This Time Had Better Be Different: House Prices and the Banks Part 1

Flattr this!

Click here for this post in PDF

Before the US house price bub­ble burst, its banks and reg­u­la­tors claimed (a) that there was­n’t a bub­ble and (b) that, if house prices did fall, it would­n’t affect the sol­ven­cy of the banks.

The same claims are now being made about Aus­tralian house prices and Aus­tralian banks. On the for­mer point, Glenn Stevens recent­ly remarked that:

There is quite often quot­ed very high ratios of price to income for Aus­tralia, but I think if you get the broad­est mea­sures coun­try-wide prices and coun­try-wide mea­sure of income, the ratio is about four and half and it has not moved much either way for ten years.

That is high­er than it used to be but it is actu­al­ly not excep­tion­al by glob­al stan­dards.
(SMH March 16th 2011)

On the lat­ter, APRA con­duct­ed a “stress test” study of Aus­tralian banks in 2010, with the stress­es includ­ing a 30% fall in house prices over 3 years:

Table 1: APRA Stress Test Table, APRA Insight 2010/2, p. 9

2009/10 2010/11 2011/12
GDP growth (%) (-3.0) 2.1 3.5
Unem­ploy­ment (%) 9.8 10.8 10.7
House Price Growth (%) (-11.8) (-12.1) (-1.7)
Com­mer­cial office prop­er­ty growth (%) (-21.5) (-9.4) 1.5

APRA’s con­clu­sion was:

The main results of the stress-test for the 20 ADIs, tak­en as a group, are as fol­lows:

  • none of the ADIs would have failed under the down­turn macro­eco­nom­ic sce­nario;
  • none of the ADIs would have breached the four per cent min­i­mum Tier 1 cap­i­tal require­ment of the Basel II Frame­work; and
  • the weight­ed aver­age reduc­tion in Tier 1 cap­i­tal ratios from the begin­ning to the end of the three-year stress peri­od was 3.1 per­cent­age points. (APRA Insight 2010/2, p. 10)

So there’s noth­ing to wor­ry about then? No bub­ble to pop, and no prob­lems for the banks if house prices fall any­way? In this post I’ll con­sid­er the argu­ment that there is no bub­ble because changed eco­nom­ic fun­da­men­tals jus­ti­fy Aus­trali­a’s rel­a­tive­ly high house prices. In the next I’ll con­sid­er what the pop­ping of the bub­ble could mean for Aus­tralian banks.


Glenn Stevens’ claim that the house price to income ratio was “about four and a half” was almost cer­tain­ly rely­ing on research by Ris­mark. Ris­mark MD Chris Joye recent­ly assert­ed that the house price ratio in Aus­tralia was 4.6, and though he con­ced­ed this was some­what high, he argued that it was jus­ti­fied by changes to eco­nom­ic fun­da­men­tals. He ridiculed the claim, made by The Econ­o­mist on the basis of a com­par­i­son of house prices to rents, that Aus­trali­a’s house prices are 56% over­val­ued:

The Econ­o­mist does not ques­tion whether the old hous­ing ratios might be non­sen­si­cal to today’s home own­ers as a result of:

  • Fun­da­men­tal changes in the struc­ture of the econ­o­my wrought by the fact that inter­est rates over the past 15 years have, on aver­age, been 43 per cent low­er than inter­est rates in the 15 years that pre­ced­ed that peri­od;
  • The fact that aver­age infla­tion since the mid­dle of the 1990s has been 55 per cent low­er than infla­tion in the 15 years pri­or; or
  • The fact that the rise of two-income house­holds and the female par­tic­i­pa­tion rate in con­cert with a near halv­ing in the nom­i­nal cost of debt might have trig­gered a once-off upward increase in house­hold pur­chas­ing pow­er, and hence hous­ing val­u­a­tions… (Chris Joye, A prop­er­ty bub­ble long shot, Busi­ness Spec­ta­tor March 25 2011)

For­mer RBA staffer and now HSBC econ­o­mist Paul Blox­ham was equal­ly adamant: Aus­tralian house prices are a tad high, but they are jus­ti­fied by changed eco­nom­ic fun­da­men­tals over the last 15 years:

… a large struc­tur­al adjust­ment that occurred in the Aus­tralian hous­ing mar­ket between 1997 and 2003… involved low­er inter­est rates, bet­ter-anchored infla­tion expec­ta­tions, and increased avail­abil­i­ty of hous­ing cred­it. With­out some rever­sal of these struc­tur­al changes – which is a vir­tu­al impos­si­bil­i­ty – we do not expect Aus­tralian hous­ing prices to fall…

Since late 2003 the dwelling price to income ratio has been broad­ly sta­ble at between 3.5 and 4.5 and has aver­aged 4 (see chart)…

We view the risk of a sharp fall in hous­ing prices as very low. (Paul Blox­ham, The Aus­tralian hous­ing bub­ble fur­phy, Busi­ness Spec­ta­tor March 18 2011)

Fig­ure 1: Ris­mark’s Dwelling Price to Income Ratio Chart

In oth­er words, this time is dif­fer­ent.

