many — and an increasing number of — OECD countries. Instead, the SIH presents the distribution in the form of quintiles (i.e. 20% divisions), which are supplemented by inter-decile ratios (the relative distance between percentiles at least a decile apart, such as the difference between the 90th and the 10th percentile — known as the P90/P10 ratio. The ASNA publishes a national household balance sheet supplemented by quintile divisions using the SIH after the results have been ‘equivalised’ (see the below discussion of a report by the Productivity Commission for the meaning of this concept).
The Evatt Foundation’s estimates for the Top 10%, Top 5% and Top 1% are based on the OECD Wealth Database, which was established in 2015. Beginning with 17 countries, including Australia, the Database now covers 28 of the 36 OECD countries. Exemplifying how far Australia’s official statistics have fallen behind world’s best practice, the ABS does not publish a domestic report on these wealth-shares, even though it supplies the OECD with the figures for Australia covering the top decile on an internationally comparable basis from the SIH. In using the OECD Database to arrive at wealth-shares for the Top 10%, Top 5% and Top 1%, our methodology entails three adjustments that distinguish the results from other Australian wealth distributions, as follows.
First, the OECD figures do not include superannuation. The exclusion may be regarded as sound because measuring the value of superannuation and pension wealth is both inherently complex and subject to substantial differences in the institutional arrangements across countries, which undermines international comparability. In principle, the wealth held in superannuation and pension schemes is equal to the current capitalised value of the flow of income that households expect to receive in the future, which is contingent on replacement rates (i.e., the ratio of the benefits to present earnings), the duration over which the benefits are paid (which depends on mortality rates), retirement ages, and indexing for costs and standards of living. In the Australian domestic case, the SIH simply records current superannuation balances, begging the question of whether these should be enhanced by or discounted for the value of public pensions. Furthermore, we are comfortable with the exclusion because, although the wealth accumulated in superannuation accounts is important for estimating the nation’s wealth for macroeconomic purposes, it is generally illiquid from a household perspective, telling us little about the present distribution of the discretionary command over economic resources and all that implies. A related argument could be made for the exclusion of a certain level of owner-occupied housing stock because, being necessarily consumed by the population, it is not available as wealth for discretionary use. However, whereas superannuation and pension values are too complex to include in the distribution, the value of non-discretionary housing stock is too complex (not to mention too unconventional) to exclude. We therefore consider a conservative approach to using the existing data is appropriate.
Second, the figures do not include non-income earning household consumer durables, such as clothing, furniture, appliances, motor vehicles, etc. Durables are included in the SIH but excluded from the ASNA on the basis that they are intrinsically consumption items. This exclusion sharpens the realism of the distribution for a range of reasons. First, the SIH’s value for durables is likely to be exaggerated at the bottom of the range. This is partly because these values are self-reported, and the fewer durables people own the higher they will tend to value them (and vice versa). It is also partly because the SIH’s durable valuation method differs from other assets in being given their insurance value on a ‘new for old’ basis (rather than current market value after depreciation). The values are also exaggerated by virtue of the SIH excluding people who have few durables, such as people living in nursing homes or living on the street. Some idea of the degree to which the SIH exaggerates the value of household durables is given by the comparison with the memorandum item for durables reported by the ASNA, the latter being less than half the SIH’s value ($358bn compared with the SIH’s $846bn). While the absolute effect of this exclusion is to reduce the recorded value of the total wealth owned by households at the bottom of the range, the amount broadly cancels out across the range. The one significant caveat is the exclusion of a value for durables unavoidably excludes works of art and other valuables that are disproportionally used as stores of wealth at the top of the range. One may therefore surmise that the effect of excluding durables on the measured overall distribution of wealth is conservative. It is a treatment consistent with the approach taken to durables by Thomas Piketty and the many scholars associated with the World Inequality Database, as well as with the ASNA.
Thirdly, the most recent wealth distribution statistics are calculated by projecting the OECD’s most recent ratios for the top decile onto the most recent SIH figures for 2016 (after excluding superannuation and household durables). This follows the method of our earlier report for the Evatt Foundation on The Wealth of the Nation (2016), where we argued that the ensuing estimates were likely to be conservative.1 Our argument has since proved true. The OECD has now published two sets of figures for the top decile in Australia for the years 2011-12 and 2013-14. In following the same methodology in projecting the 2011-12 figures onto 2013-14, our earlier wealth-share estimates and the OECD’s subsequently published results are as follows: (1) Top 10%: Evatt estimate 49.1 per cent; OECD actual result 49.6 per cent; (2) Top 5%: Evatt estimate 35.2 per cent; OECD actual 35.7 per cent; (3) Top 1%: Evatt estimate 14.5 per cent; OECD actual 15.0 per cent. Thus, our estimates of the extent of inequality turned out to be consistently a little more conservative than the OECD’s. We expect our latest estimates will also prove to be conservative when a further actual OECD update for 2016 becomes available in 2020-21.
