‘Australia – a nation of millionaires‘. Catchy headline – but is it true?
Of course not. But the ABS came in for some criticism over the last month or so, after they issued a press release that led to widespread reporting of this factoid, with accusations of political partisanship and glossing over clear indicators of growing inequality in the data.
So, headlines aside, what are the facts? Is Australia becoming a more unequal society? The answer, of course, is ‘it depends’. Are we talking about inequality in income or wealth? When you talk about the ‘average’ Australian, are we talking mean or median averages (in either case, these definitions are important, as they can change the answer).
In this piece, Glenn explains why, as a statistician, he’d use the median, not the mean average when analysing changes in wealth and income; why, as a demographer, he’d go with the Gini coefficient as a better measure of the total level of inequality, and why he’d point to house prices as the cause of growing wealth inequality.
In July, the ABS released the latest version of their biennial survey, “Household Income and Wealth, Australia” (6523.0).
This has generated quite a bit of interest. Because it does show a slight increase in wealth inequality in Australia over time, and the ABS also put out a media release saying that average household wealth is now over $1 million per household – true, but it nicely illustrates a difference between means and medians as a measure of average. So overall wealth is going up, but it is more unequally distributed – is this a concern?
How do we measure wealth and income equality?
Every two years, the ABS runs the survey of Income and Housing Costs with a random sample of around 15,000 households to measure the value and distribution of household wealth and income.
The 2017/18 survey showed that household wealth (assets owned by households minus liabilities) in Australia has indeed topped $1 million per household for the first time (in 2017-18 dollars inflation-adjusted, it was $1,022,000 per household). This is a simple mean average– divide the total wealth of all households by the number of households. The median (mid-point of all households) is only $558,900.
The ABS did put out a media release about this and was criticized for using the mean rather than the median – though both numbers are on their publication page, just not in the media release. It’s quite a catchy headline to say that household wealth is now over $1 million – it doesn’t necessarily imply a political motive. But the median does tell a different story.
Why median works better for income and wealth
The mean is skewed upwards by a few very high net worth households at the top end – only about 30% of households had wealth above the mean of $1,022,000, so about 70% were below it.
A difference between the mean and the median is an indicator that the wealth is not evenly distributed (if it were, the mean and the median would be equal).
But in reality, the fact that the distribution can skew further at the top end than the bottom of the distribution (zero is the bottom, but there is no upper limit) means that mathematically, the mean for income will always be higher than the median.
This is why you’ll usually see the median used when dealing with numbers that have a hard lower limit but no upper limit – particularly things like house prices – where you almost never see the mean used. Distributions with a long tail to the right on a chart will always have a higher mean than median.
How have average household income and wealth changed?
When you look at both of them for this dataset, Mean household wealth increased 6.1% since the last survey (about $58k increase) while the median increased by 2.3% (only $12k). This indicates that the top end of wealth is increasing at a faster rate and pulling the mean upwards.
Figures are inflation-adjusted to 2017-18 year and sourced from the ABS publication Household Income and Wealth, Australia (6523.0)
The main drivers of this increase have been increases in house prices and superannuation balances. These disproportionately affect homeowners, who are more likely to be in the higher ranges of the wealth distribution (but not necessarily the income distribution).
The difference between mean and median can provide an indicator of inequality – but there is a more precise way of measuring the distribution.
The Gini Coefficient
The international measure of inequality is the Gini Coefficient. While it has its limitations, it’s a better indicator than just looking at the difference between the mean and the median. The Gini Coefficient is a number between 0 and 1. If Gini is zero, that means every household has exactly the same income. A theoretical Gini of 1 means that one household has all the wealth and all other households have none at all. In reality, all nations have Gini coefficients between these extremes, with most falling between 0.3 and 0.7.
Australia’s Gini coefficient for wealth did increase from 0.593 to 0.621 in the last 12 years. This indicates that wealth is indeed becoming more unevenly distributed.
Looking at the extremes in the detail, the top 20% of all household wealth had an average of $3,237,000 in household wealth in 2017-18, up by about $1 million in 12 years (inflation-adjusted).
In contrast, the lowest 20% of household wealth remained virtually unchanged over the same time period, at around $35,000 (it actually did appear to fall slightly, but this is within the margin of error of the survey, which is also published). The share of total wealth owned by the bottom 20% fell from 0.8% to 0.7% in the last survey period. The table below shows the same figures but equivalised for the number of adults and children in the household. The percentages are almost identical to total wealth, but the actual $ appear lower.