They would say that, wouldn’t they?

The “this time is dif­fer­ent” argu­ment asserts that low­er inter­est rates, low­er infla­tion and high­er income per house­hold (and more income earn­ers per house­hold) means that though the house prices to income ratio might high­er than before, it’s noth­ing to wor­ry about.

Tell that to a would-be first home buy­er who’s con­tem­plat­ing tak­ing out a mort­gage. In 1992, the aver­age mort­gage for a First Home Buy­er was $ 71,500. It is now $274,000.

Fig­ure 2: Aver­age First Home Mort­gage and Mort­gage Inter­est Rate

The “no bub­ble” argu­ment asserts that this has been coun­ter­bal­anced by the fall in inter­est rates—which were 12% then and are 7.8% now. So the aver­age first home buy­er’s mort­gage is 3.8 times high­er than it was two decades ago, while inter­est rates are 2/3rds what they were then. Does one—along with changes in income and demographics—counterbalance the oth­er?

Not on your life: the increase in debt and debt ser­vic­ing has far out­stripped all the fac­tors that Joye and Blox­ham rely upon to argue that Aus­trali­a’s house prices are not in a bub­ble.

I want to make this case slow­ly, so that you can see each step in the argu­ment, so let’s first look at the week­ly inter­est and loan repay­ments on a typ­i­cal 25-year First Home hous­ing loan. Back in 1992, the week­ly inter­est bill was $165; now it is $420—2.5 times as high. Repay­ments were $174; now they are $490—2.8 times as high.

Fig­ure 3: Inter­est up 2.5 times, repay­ments up 2.8 times

So have incomes risen suf­fi­cient­ly to mean that this almost three­fold increase in debt ser­vic­ing costs over 20 years is no big deal?

Not if you’re a wage earn­er! Aver­age before tax wages have risen from $505 a week in 1992 to $996 a week at the end of 2011—so they have almost dou­bled. Using an aver­age tax rate of 28%, that gives the aver­age wage earn­er $777 after tax a week today, ver­sus $394 back in 1992.

Fig­ure 4: Aver­age wages have risen by 97% since 1992

While wages have risen, the 2.8 times increase in loan repay­ments means that mort­gage pay­ments on an aver­age first home loan have gone from tak­ing 40 per­cent of after-tax income of the aver­age work­er in the 1990s to 64 per­cent now—after reach­ing a peak of 74 per­cent in late 2008 before the RBA slashed inter­est rates (the ratio fell to 53 per­cent, and it would have fall­en fur­ther had the First Home Ven­dors Boost not caused house prices to sky­rock­et again).

In the ear­ly 1990s, a young wage earn­er could aspire to financ­ing a house pur­chase using his or her income alone. Now, that’s out of the ques­tion.


He’s a (young) Working Class Man Renter…

This is what the “no bub­ble” pro­po­nents don’t get: high house prices have become a class and age issue. If you’re a young “work­ing class man” on the aver­age wage, you can no longer afford to enter the hous­ing mar­ket in Australia—since the aver­age first home loan would con­sume over 60 per­cent of your after-tax wage.

Even if you’re a “young work­ing class cou­ple”, the cost of ser­vic­ing a mort­gage from wage income alone is pro­hib­i­tive. In the 1990s, a cou­ple (where both earned the aver­age wage) had about 80% of their income free for oth­er costs after pay­ing the aver­age First Home mort­gage. The rapid esca­la­tion in house prices after Howard dou­bled the First Home Own­ers Grant in 2001 drove this down to under 65 percent—and most wage-earn­ing cou­ples sim­ply don’t have that much head­room in their bud­gets. They can’t pay the rates, the food bill, the petrol, and the edu­ca­tion fees, with less than three quar­ters of their after-tax income.

Fig­ure 5: Max Headroom–disposable income after pay­ing the mort­gage plum­mets as prices rise

Faced with this lev­el of poten­tial debt-ser­vic­ing costs, young would-be house-buy­ers are giv­ing up on the dream of home ownership—and its atten­dant night­mare of debt peon­age. They’re also sign­ing up in droves to call for a polit­i­cal cam­paign against house prices by Get­Up: see the Anti-FHOG, Anti-Neg­a­tive Gear­ing, and Buy­ers Strike cam­paign sug­ges­tions (and read David Llewellyn-Smith’s excel­lent piece on it in the Fair­fax press too).