Finally, while our wealth inequality calculations should be regarded as conservative within the terms of the OECD and SIH data, it should also be appreciated that these terms are themselves conservative. This is most vividly revealed by the large and growing discrepancy between the total national wealth reported by the SIH ($7,325bn after excluding durables) and the ASNA ($8,802bn). The discrepancy of $1477bn is the equivalent of the total wealth of the SIH’s second quintile, or more than one and half times the total wealth of the Top 1%. It is apparent that the vast bulk of the discrepancy is in forms of wealth predominantly owned by the top end of the range. Apparently missing from the SIH is $379bn from accounts held with financial institutions, $329bn in residential property assets, $317bn in property and other loans, and $150bn in shares, trusts and other equity. In part, the shortfall can be explained by the notoriously low response to inequality surveys by the rich, particularly in relation to financial assets. In part, it can also be explained by the SIH’s exclusion of people living in nursing homes, for while they own few durables, they are generally wealthier than younger people. Finally, beyond all official statistics lies the wealth hidden in tax havens, which is estimated on a worldwide basis to exceed the total official wealth of Australia’s households. In sum, our methodology is straightforward, robust and conservative. Were it able to be fully calculated, we would not be surprised to find that the real proportion of the nation’s wealth owned by the Top 10% reaches anywhere up to 55 per cent of the total.
Other measures of wealth inequality
As well as the data produced by the SIH, the OECD, the ASNA and the Evatt Foundation, there are two other major sources of information on Australia’s wealth inequality. These are the annual Global Wealth Report published by the Swiss investment bank, Credit Suisse (which is utilised by Oxfam in an annual well-publicised global analysis), and the government funded Household, Income and Labour Dynamics in Australia (HILDA) survey, conducted by the Melbourne Institute at the University of Melbourne. Both are valuable sources of information on inequality, for different reasons.
The data series produced by Credit Suisse is the boldest attempt to reconcile the various statistical sources to compare the distributions of the world’s private wealth. It differs from the OECD and SIH data in that the unit of observation is individual adults (from age 20, rather than households), which is increasingly being regarded as best practice in the international inequality literature. Splitting household wealth equally between married couples markedly lowers the mean and median wealth figures and has variable effects on the level of inequality but neutralises demographic trends (due to the decline of marriage relative to single-headed families), implies gender neutrality, and meanwhile still recognises that children typically own no wealth. The Credit Suisse data also aims to reconcile surveys such as the SIH with the data from national accounts, the ‘rich lists’ produced by business magazines such as Forbes, and income correlations, and it includes private superannuation wealth. The upshot is probably the most realistic of all the wealth estimates, although the exclusion of public superannuation schemes and pension entitlements dilutes the robustness of Credit Suisse’s international comparisons. The method yields a higher estimate of the wealth of Australia’s top decile than the Evatt Foundation: the 2018 Global Wealth Report exceeds our estimates in finding that the Top 10% own 52.7 per cent of the nation’s wealth, the Top 5% own 40.8 per cent, and the Top 1% own 22.4 per cent.
The HILDA survey differs fundamentally from the other estimates of inequality in that it does not aim to examine wealth as such. Rather, the strength of the HILDA survey is that it is a nationally representative longitudinal study of households which looks at a much wider range of aspects of life, not only the standard economic variables like income, expenditure and wealth, but also family relationships, childcare, employment, education, health, attitudes, values, life events, experiences and more. HILDA’s distinguishing methodological feature is that the same panel of households and individuals is interviewed each year, enabling the study of change across many dimensions over time. With respect to wealth specifically, the survey is idiosyncratic in including the value of vehicles (like the SIH) but excluding other household durables (like the ASNA), and in reporting its results by select percentiles. More significantly, the total wealth captured by the HILDA survey is less than the SIH (and even less compared with the ASNA and Credit Suisse). The participants in the ongoing panel do not include Australia’s richest households (the wealthiest of the survey’s initial participants owned assets valued at less than $22 million, a small fraction of the cut-off point to qualify for inclusion in Australia’s top 200 rich list).2 Recent research on income inequality has highlighted how the exclusion of extremely rich households, while tiny in number, makes a significant impact on the measurement of inequality overall.3 While the HILDA results are not directly comparable with the wealth estimates presented in this report, it is worth noting that the survey has also reported increasing inequality at the top of its range in the time between its original 2002 report and the release of its most recent wealth data in 2014.
