Australian House Prices down 10% from Peak
There are several providers of statistics on Australian house prices, but only one that doesn't have a vested interest in the direction house prices actually move in: the Australian Bureau of Statistics. So despite the criticisms of this series—that it's based on detached dwellings only, based on median sales data, too infrequent, not adjusted for "hedonic" differences between houses, etc., it's the only one I trust.
Click here for data in Excel: Debtwatch; CfESI
Click here for more data in Excel: Debtwatch; CfESI
Click here for this post in PDF: Debtwatch; CfESI
Chris Vedelago had a very nice piece about how confusing the various commercial house price statistics are:
The Real Estate Institute of Victoria said the city's median house price rose 0.9 per cent in the March quarter. Except that, according to RP Data-Rismark, it fell 1 per cent. Australian Property Monitors, which is owned by Fairfax, believes prices rose 1.6 per cent in the three-month period. Residex, on the other hand, estimates values fell 1.9 per cent… ("Confused about the market? We all are“, The Age April 29)
I’m happy to ignore these numbers—and even more so the spin doctoring that goes with them. The ABS numbers are in, and they show a 1.1% national fall over the March quarter. Sydney house prices fell 1.8% according to the ABS, whereas Australian Property Monitor alleged they rose 1.4%—the latter being the basis for Andrew “Always Look On the Bright Side” Wilson’s latest piece “Confidence rises as prices bounce back” (SMH April 28). Yeah, right.
Australian House prices have now fallen 6.1% from their peak, and have been falling for 21 months, which is the longest downturn in nominal prices ever recorded by the ABS—the previous longest being the 12 months from the beginning of the GFC (which was terminated by my favourite government policy of all time, the First Home Vendors Boost).
Figure 1: Nominal house prices have fallen 6.1% since June 2010

I’m sure the usual spruikers will come out with why this is now the bottom, and it’s a good time to buy, and there wasn’t an Australian house price bubble, and the shortage will drive up prices, and… So let’s put the current data in the context of the bursting of acknowledged overseas house price bubbles.
Firstly the inflation adjusted data: in real terms, house prices have now fallen 10% from their June 2010 peak, and are back to a level they first reached in late 2007.
Figure 2: Real house prices have fallen 10% since June 2010

Now let’s compare the Australian experience to date with the Japanese and US experiences—where no-one, not even Alan Greenspan, denies that there was a housing bubble. The Japanese bubble peaked in June 1991; the US bubble peaked in in May 2006; and Australian house prices peaked in June 2010. Figure 3 shows the three declines from the peak, and while the Australian experience so far is clearly better than the USA’s, it’s only a whisker better than the Japanese experience to the same date after the peak.
Figure 3: Comparing Japanese, US and Australian house prices from their peaks

Anyone who takes comfort from that should also consider the longer term perspective—see Figure 4.
Figure 4: The long term picture for Japan and the USA

The motive force behind Australia’s bubble was the same as in the USA and Japan: accelerating debt drove rising house prices during the boom. Now in both those countries, decelerating debt is driving house prices down. The same pattern applies in Australia—see Figure 5 .
Figure 5: Mortgage acceleration drives change in house prices

Don’t take heart from the uptick in acceleration at the end of the series there: for that to be sustained into the future, ultimately Australian mortgage debt would need to start rising (compared to GDP). But mortgage debt grew more rapidly here and reached a higher peak than in the USA (see Figure 6); the odds that it will rise again are slim.
Figure 6: Australian mortgage debt exceeded the USA’s

And even though the actual level of mortgage debt is still rising, it’s doing so at the slowest rate ever recorded by the RBA (see Figure 7).
Figure 7: Annual growth in mortgage debt (with series break in 1991)

The odds are that the rate of decline will accelerate in the next year—since as Leith van Onselen pointed out yesterday, many Baby Boomers are relying on rising house prices to secure their retirements. Now that house prices are falling, and have been doing so for almost 2 years, many of these Boomers—74% of whom earn less than $80,000 a year, with the average investor losing over $9,000 a year on these “investments”—could decide to get out rather than continue to absorb losses. The unwinding of their leveraged positions could push mortgage growth below zero, and of course accelerate the house price fall.