Table showing share of income and wealth by quintile (20% of households in each group)
Figures are inflation-adjusted to 2017-18 year and sourced from the ABS publication Household Income and Wealth, Australia (6523.0)
It’s important to note that these are surveys, with a sample size of around 14,000-15,000 households, are subject to sampling error and the relative sampling error is always higher among the lowest wealth and income groups. Nevertheless, it’s clear that the majority of increases in wealth have been in the higher parts of the wealth scale. The share of wealth owned by the bottom 20% of households is very very small, while the top 20% own nearly two-thirds of the wealth. What causes this?
Wealth changes over a lifetime
If you look at the ‘Age of reference person‘ table for this research, the majority of the lowest 20% of household wealth consists of young people who don’t own their own home (and also some elderly pensioners who are renting – these are arguably the more vulnerable group). These households do move into the higher ranges over time, so the bottom 20% don’t always remain the same households (nor do the top 20%), however as this is not a longitudinal survey, we’re not able to demonstrate how individual households’ fortunes change over time.
It’s clear that there is a very strong correlation with age, however (Table 10.2 in the data) – from 15-24 through to 65-74, net worth increases in each age bracket. This is clearly due to home ownership, and paying off a dwelling – Only 1.3% of 25-34 year olds fully owned their home (most were renting), while 69.3% of 65-74 year olds did. At 75+, wealth declines a little, probably due to the impact of aged care, or simply that those in this age bracket didn’t have superannuation built up during their working years.
In general, though, wealth increases with age – and home ownership is the key.
Are the poor getting poorer?
It’s tempting to say “the rich are getting richer, and the poor are getting poorer”, but this is not quite true.
Firstly, the poorest households are not getting poorer – their relative share of Australia’s wealth, already low, has got lower, but it has kept pace with inflation, being roughly the same over 12 years.
It is clear, however, that the large increase in housing prices seen in Australia has lifted the wealth of anyone who owns housing, so there is a real divide between those lower wealth households who mostly don’t own housing and the higher wealth households who mostly do. As housing prices increase it will naturally lift the wealth of those in the higher ranges. Wealth is more unequally distributed because it tends to increase with age, far more so than income.
Perhaps this is why the ABS in their second media release chose to focus on income – which doesn’t show the same level of change, nor the same level of inequality.
What does Gini tell us about inequality?
The Gini coefficient for income has barely changed in 12 years – at 0.439 for gross income, and 0.328 for Equivalised income, which is adjusted based on the number of adults and dependent children in the household. The mean household income is $1,062 per week, and is relatively unchanged in the past 10 years, when adjusted for inflation (in fact the ABS does not regard the difference shown to be statistically significant).
Even the top 20% of incomes haven’t increased much in that time ($2,085 per week to $2,142 per week). In the same time, the lowest 20% of incomes have risen from $371 per week to $399 per week – also a relatively small change but larger in percentage terms than the top 20%.
The importance of equivalisation
This equivalisation factor is really important – and not well understood. Basically, equivalisation allows you to compare incomes for houses of different sizes and compositions. Without equivalisation, you couldn’t compare the income of a household where a single parent supports three dependents, with that of a ‘double income no kids’ household, or a share-house with four people in their early 20’s all earning an income.
Income is a lot more unequally distributed in terms of the actual $ per household, but often this is a product of larger households having more income. So when you factor in household size, the income inequality drops. We use this equivalisation on the .id community profile to allow users to compare incomes for households of different sizes and compositions.
Ok.. so is there more or less inequality in Australia, compared with 10 years ago?
It’s clear that the increase in wealth at the top end is largely due to the increase in the value of the family home – which doesn’t generally generate an income. And when you look at incomes, while there is certainly inequality, it is far less than looking at wealth. Wealth is increasing nationally, but inequality in wealth is also increasing – the lower wealth ranges are increasing far more slowly than the upper ones, because the lower quintiles generally don’t own housing, and haven’t benefited from the increases in house prices.
There is no doubt that the value of housing has lifted home-owning households’ wealth, but it hasn’t really lifted their income over time. This may be why the ABS chose to focus on this relative stability in their media release, rather than any political conspiracy.
In fact, the CIA World Factbook shows that incomes in Australia are among the most equal in the world using the Gini coefficient method. We rank 134 out of 157 countries that they list, for income inequality (where #1 is the most inequality).
As for whether or not the ABS was right to issue a press release that cited the mean, not the median, household wealth – I guess that depends on your politics. I know which one makes a better headline, but does that headline help us better understand inequality in our communities?
