For a supposedly classless society, we get surprisingly hot under the collar about the issue of ''middle-class welfare''.
The idea that well-off families, living comfortably on two wages, in a house they own in a swanky suburb may be living off the largesse of our taxes through barely deserved benefits and poorly targeted payments from the government makes our blood boil, and politicians recoil.
The Canberra Times highlighted this in a recent piece by Matt Wade (''Household inequality revealed in study'', November 15, page 3), which reports on a paper given at an Economics Society seminar last week by the Australian Bureau of Statistics. Wade points out ''the most revealing finding was the amount of government assistance being transferred to high-income households.
''The preliminary results showed the richest quintile of households received about 12 per cent of social assistance benefits while the next-highest quintile got 11 per cent … the findings not only highlight big income disparities across the community but raise questions about the scale of 'middle-class welfare' flowing to well-off families.''
These figures are surprising because previous studies show Australia targets the poor more than any other country in the OECD.
The new figures come originally from an ABS report on ''Government Benefits, Taxes and Household Income'', published in June this year and refer to the situation in 2009-10.
This report is the most comprehensive study of the distribution and effects of government social spending in Australia, and is invaluable for those developing and analysing government social policies. The survey collects information on the distribution of ''social assistance benefits'', the cash payments made by Centrelink such as age pension, disability support pension, family tax benefit, parenting payments, Newstart and so on.
Information is also collected on direct taxes paid by households, and the study models how much households receive in the form of benefits in kind (e.g. Medicare, pharmaceutical benefits, public housing, school and university education and childcare assistance) and how much is paid in indirect taxes.
On top of that, the 2009-10 study collected information on household net worth - or the value of assets held by households minus liabilities, such as mortgages.
The most important benefits in kind - the universal health insurance Australians receive from Medicare and the benefits of public education - are not income-tested, so households at all income levels benefit.
Family tax benefits give some help to a majority of families for reasons of horizontal equity - the presence of children increases household needs not just for the poor.
However, pensions and allowance payments are income-tested, so it appears surprising the richest quintile (20 per cent) of the population get as much as 12 per cent of these benefits. Since income-testing is intended to target assistance to those most in need, surely these new figures show that a lot of our benefit spending is going to households who don't need help?
Well, maybe not. Underlying these studies are a lot of technical judgments about ways of analysing data and presenting results. One of the most esoteric of these issues is how to rank households when dividing the population into different groups such as quintiles.
Households can be ranked from poorest to richest on a number of different measures, including the private incomes they have before they receive any benefits, the disposable incomes they have after they receive benefits and pay income taxes, the final incomes they have after their receipt of benefits in kind and payment of indirect taxes, or by their net worth.
Ranking by different measures produces widely divergent results. For example, when we rank households by their private incomes - which is what Centrelink uses to calculate people's benefit entitlements - the richest quintile only get 1.7 per cent of total spending, not 12 per cent.
But if households are ranked by their net worth, then the wealthiest quintile do get more than 10 per cent of all income-tested benefits.
How can this happen? To take an example, a 75-year-old widow might have no other income than her age pension of just over $20,000, but own her home worth $500,000, not counted in the pension assets test.
A 40-year-old couple, with children, earning $150,000 or more could be buying a house worth $1 million, but have a $500,000 mortgage. These two households would have the same net worth in housing, so when households are ranked by net worth they would be placed in the same wealth quintile, but would be in completely different income quintiles.
In Australia we have many older people with low incomes and high net worth, because of high levels of home ownership among older people, while younger people with much higher incomes are still paying mortgages. In fact, there are twice as many over-65s in the highest wealth quintile than in the lowest wealth quintile. So does this mean we should more tightly target pensions on the basis of people's assets?
It would certainly be possible to target more by including the family home in the assets test, and raising the threshold so that people with low incomes do not need to sell their homes to have enough to live on.
Alternatively, the value of housing above a high threshold could be counted as assets if we thought people in houses worth $1 million or more should receive lower pensions.
Any government that wanted to include the family home in the pension assets test would be, in the words of Sir Humphrey, ''courageous''.
But more importantly, would it be good public policy to have a system that denied public pensions to older people because they owned their own home? I suspect that's not the kind of attack on ''middle-class welfare'' anyone would want to see, especially our politicians.
Professor Peter Whiteford works in the Crawford School of Public Policy at the Australian National University.