the debtwatch link to the first excel file is broken.
I’ll fix it later today–I did that post in a screaming hurry as noted earlier. I’ve also spotted an error in my Case-Shiller index importing routine which doesn’t affect this post, but has to be corrected for the data. Hopefully I’ll have that done by tomorrow.
The RBA weaved its magic on the Australian economy yesterday. The popular press is praising the benefit of a 0.5% reduction in interest:
http://www.businessweek.com/news/2012–05-01/stevens-seen-cutting-rate-to-record-by-swaps-australia-credit
Same commentators are looking skeptically at the banks to see that it is passed on.
So if the RBA gets its way Australians will rediscover their appetite for growing debt — as seems to be happening in the US with savings rate now declining again.
There is an interesting pie chart taken from a US Fed publication showing the composition of student loan debt by age:
http://libertystreeteconomics.typepad.com/.a/6a01348793456c970c016762736bfe970b-450wi
I wonder if the over 60+yo group are hopeful of selling their property to pay back their student loans. Maybe they are slow starters and are intending to work till they get in their 90s.
I don’t think there is a similar situation with student debt in Australia — yet. But only a matter of time before HECS recipients reach retirement age.
At some point, irrespective of low interest rates, the principal has to be repaid.
Hi, Steve,
I have always read your argument about housing price bubble in Australia with interest. However, even if it is bubble, it doesn’t have to “pop”. It can stagnate while the economy grows, or the air can sip out slowly.
One data point that turn me toward it is the large gap between housing completion and population growth Australia had in the recent years.
Is there any reason why you think it will “pop”?
Vancouver BC
If the numbers are accurate we are looking at a $128 000, or 10.5% drop in April 2012’s average price, since the Feb 2012 peak. And ~49k drop from March –April 2012, or ~4% drop in price mnth / mnth.
Vancouver apartments, according to this preliminary data, increased by $465 since end of March 2012 to now. Looks like a downsizing trend, and a move to the east, or where it’s least trend.
Question is, what areas is he including, the whole of RBGEV stats? Or Just Vancouver?
http://www.yattermatters.com/2012/05/its-cold-outside-for-vancouver-real-estate/
We will have to wait for the full stats from the REBGV.… or Steve, do you have any idea where more reliable 3rd party data on Vancouver Real Estate prices might be?
The only other data I found is attached. It comes from here, and when I calculated all of the areas listed, I got a 333k drop avg since March up to April 21st.
It did not include all areas on the REBGV MLS, but did include West/East Van, Richmond, North Vancouver.
Vancouver BC
If the numbers are accurate we are looking at a $128 000, or 10.5% drop in April 2012’s average price, since the Feb 2012 peak. And ~49k drop from March –April 2012, or ~4% drop in price mnth / mnth.
Vancouver apartments, according to this preliminary data, increased by $465 since end of March 2012 to now. Looks like a downsizing trend, and a move to the east, or where it’s least trend.
Question is, what areas is he including, the whole of RBGEV stats? Or Just Vancouver?
http://www.yattermatters.com/2012/05/its-cold-outside-for-vancouver-real-estate/
We will have to wait maybe for the full stats from the REBGV, or maybe someone has access here?
Steve, do you have any idea where more reliable 3rd party data on Vancouver Real Estate prices might be?
The only other data I found here: http://www.laurenandpaul.ca/Dexterselectmarketstats.ubr
When I calculated all of the areas listed, I got a $333,000 drop in average price, since the end of March 2012, to April 21st.
It did not include all areas on the REBGV MLS, but did include West/East Van, Richmond, North Vancouver.
What prompted my searching, was a report from Garth Turner mid April 2012, on his blog, that prices had dropped 100k on the REBGV MLS.
I had prevously posted my doubts about the apparent causal link betwen two data sets in one of the plots. I set up an absurdly simple model using Matlab-Simulink to look at apparent causality and quickly realised my simple observation of lead/lag with a pair of signals was no good.
The model comprised just a signal generator feeding a fixed delay — what could be simpler?. This corresponds to a cause (generator) and an effect (delayed generator).
If you apply a slowly increasing ramp signal, the cause clearly leads the effect. This corresponds to the observation of a long time period relative to the delay.
However, if the variation is cyclic, the apparent cause/effect can be anything you like, depending on the cycle period to delay time ratio. When the period is slightly longer than the delay, it looks like the ‘effect’ is pushing the ’cause’.
In engineering terms the phase delay is proportional to the input frequency, which is a fundamental frequency-domain property of a time delay. I knew this, but hadn’t thought it through in this context — interesting.
@ Steve Hummel May 2, 2012 at 12:38 pm
One advantage of Glass-Steagall over most other ad-hoc proposals is that it has proven to be effective in curbing speculation over many decades.
Lyonwiss,
Yes. Its very effective for business entities. And if we had a dividend and regulated how much of it went where, like 95% to retail spending or savings and maybe 1–5% to stocks or other non-derivative speculation it would effectively regulate individuals. With freedom comes responsibility after all. These kinds of commonsensical regulations of the financial/monetary system could keep the vices of economics largely out of effect.
