According to key economic commentators like Michael Stutchbury (News Ltd) and Alan Mitchell (Fairfax), Australian housing is very expensive by international standards. This mantra is recycled in the media on a weekly basis. It’s also supported by some economists who incorrectly claim that Australian property prices are 7-8x disposable household incomes.
The high cost of Australian housing is often pinned on our purportedly concessionary tax system that exempts the family home from CGT. The Henry Review should be directed at remedying this failure, some pundits say. And the RBA is advised to chase down house prices using the blunt monetary policy instrument, even though this is neither part of the RBA’s mandate, nor likely to be effective given the collateral damage that it would inflict on the wider economy (as the Treasury’s Dr David Gruen and the RBA’s Dr Guy Debelle have both independently observed).
The hard empirical fact is that these journalists and economists are wrong. Before we show why, think about it intuitively.
Australia has an internationally high rate of home ownership (around 70 per cent at the time of the 2006 Census). Yet we also have amongst the lowest mortgage default rates in the world with just 0.66 per cent of the five million borrowers with a loan materially behind on their repayments. This is nearly one-tenth or one-quarter of the default rates observed in the US and UK, respectively.
Based on the RBA’s preferred benchmark, housing affordability is no worse than it has been on average over the last 28 years. Even during the recent nadir in the early 2000s, affordability was better than it was in the late 1980s and early 1990s.
Finally, Australian house prices are rising, not falling despite the fact that the deposit requirements and credit rules imposed by lenders today are the toughest they’ve been in over 15 years. Even when mortgage rates hit 9.6 per cent in 2008, house price falls were only modest.
If Australian housing was unusually expensive, you might expect to see low ownership levels, high default rates, historically poor affordability, and weak demand as manifest in negative price growth. But we do not.
In a speech earlier this year, Glenn Stevens disclosed the RBA’s estimate of Australia’s ‘dwelling price-to-income ratio’, which is a metric commonly used to compare the cost of housing around the world. The RBA found that dwelling prices in Australia’s capital cities were just 4.8x disposable household incomes, which is nearly half the ratio proposed by many pundits.
In private correspondence with the RBA, I highlighted several problems with this analysis generally (notably, not the RBA’s specifically).
First, one quarter of all homes are not the more expensive free-standing ‘houses’; they are semis, terraces and apartments. So one cannot compare ‘house’ prices to incomes – you need to use a ‘dwelling’ price proxy that includes all property types (as the RBA correctly does).
Second, 40 per cent of Australia’s housing stock is not located in the capital cities; it is found in cheaper outer-metro and regional markets.
When RP Data-Rismark accounts for both these facts, we find that Australia’s median dwelling price is just $370,000, which is substantially less than the $500,000 figure often quoted in the media.[1]
Figure 1 below shows the time-series change in nominal median dwelling prices located across Australia and in capital cities since 1980. Figure 2 overlays Australian household disposable incomes against these two variables.[2]
Source: RP Data-Rismark (Click to enlarge)
Source: RP Data-Rismark (Click to enlarge)Taking into account all homes in all regions, and using the RBA’s definition of ‘disposable household incomes’ (remembering that there are on average 1.3 employed persons per household, and ‘income’ includes all earnings from savings, investment and labour sources – ie, not just wages), Australia’s dwelling price-to-income ratio is just 4.1x today. This is a fraction of the 7-8x income estimates often advanced by experts in search of a headline.
And so contrary to popular conjecture, Australian housing does not appear to be overly expensive by international standards.
Figure 3 visually illustrates the change in the ‘capital city’ and ‘all regions’ dwelling price-to-income ratios since 1980.[3] While there is a slight difference between these lines, this is likely explained by the fact that the denominator is Australia-wide disposable incomes. If we divided through by capital city incomes, the discrepancy would presumably disappear.
Source: RP Data-Rismark (Click to enlarge)
Source: RP Data-Rismark (Click to enlarge)As both we and the RBA have argued previously, the increase in household debt levels, and other metrics such as price-to-income ratios, during the 1990s would seem to have been once-off ‘level-effects’ associated with the secular reduction in inflation and nominal interest rates brought about by (in part) the change in the monetary policy framework, which formally shifted to an ‘inflation targeting’ approach, the adoption of an explicit CPI target of between 2-3 per cent per annum, and the establishment of an independent central bank during the 1990s.
To examine this issue more closely, Figure 4 compares nominal mortgage rates with headline CPI since 1980. The structural downward shift in inflation and nominal interest rates is easy to identify. In Figure 5, we add nominal mortgage rates to the capital city and all-regions dwelling price-to-income ratio series. Observe in this latter chart how the blue ‘all-regions’ dwelling price-to-income ratio shifts up from a relatively stable 3x household incomes to 4x incomes during the late 1990s, where it has stayed static ever since. This is entirely consistent with once-off ‘level-effect’ explanation outlined above.
