The misleading housing statistic that just won’t die

It’s the misleading housing statistic that just won’t die. But it isn’t a claim being made by politicians in the heat of an election campaign. Instead it keeps appearing in the reports of some of Australia’s most prominent housing researchers.

The Australian Housing and Urban Research Institute (AHURI) argues that most new housing built in Australia is too expensive for low- and middle-income earners. The claim keeps being repeated, but it’s based on flawed research already de-bunked by Grattan Institute. The sooner these statistics die, the better Australian housing policy will be.

The claim

A 2017 report for AHURI claimed that:

“Most of the growth in housing supply has been taking place in mid-to-high price segments, rather than low price segments.” (p.2)

“Almost 80 per cent of house approvals are to be found in the 6th to 9th deciles, a range covering transactions between $306,000 and $795,000 in 2013–14.” (p.16)

The claim appeared again in a 2018 report on homelessness by the Centre for Social Impact, co-authored by other AHURI researchers:

“Further evidence for the idea that Australian housing unaffordability and homelessness is not simply due to insufficient housing supply is found in the fact that we have seen increases in both housing supply and housing prices in recent years. Largely, this is because growth in housing stock has been concentrated in the middle and upper end of the market: almost 80 per cent of new stock was in the 6th to 9th price deciles.” (Ong et al., 2017)

And it appeared again last week in a McKell Institute report evaluating the potential impacts of federal Labor’s policy on the National Rental Affordability Scheme (the so-called NRAS 2.0 policy):

“Without regulation and a government policy that incentivizes private investment, the market does not deliver affordable rental options for those on low or even moderate incomes, with almost 80% of new housing stock priced at the upper end of the market.”

Why it’s wrong

The trouble is, the statistic is just plain wrong.

For a start, it defies basic common sense. Where are most new houses being built today? On the greenfield fringes of our major cities. Are the new houses built in Werribee, Wyyndham Vale or Donnybrook in Melbourne’s west and north, or in the far reaches of Greater Western Sydney, really among the most expensive in Australia?

The AHURI claim implies that half of all houses in Australia were worth less than $306,000 in 2014. But the median Australia-wide house price in 2014 was $571,000.

How did they get it wrong?

The original AHURI researchers presented analysis of building approvals in each Local Government Area in Australia. They counted the number of new house and apartment approvals in each of 2005-06 and 2013-14 in each LGA, and then ranked LGAs from lowest to highest according to the real median prices of houses and apartments. They then divided the LGAs into 10 groups (or deciles), each containing the same number of LGAs.

The authors found that over the past decade almost 90 per cent of houses and 95 per cent of units were built in the 50 per cent of LGAs with more expensive housing. Almost 80 per cent of house approvals were in the 6th to 9th deciles.

The critical flaw is that when they grouped LGAs into deciles, the AHURI researchers failed to weight the LGAs by the existing number of dwellings in each.

This is more than a rounding error, because LGAs have very different populations, ranging from 62 people in Maralinga Tjarutja in far-west South Australia to 1,184,215 in the LGA covered by Brisbane City Council. Brisbane has 459,227 houses and units; Maralinga Tjarutja has 14.

Just 5 per cent of Australians live in the smallest half of all LGAs, whereas half of all Australians live in the 10 per cent of LGAs with the largest populations — all of which are in or close to the major cities.

Small LGAs are not randomly distributed around Australia – most are in rural and regional areas, where few jobs are being created. These aren’t the places where we would expect much new housing.

It is hardly surprising that most new housing is built in the top three LGA deciles – they already house most of the population. These LGAs cover the vast bulk of Australia’s major cities, where population and jobs are growing faster than regional areas. And so it is misleading to conclude that new housing has been disproportionately built to serve the top end of the market.

Our analysis of the same data, updated to 2016-17 and accounting for the different sizes of LGAs, shows that most new houses are being built in cheaper-than-average suburbs on the fringes on the major cities. Two-thirds of all new houses built in 2016-17 were in areas with cheaper house prices than average, and 16 per cent of new houses were built in the cheapest 20 per cent of LGAs.

Better policy requires better analysis

Grattan first published our critique in Inside Story in February 2018. We then presented a peer-reviewed paper to the Australasian Housing Researchers Conference in June 2018.

AHURI has not publicly responded to our critique. Its report remains on the AHURI website. Worse, the claim continues to be cited in debate on housing policy in Australia today.

Some advocates of affordable housing are using this research to argue that boosting the supply of market-rate housing won’t make housing more affordable. That’s wrong – and it allows politicians to avoid politically difficult planning reforms that would make housing substantially more affordable in the long term.

Australians deserve – and people at risk of poverty and homelessness need – evidence-based policy based on high-quality research.