Another thing to consider is how difficult it would be for regressive forces to overturn a dividend once it was instituted. Socail Secuirty has survived 80+ years.…and it is re-distributive. A distributive dividend steals/taxes no one and lasts a lifetime serving the individual more effectively and longer while simultaneously eliminating the necessity for welfare and SS and their respective tax burdens at the same time. Its win/win.….except as it impacts the duopoly/triopoly of credit and their largest market, consumer finance.
Henessy Avenue
A lot of the year to year population changes are due to changes to birth rates, which might affect changes to demand to larger homes but wont affect the requirement for total new homes directly. That one of many reasons why you cant use population levels alone as a proxy for housing demand, you have to look at household formation levels & also whether or not they affect stock levels (such as people living longer reducing stock turnover) or the flow of net additional households. Also different components of demographic change will have different lags on household formation. For example a baby wont be demanding a home of their own for at least 18 years.
The interelationship is also two way — for example if houseprices are expensive you get more concealed households (refusal to leave nesters, couples moving back in with parents, people forced to share etc.)
Steve
It should not be too surprising that correlations survive time accuracy issues for two reasons. Firstly it is irrational to sit on an extension of credit for two long, you have to pay interest on it so you try to put it to work straight way or in having a mortgage extended issue the bankers draft on the same day as you exchange contracts, which in most countries I know of will be done by your lawyer on the day of sale. If there is variation of delays in spending for credit advanced in excess of house purchase costs we would expect it to follow some kind of power law distributions evening out variations. Secondly even averaging out paths for change in debt over 1 year there is only so many paths that this could follow month to month and again we would expect these to be distributed around the straight a to b path in a regular way.
Thanks Brian, very nice to have that feedback about feedback. I am a sceptic about the applicability of conventional econometric tools to nonlinear dynamics, and this is a nice illustration of the same.
Could you send me the Matlab-Simulink modules?
Hi Steve,
I thought I’d let you know your compelling graph comparing US, Japanese and Australian house prices was shown by Alan Kohler last night on the 7:00pm ABC News. He did correctly quote you as the source.
Thanks e2; Alan rang me yesterday to ask permission to do so.
Yes, I was very pleased to see your graph on ABC, nice one Alan!
Hi Steve, enjoy your work. Any chance of a comparison of the debt profiles between Japanese,US, and Australian consumers during each of these deflation profiles? This would provide further insight into what to expect. It would also serve as a point of contrast as the Japanese are know for there savings rates; hence a lower debt to GDP and debt to income ratios.
link still broken, and the ABS series is 6416, not 4616
Sorry David–too much happened today, tomorrow is a conference all day. Will have to leave for the weekend.
Interesting to break this down between states. If you look at the ABS data for recent months you can see some continuing to power ahead, eg WA, NT, maybe ACT. But the drop has been in NSW and Vic. More interesting to me is that NSW has been treading water really for the last 10 years or so compared to other states.
Hi there Steve. I’m an admirer of your work.
I’ve just started trying to look what is happening with the Australian property bubble and have been searching for data.
I’ve managed to find statistics by the Supreme Court of Western Australia on “Civil Property Possession Applications” here:
http://www.supremecourt.wa.gov.au/content/about/statistics.aspx
However, I haven’t been able to find stats on this from any of the other states, but I know they are out there somewhere, because these people have the most recent figures:
http://au.news.yahoo.com/today-tonight/latest/article/-/12486561/mansion-repossessions/
Do you know if data from the other states is available online anywhere?
Thanks
Jeff
Thanks Jeff,
Unfortunately I don’t know that. It’s useful data though–the sort of thing I’d like to collect–but I haven’t had time to do so. Hopefully the behind the scenes changes we’re making to this blog–which will turn up in the next couple of months–will enable us to automate collecting stats like that.
Please keep us up to date with what you find.
You might also find that my good mates at Prosper Australia–www.prosper.org–have their fingers on this stuff.
Hi Steve,
I’ll certainly keep you up to date if I find anything.
I’m looking at Prosper Australia’s site now. (BTW you omitted the .au but I managed to work that out)
Hi Steve,
The best thing to do with your data series on mortgage demand and house prices is to apply empirical mode decomposition on each series, and see if any of the IMFs are correlated. EMD is much better than FFT for series that are either non-linear or have duration mismatch, and it sounds like your data is both.
Cheers,
JLP