The Conversation is fact-checking claims made on Q&A, broadcast Mondays on the ABC at 9:35pm. Thank you to everyone who sent us quotes for checking. Viewers can request statements to be FactChecked via Twitter using hashtags #FactCheck and #QandA, on Facebook or by email.
Australian statistics show that we are at the most unequal we’ve been in 75 years. – Leader of the Opposition, Bill Shorten, speaking on ABC TV’s Q&A program, September 21, 2015.
Members of the Opposition are fond of saying that inequality in Australia is at a 75-year high. That is, that inequality is worse now than it has been since about halfway through last century.
Is that statement supported by the research?
Checking the source
The source for Shorten’s stat is research by Shadow Assistant Treasurer Andrew Leigh, a former professor of economics at the Australian National University.
A spokesperson for Leigh directed The Conversation to the following graph, using data published in Leigh’s 2013 book, Battlers and Billionaires: The Story of Inequality in Australia.
The chart shows inequality since just after federation, defined as total income share held by the top earners. A higher figure is more unequal; lower means more equal, until one (when 1% hold 1% of total income).
Leigh says on his website that his analysis “is based on crunching tax data, national accounts figures, and population statistics.”
Leigh and fellow economist Tony Atkinson argued in a co-authored paper that inequality fell between the 1950s and the late 1970s. The same paper notes that for the top Australian earners:
There is a clear spike in 1950, mainly due to the peak wool prices which sheep farmers received in that year.
Commodity price anomalies aside, the overall trend in Leigh’s graph supports the narrative that income equality in Australia improved after the 1940s, began worsening in the 1980s, and is now back at levels not seen since the middle of last century.
So Shorten’s representation of Leigh’s data is perhaps a slight exaggeration but not a major one.
Technically, there was a spike in income inequality in 1950 so perhaps some sticklers will say Shorten should have said the most unequal in 65 years. Leigh’s data shows that apart from the spike, Australia is back to the level of top income shares of the 1940s (except for the war years 1944 and 1945).
What does other research say?
Work by Professor Peter Whiteford, a researcher on inequality at the Australian National University, shows that higher average incomes do not benefit all Australians if gains are only held by the wealthy few.
And Whiteford wrote last year on The Conversation:
The most common measure of inequality is the Gini coefficient, which varies between zero and one. If everyone had exactly the same income then it would be zero (perfect equality). If one household had all the income then it would be one (complete inequality)… Research by economists David Johnson and Roger Wilkins found that the Gini coefficient increased from around 0.27 in 1981–82 to around 0.30 in 1997-98. Subsequently, the official ABS income statistics show that the Gini coefficient increased to 0.34 just before the global financial crisis in 2008, then fell to 0.32 in 2011-12.
The most recent income survey released by the ABS tracks, among other things, how Australia’s Gini coefficient has changed since the 1990s.
It shows that inequality peaked in 2007-08 and then fell, but in the most recent year went up again but not quite to the 2007-08 level. That suggests inequality was a little bit higher seven or eight years ago, compared to now.
A quick qualifier: tracking the Gini coefficient is valuable because it gives information on the bottom income earners as well as the top. However, the Gini can be constructed using broad or narrow income measures, different data sources, and can look at either household or individual incomes. So it is not a simple matter to compare Gini coefficient results. The ABS, in their Table 1.1 data release, urged caution in interpreting the recent changes in Gini coefficient results, saying:
Estimates presented for 2007–08 onwards are not directly comparable with estimates for previous cycles due to the improvements made to measuring income introduced in the 2007–08 cycle.
Much of the discussion of inequality is about the pattern of change over time - the trend, not the year-to-year differences. Many would argue insufficient consistent figures exist even since 2007-08 to identify a trend.
Finally, the ABS Gini estimate is based on surveys (which the tax based measure of the 1% share used by Leigh isn’t), with survey errors (see ABS explanatory note 78), and so each Gini is a mid-point and needs a window drawn around it, to reflect the possible range of the true value. This window might show that the very small differences in the estimates since 2007-8 actually aren’t statistically different.
We should be cautious not to draw conclusions from any sole figure; any single inequality measure provides only a partial picture. Looking at a broad range of inequality measures and trying to grasp the trend gives a better picture of inequality in Australia.
It all depends on what figures you use.
Shorten’s representation of his ALP colleague Andrew Leigh’s data is perhaps slightly exaggerated but broadly correct – give or take a few years. There are not many inequality analyses going back as far as 75 years and most research supports the proposition that inequality has been rising in Australia.
There is data recently released by the ABS using the Gini coefficient that suggests inequality may have been a little bit higher seven or eight years ago than it is now. But there are strong doubts about whether this is true statistical difference or a trend. So it is too early to say whether inequality was stable or falling over the period from 2007-8 to now.
The fact check is a good summary of the available data on income inequality and, importantly, the difficulty in measurement. The spike in the 1950s was probably due to one-off factors, so it is fair to discount it. What we should emphasise is that our choice of index matters. Different indices focus on different parts of the distribution. Do we care more about the middle compared to the bottom of the distribution, the top compare to the middle, or the very top of the distribution (the famous 1%)? Whatever index we use, it is reasonably clear that inequality is not falling over the long term in Australia, and is a key area of policy concern.