The public policy think tank, Per Capita, has just released a new multidimensional measure of inequality, the Australian Inequality Index. The index combines various measures of inequality in seven areas: income, wealth, gender, ethnicity, disability, and intergenerational and First Nations inequalities. The measures are weighted and give a measure of inequality for each category, and for overall inequality.

Given my ongoing critique of the use of mono-dimensional and often misleading household income statistics as the primary measure of inequality, I welcome Per Capita’s initiative. In previous posts I have tried to develop consistent indicators of inequality in a number of similar areas, mainly in terms of shares of total income. However, the Per Capita inequality index is broader than simply income (or economics) and includes a range of social measures – something that makes its methodology bolder, and more fraught (more below).

Results

Because the Inequality Index combines different types of measures (for instance, gender inequality includes political representation ratios, crime victimisation rates, and the gender wage gap), it is difficult to get a common language and measure. Per Capita solves this partly by using indexes with values between 0 (perfect equality) and 100 (the furthest distance from equality). So, for instance, if wealth inequality was rates at 70, this would mean that the wealth of the highest income group would need to decrease by 70% for equality to be achieved. These index numbers are then tracked in each area (and combined) for the years since 2010.

The outcome is summed up in the graph below from the Summary Report. It shows stability in the immediate years after the GFC, followed by a bumpy decrease in inequality from 2013 to 2018, primarily due to improvements in gender and ethnic equality, and in some measures of equality for First Nations’ people. However, the first two of these indexes turned around late in the decade and, coupled with rises in income and wealth inequality, the index shows a resurgence in inequality.

Interestingly, alongside these index numbers, there is also an estimate of the time that it would take to reach equality – at the current rate of progress, and with a 1% per annum rate of catch-up. I have created the table below from the Index to sum up the key findings in each area of inequality.

Measure	2021 Index Number	Change since 2010: Index Points	Years to Equality based on trend over last 10 yrs	Years to Equality at 1% p.a. Catch Up
Income	45.4	-4.2	86 yrs	45 yrs
Wealth	64.4	7.7	Never	85 yrs
Gender	21.0	-9.8	18 yrs	19 yrs
Intergenerational	28.5	1.2	Never	29 yrs
Ethnicity	45.9	-4.1	37 yrs	45 yrs
Disability	63.4	16.8	Never	63 yrs
First Nations	36.5	-6.7	51 yrs	63 yrs
Overall Inequality	43.6	0.1	361 yrs	40 yrs

Statistical Issues in the Inequality Index

Given the Inequality Index was put together by a relatively small think tank rather than a government statistical agency, I think there are some minor anomalies in a few places, but this should not distract from the usefulness or ambition of the project. However, there is still devil in the detail.

The full methodology and assumptions have not yet been published, so what follows is based on the information in the Summary Report.

The bringing together of multiple dimensions of inequality requires weighting the relative importance of the various components – otherwise a few areas of harsh inequality potentially impacting relatively few people may overwhelm the index. Yet weighting is tricky: how do you weigh the relative importance of women’s political representation, with the gender wage gap – let alone those issues with rates of Aboriginal incarceration or numbers of people with disability reporting discrimination?

The process is inherently subjective, but all statistics are subjective in that their definitions reflect subjective or theoretical assumptions. The bigger question is whether the weighting and the subsequent index is statistically robust. That is, would the index or the trends be significantly changed by minor changes in the weighting or categories. I don’t have the data (or the statistical skill) to make that judgement, but I am prepared to take the Inequality Index on face value – not least because the important thing about indexes is not so much how they are constructed, but their ability to show trends over time. In some senses, as long as they capture key elements well enough it becomes more important to maintain consistency over time then to continually adjust to political nuances.

Broader Critique

The question then is whether the Inequality Index does actually capture the key elements of inequality well enough – and here I do have some questions and critique. For instance,

• The income inequality data is based on standard measures household income, which I have argued previously are misleading as they fail to take account of housing incomes, social-transfers-in-kind, capital gains and capital income.
• The intergenerational inequality index focuses on current differences between age cohorts in rates of poverty and intended retirement age. It is too easy to dismiss these as life-cycle effects, and to me, the bigger intergenerational issues are long-term: what sort of economy, infrastructure, natural resource base and environment are we bequeathing the next generation? Is the next generation going to be better or worse off than the current or previous ones at same point or overall in their lives? These aspects of intergenerational inequality are not considered in the index.
• Both the ethnicity and disability inequality indexes are based on rates of reported discrimination and labour force participation, but there is no accounting for the income that comes from that participation (or not). I wanted to know the share of income and wealth held by those groups.
• Similarly, the First National inequality measure has 12 different components, but not one relating to income.

There are all sorts of good reasons for the choices about what to include and leave out, including the challenge (or impossibility) of getting robust and continuous public data on some of the issues above. For that reason, my main critique of the Inequality Index is not the points above (although they remain important), but two areas where I think the issues are of a much greater scale – the omission of class and geographic inequality.

Class

Despite the Summary Report’s Introduction acknowledging the importance of Thomas Piketty’s work, the Index uses the very bald income quintiles which Piketty criticises, and it does not examine the top 10% and top 1% where Piketty sees inequality growing at its most obscene. More importantly, the distribution of (some) income across a stratified income spectrum arbitrarily divided into quintiles does not capture class inequality or the structural inequality of the distribution of income between labour and capital.

The labour share of GDP is a much more robust measure of class inequality, and one for which there is robust ongoing data. While capital and labour incomes eventually land (differentially) in households on the household income spectrum, so to do the wage differentials of the gender wage gap and the differing incomes of varying labour force participations of other groups. This is no reason to exclude class inequality or assume it is covered by household inequality. I would have liked to have seen class, measured by the labour share of GDP included as an eighth sub-index, separate to and alongside the household income data.

Regions

The other significant omission from the Inequality Index is geographic inequality. The Australian population, income and wealth is concentrated in a small number of cities, and the data is clear that residents of some states (SA and Tasmania in particular) and people in regional and remote communities have significantly lower average incomes than those in the capital cities.

Beyond simply income, geography matters in terms of inequality in access to services. Many services cost significantly more (e.g. telecommunications) or are simply unavailable in many regional areas. Differences in access to health, education and other services can be seen as inequalities in the social wage, but they also have direct impacts on quality of life and the sustainability of communities. To ignore the geographic dimensions of inequality is a major oversight in measuring inequality in Australia.

Conclusion

Despite these queries and critiques, I still regard Per Capita’s Inequality Index as a bold and important initiative – a significant step beyond the narrow and flawed income measures used in much inequality analysis. I hope that in time the Index can be revised to incorporate some of the measures noted above, but either way, if Per Capita can sustain the methodology and index, it will be a valuable tool for understanding whether (and where) we are becoming more, or less, equal.

However, any socio-economic index is a tool, not an end in itself, and I suspect the greatest challenge for the Index is not its construction but its use. There are other indexes (e.g. the UN HDI, the Genuine Progress Indicator, and most recently, Wellbeing Budgets) that also reflect multiple dimensions of equality and wellbeing, but they pale in comparison to the use and status of economic statistics like GDP, the unemployment rate and CPI. Those official measures are sometimes misused, misunderstood or politically dubious, but they dominate economic discussion. They do so, not because they are the best scorecards of well-being or economics, but because they are causal variables (within ruling economic theories) used in economic management.

Accordingly, to be truly effective, the Australian Inequality Index will need to be not just a scorecard, but an active instrument of policy. Whether it has the theoretical framework and the mobilising power to play that policy role remains to be seen, but it is a start to build upon– and given my statistical efforts at measuring inequality, I am slightly jealous!