Economists love data. For some it is a case of the more the merrier….or, at the least, the more data the more articles than can be published. Whether this contributes much to the overall good is doubtful. When it misleads it needs to be called out.
So it almost certainly is with the Productivity Commission’s recent report on inequality in Australia, Rising inequality? A stocktake of the evidence. The Commission does excellent work – it has skewered more than one sacred cow in its time, and I trust it will continue to do so. But it is not infallible, and from time to time it can become lost in data when what it really needs to do is stand back and ask whether the available data really make sense. Especially when it conflicts with other data that is readily available.
The Executive Summary of the Commission’s report runs to five pages, the full report to 162 pages. It is replete with tables and graphs, and those dealing with income and wealth inequality are based on two excellent data sources: the Australian Bureau of Statistics Survey of Income and Housing (SIH) and the Melbourne Institute’s Household, Income and Labour Dynamics in Australia (HILDA) survey. The headline in the report, the very first conclusion offered, is:
“Over nearly three decades, inequality has risen slightly in Australia.”
No wonder that the Productivity Commission launched its report at the National Press Club in Canberra, and not in a suburban or country pub. The reception would likely have been very different. A little further on in the report (page 69) it offers:
“Wealth inequality increased over the same period…..However not all measures show a clear upward trend.”
And further on again, at page 75:
“The top decile’s share of wealth has been fairly constant”
By now the quizzical looks in the front bar are turning to frowns and frank disbelief, and in some corners no doubt to righteous indignation. But are the punters right? Could it be that the Commission is wrong? On this occasion, and on these particular conclusions, the answer is yes. As the Commission itself realises. Why? In short, the limitations of the evidence the Productivity Commission relies upon, and the strength of the evidence in another data set the Commission chose not to examine – the Household Balance Sheet in the National Accounts (ABS 5232, Table 34.)
The Survey of Income and Housing and the Melbourne Institute’s HILDA Survey are both that – surveys. Both are well designed and well administered, but conclusions cannot be drawn from them that they were never designed to support. The SIH is based on a sample of around 17,700 households; HILDA on a sample a bit more than half that. Or in other words 0.2% and 0.1% of households respectively. And “households” in the SIH are defined to exclude at least one important sub-group – those living in institutionalised care such as nursing homes. As a group we have little hard evidence about their wealth. We do know that as a group older Australians are on average wealthier than younger Australians, including those approaching retirement. That wealth does not vanish, though it will diminish over time. Leaving 200,000 relatively wealthier persons out of the SIH means it will under-report wealth overall, and towards the top end of the distribution.
Both surveys cannot – and do not claim to – estimate the wealth or trends in the wealth of high wealth households. As the ABS readily acknowledges, the statistical uncertainty for SIH estimates of mean household wealth are such that beyond $5 million they urge caution; beyond $20 million the uncertainty is such they simply refuse to release estimates to the public. The HILDA sample is smaller, so can provide even less assurance about trends with regard to higher wealth households. The limitations of the SIH and the HILDA data set compared to the National Accounts data was pointed out by the Reserve Bank in its 2016 commentary on the inequality data.
The National Accounts are our most important economic data set. They lack the richness of SIH and HILDA, but they tell a story nonetheless. The National Accounts report estimates of Household Net Worth on a regular basis. They are invariably higher than the SIH estimates, and much higher than the HILDA estimates. And the gap has recently grown – from $350 billion in 2013-14 to over $800 billion in 2015-16 for the SIH. (The equivalent gap for HILDA is $1,870 billion as at December 2014.) The ABS, well aware of the gap, goes through a torturous exercise to attempt to reconcile the estimates – after all, they are intended to measure the same thing. They managed to get the gap down to 4% in 2013-14, but the latest exercise could only reduce it to 8%, or some $626 billion. The public bar may well take the view that is rather a lot to misplace.
So where is it? We don’t know. Perhaps there is a material flaw in the National Accounts and it doesn’t exist at all. (That prospect should give Treasurer Josh Frydenberg and Secretary Phil Gaetjens palpitations.) But perhaps the National Accounts are right about aggregate household wealth, or more right than the SIH (and certainly better than HILDA.) Perhaps a very large part of the gap is actually sitting in high wealth households – the top decile. In which case some of the Productivity Commission’s conclusions look decidedly unsafe. As they well know. Let’s hope they do something about it.
Joe Roach is a retired public servant with experience in the department of Finance and Defence. He is currently researching ways of improving taxation, including a tax on bequests between one generation and another.