Source: RP Data-Rismark (Click to enlarge)Australia’s lower than expected dwelling price-to-income ratio also fits well with IMF analysis that we recently presented to the annual Melbourne Institute conference, which showed that growth in real Australian house prices relative to incomes, and house prices in isolation, had been no greater than our median peer country over the last 10 and 20 years, respectively.
A related fallacy here is that Australia’s housing tax regime is unusually generous. But as Ernst & Young have shown, Australia’s tax treatment of housing (including the CGT exemption and negative gearing) is in line with the practices found in the UK, New Zealand, France, Canada and Germany.
Commentators also typically overlook the fact that families pay a big price for their CGT exemption: the interest repayments on their home loans are not tax-deductible, as they are for all other asset-classes.
One country that does levy CGT on the family home in exchange for interest deductibility is the US. Yet this model encourages households to leverage up, which results in higher loan-to-value ratios, longer mortgage pay-down periods, and higher default rates in comparison to Australia where borrowers service debt out of post-tax income.
Many sociologists have argued that the Henry Review should seek to remedy the (non-)concessionary tax treatment of housing by recommending a new land tax. But here they also tend to forget that Australians already pay hefty land taxes of circa one per cent per annum of the value of their homes via local government rates and charges.
There is, to be sure, a serious policy problem afflicting the cost of Australian housing, but it has nothing to do with established house prices or the tax-treatment of property owners. As the Prime Minister’s Home Ownership Task Force first documented in 2003, the challenge we face is government-imposed frictions on the supply-side of the market that have caused a growing ‘housing needs-production mismatch’.
Table 1 below shows that total local, state and commonwealth government taxes and charges in Sydney can now amount to up to nearly $160,000 of the cost of a new dwelling. The rise in these levies has been principally driven by local and state government duties, where property taxes as a share of revenues have grown fivefold over the past quarter century according to the PCA.
Source: RP Data-Rismark (Click to enlarge)Combined with zoning restrictions that artificially inflate the price of land over time, it is now difficult for Australia’s building industry to produce a new house in Sydney or Brisbane for less than $550,000 (see Table 1). This is well above the market-clearing price of a home for many prospective buyers (and particularly lower-income consumers). The consequence is that the supply-side is not generating the number – nor the type – of homes that the country requires, with dwelling starts per capita near their lowest level since the late 1960s (see Figure 6).
Source: ABS; RP Data-Rismark (Click to enlarge)The supply that is coming online is also a poor substitute for existing homes because it does not have the same ‘amenity’ value (think location, schools, hospitals, transport etc). This creates additional risks for investors since the high cost and poor location of new supply means that it has questionable demand prospects, which is one reason why the finance available for development has dried-up.
There are two obvious solutions to Australia’s housing cost problem that Kevin Rudd, Wayne Swan and Tanya Plibersek can avail themselves of. First, they should relax density restrictions in existing urban areas where high housing demand exists. Bringing new supply online on the fringes of cities, which is far removed from key labour markets, has weak infrastructure support, low consumer demand, and much higher environmental impacts is emphatically not the answer.
Second, the Henry Review should advocate substantially reducing the tax burden on the supply-side of the housing market, which will help reduce its cost of production and encourage investors to commit scarce capital that is competing with the resources sector to the supply of lower-cost shelter.
Footnotes:
[1] This is the simple median price (or 50th percentile observation) derived from all home sales collected in Australia by RP Data Ltd, which represents 99% of all transactions nationally. This is different to the median values reported by APM, RP Data-Rismark, and the ABS for several reasons. First, it includes all property types (ie, not just detached houses, like the ABS). Second, it covers all regions rather than just the capital cities, which APM, RP Data-Rismark and the ABS normally focus on. Third, the median value reported by APM is calculated using a ‘stratification technique’, which is different to the simple 50th percentile observation used here. RP Data-Rismark’s ‘median values’ must also be interpreted differently. These are, in fact, the index values attributable to the RP Data-Rismark ‘hedonic index’, which was originally based at inception on median automated property valuation estimates (ie, the median of a statistical valuation of all capital city homes). The change in the index value over time reflects the underlying capital growth rates generated by residential property in the relevant region. These growth rates are not influenced by capital expenditure on homes, compositional changes in the types of properties being transacted, or variations in the type and quality of new homes manufactured over time. The RP Data-Rismark ‘median values’ are not, therefore, the same as the ‘simple median price’ associated with all homes sold during a given period.
[2] We use the disposable income definition employed by the RBA, which is the National Accounts measure divided by the number of dwellings. Here the RBA comments: “For income, our conventional measure is household disposable income before interest payments and excluding unincorporated enterprises obtained from the national accounts. Our view is that interest payments should not be deducted from household disposable income because these are funds that are available to service loan repayments, while most income from unincorporated enterprises is profits on capital rather than from labour. To make it per household, we use the number of households which is reported by the ABS…This provides a yearly time-series of the number of households, which we can use to interpolate the other quarters.”
[3] The gold star depicts where some economists have claimed that Australia’s price-to-income ratio sits. The red star is the latest estimate from the Demographia survey (see here for our comments on this latter study).