The Productivity Commission’s wealth estimates
The Productivity Commission has also sought to contribute to the data on wealth inequality in Australia with a 2018 report titled Rising Inequality? A Stocktake of the Evidence. This contribution is problematic for reasons that we have detailed at length elsewhere, and they need only be summarised here.4 Broadly, the Commission’s results were consistent with the other sources in finding that wealth inequality has increased steadily in Australia over the 12 years to 2015-16. More controversially, it also found that the distribution has been ‘fairly stable’ since 2010. In support of this claim, rather than undertaking a stocktake of the existing evidence, the Commission presented novel estimates of the top decile’s wealth based on ‘equivalised’ households. This methodology produced a curve in a diagram tracking wealth inequality over time, purporting to show a ‘fairly constant trend’ overall, but with a fall in inequality in 2015-16, the year we have reported on here. In our detailed analysis, we argue that the Productivity Commission’s claim that the trend has generally been stable before a recent downturn is misleading, whether by error or design, and has only served to deny the scale and severity of the growing wealth inequality in Australia.
It is not possible to be definitive about the reasons for the anomalous finding, as the Commission effectively went out of its way to present data that was incommensurate with other wealth statistics and defy reproducibility. While the Commission published a good deal of data in its report, it was conspicuous in not publishing the raw figures (or the computer codes) on which its Top 10% estimates were based. According to the notes for the sources, the estimates drew on microdata from the SIH, which were then ‘equivalised’. The aim of adjusting inequality statistics for households by using an equivalence scale is to reflect differences in household size and composition, and perhaps also to recognise economies that can arise from households sharing resources. In analysing the inequality of household incomes, there are internationally recognised equivalence scales that are widely applied in the literature and they are particularly relevant for estimating the effects of taxes and transfers. This is not the case for wealth statistics. With respect to household wealth, as the OECD has observed, ‘no internationally agreed equivalence scales exist, and there is no consensus on whether the scales used for income are appropriate for wealth’.5 Some of the reasons are obvious. Unlike the flow of income, which is used to finance household consumption, assets and debts are typically owned (or owed) by named individuals who don’t include children. In principle, we would strongly encourage the construction of both income and wealth data that can be decomposed to produce the best possible estimates for differences due to age, gender, children, people with disabilities, localities, and so on. However, such constructions must be in addition to rather than substitutes for using a homogeneous unit of observation, which is fundamental for reliable comparisons of collections in the same country and obviously for international comparisons. In treating wealth as if it is a flow of income, the Commission has effectively contradicted the definition of wealth, a category error.
The difficulties in the case of the Productivity Commission’s finding are compounded because the microdata that has been equivalised has been drawn from the SIH’s ‘confidentialised unit record files’ (CURF), which the ABS only makes available to approved users. This microdata was independently researched under the auspices of the Evatt Foundation’s inequality project prior to the publication of the Commission’s report and was found to be of questionable accuracy (in their unequivalised form). Primarily this was because the data are incompatible with the full SIH, of which the CURF data are a subset. The obvious explanation is that the microdata for the Top 10% utilises a much smaller sample than the SIH’s Top 20%. It therefore has a higher margin for error in measuring the top end of the range, which has such a famously high margin for error in any event that best practice is to over- rather than under-sample. As set out in our more detailed analysis, our assessment is that the Productivity Commission’s data is so improbable that it can be regarded as virtually impossible.
These reflections on the flawed quality of the Productivity Commission’s data highlight the need for reforming Australia’s inequality statistics to bring them into line with current international practice and emerging directions. This is not a criticism of the Australian Bureau of Statistics, but rather the constraints within which its undoubted professionalism is bound to function. The most glaring shortcomings are in the measurement of the distribution of wealth, where the existing framework has effectively been superceded or at least requires augmentation. Over recent decades, the increase in economic inequality has largely been driven by a startling rise in the income and wealth accruing to the top of the distribution, namely the Top 10%, Top 5%, Top 1%, Top 0.1% (and indeed, the Top 0.01% and 0.001%). These divisions are increasingly being supplemented by measures of the ‘middle class’, which refers to the 40% of households (or adult individuals) that lie between the Top 10% and the bottom 50%. These divisions are now routine in the flourishing scholarly literature on inequality around the world and have been adopted by international agencies such as the OECD, the IMF, the World Bank and the United Nations.
The upshot is that Australia is increasingly unable to speak the common language of the international conversation on inequality. Australians have traditionally prided themselves on the high national value placed on egalitarianism and are given to the thought that economic inequality in this country is not as bad as overseas. Yet in many vital respects, we simply don’t have the statistics to show this to be true. A major aim of the Evatt Foundation’s inequality research is to produce data for the wealth of the top decile that will challenge governments and others to produce better data on this critical statistical link in Australia’s official inequality data.
Bringing Australia’s data up to international standards will not be a simple process. The Wealth of the Nation also identified anomalies or at least difficulties associated with the treatment of superannuation and pensions, consumer durables, valuables, the sampling of rich households, and the integration of the survey results with the national accounts. Ideally, the SIH should be refined and integrated with both the national accounts and tax data, so that it will become possible to track the evolution of income and wealth levels from the bottom to the top in a systematic manner and allow for comparisons to be made both within the country and with other countries. The ultimate objective should be to develop ‘Distributional National Accounts (DINA)’.6 The concept entails capturing 100% of national income and wealth to allow for economic growth rates to be calculated for each division of the distribution in a manner consistent with the calculation of Australia’s macroeconomic growth rate.
When we have estimates for the distribution of both pre-tax and post-tax income and wealth, it will become possible to provide a comprehensive view of how government policies affect inequality. Access to more and better data is critical to monitoring the nation’s inequality dynamics. It is the key to understanding the present and the trends that are likely to dominate in the future, and for designing potential policy responses. DINA accounts will allow for a more informed public discussion on inequality by bringing much more complete data to all sides in the debate, thus helping to ground discussion, deliberations and decisions. At present, the opacity in the inequality of income and wealth undermines rational democratic discussion. Perhaps it needs to be remembered that economic issues don’t belong to economists, statisticians, officials, or business people. They belong to everyone.
This is the text of the 2015 Annual Evatt Lecture, presented by Lesley Hughes at the H. V. Evatt Memorial Dinner, introduced by Trish Doyle, MP, and hosted by the Evatt Foundation in association with the Katoomba branch of the ALP & the Blue Mountains World Heritage Institute at the Carrington Hotel, Katoomba, on 17 October 2015.
Visit the Climate Council website: climatecouncil.org.au.
You can take Professor Hughes’ free short online course on climate change through Open Universities Australia. The course runs over 4 weeks (repeating) and takes about an hour a week (each lecture is only 6 minutes). If you enroll in the last week, you can do the full course at once: open2study.com/courses/climate-change
1. Sheil, C. and F. Stilwell (2016) The wealth of the nation: current data on the distribution of wealth in Australia (Sydney: Evatt Foundation).
2. Headey, B. (2003), ‘Income and Wealth — Facilitating multiple approaches to measurement and permitting different levels of aggregation’, HILDA Project Discussion Paper Series No. 3/03, Australian Government Department of Family and Community Services, p. 10.
3. See Alvaredo, F. (2011),’A note on the relationship between top income shares and the Gini coefficient’, Economic Letters, 110, pp. 274-277.
5. OECD (2013), OECD Guidelines for Micro Statistics on Household Wealth (Paris: OECD Publishing) p. 169.
6. See Alvaredo, F. and A. Atkinson, L. Chancel, T. Piketty, E. Saez, and G. Zucman (2017), ‘Distributional National Accounts (DINA) Guidelines: Concepts and Methods used in WID.world’, WID.world Working Paper Series No 2016/1; Facundo Alvaredo, F. and L. Chancel, T. Piketty, E. Saez, and G. Zucman (2017), ‘Global inequality dynamics’, Working Paper 23119, NBER Working Paper Series (Cambridge, MA).
Sheil, Christopher and Stilwell, Frank, 'Wealth inequality in Australia: 2012 to the present', Evatt Journal, Vol.18, No.1, May 2019.<https://evatt.org.au/wealth-inequality-in-australia-2012-to-the-present>