A “Buy­ers’ Strike”, whether orga­nized or not, is what will end the Ponzi Scheme of debt-inflat­ed house prices, because like all Ponzi Schemes it only con­tin­ues to work so long as new entrants out­weigh those try­ing to cash out.

Those who are try­ing to cash out—existing house own­ers who are sell­ing as spec­u­la­tors, or sell­ing to real­ize a paper cap­i­tal gain and upgrade to a more expen­sive house, or sell­ing an invest­ment prop­er­ty to fund their retirement—are now sell­ing into a dwin­dling mar­ket.

The first effect of this imbal­ance between demand and sup­ply is an increase in the time to sell, and in the num­ber of unsold prop­er­ties on the mar­ket. The sec­ond effect is a mod­er­ate fall in prices, once sell­ers who have to sell real­ize that they have to take a hair­cut. The third effect in Aus­tralia may well be an increase in sales by prop­er­ty spec­u­la­tors, if they see their cap­i­tal gains dimin­ish­ing the longer they hold on to their “invest­ments”.

The Scheme could be kept alive by a reduc­tion in inter­est rates to entice new buy­ers into the market—Australia’s float­ing rate mort­gages make it much eas­i­er for the Cen­tral Bank to manip­u­late mort­gage rates here than in the USA—but even there, there’s a lim­it. To get mort­gage pay­ments back to 20% or less of the after-tax income of a cou­ple earn­ing the aver­age wage— with­out mort­gage lev­els falling, and hence house prices falling—the mort­gage inter­est rate would need to fall to 3%. This would require the RBA to drop its cash rate to zero from its cur­rent lev­el of 4.75 per­cent.

Even if it does do that, it will take a very long time to do so—remember that Aus­trali­a’s Cen­tral Bank was still rais­ing inter­est rates well into the GFC (it increased the cash rate to 7.25% in March 2008, and only start­ing cut­ting it in Sep­tem­ber when the cri­sis was already a year old). Mort­gages and house prices will have plen­ty of time to fall before that hap­pens.

Fig­ure 6: Aus­trali­a’s Cen­tral Bank rate is almost 5% high­er than the USA’s

This rais­es two ques­tions: how much could house prices fall, and what could be the impact of a fall on the financiers of this Ponzi Scheme: the banks?

I’ll con­sid­er the sec­ond ques­tion in a post next week; for now let’s do some­thing the “no bub­ble” crowd reg­u­lar­ly refuse to do, and con­sid­er long-term data on house prices and incomes.

Fighting Magoo-nomics with long-term data

I some­times feel like I’m fight­ing Mr Magoo when I debate the non-bub­ble set: they choose a short-term data set and then tell me that what I’m pre­dict­ing can’t hap­pen because it has nev­er hap­pened before. Yet there is long-term data to show that it has hap­pened before. They either ignore it, or find rea­sons to dis­miss it because it does­n’t meet their qual­i­ty stan­dards.

This is self-serv­ing. Old­er data will almost always not meet mod­ern stan­dards, sim­ply because it is old and, in most cas­es, sta­tis­ti­cal prac­tices have improved over time (one obvi­ous excep­tion to this is gov­ern­ment report­ing of unem­ploy­ment and infla­tion, where stan­dard have been dropped for polit­i­cal expediency—see Roy Mor­gan’s fig­ures on the actu­al unem­ploy­ment rate in Aus­tralia, and John Williams’ “Shad­ow­stats” infor­ma­tion on actu­al unem­ploy­ment and infla­tion in the USA). But the data exists, and unless it is out by a huge mar­gin, the infor­ma­tion it con­tains is worth con­sid­er­ing.

Joye’s points above about inter­est and infla­tion are a case in point here: “inter­est rates over the past 15 years have, on aver­age, been 43 per cent low­er than inter­est rates in the 15 years that pre­ced­ed that peri­od… aver­age infla­tion since the mid­dle of the 1990s has been 55 per cent low­er than infla­tion in the 15 years pri­or”. That’s all true—but if you look back fur­ther in time, inter­est rates and infla­tion were low­er in the 1960s than they are today. In the 1970s, though inter­est rates were high­er than today, they were low­er than the rate of infla­tion, so that the real inter­est rate was neg­a­tive.

Time Peri­od Nom­i­nal Mort­gage Rate Infla­tion Rate Real Mort­gage Rate
1960–70 5.37 2.43 2.94
1970–80 8.62 9.76 -1.13
1980–95 12.71 6.69 6.02
1995–2011 7.55 2.78 4.77

If Joye and Blox­ham’s “struc­tur­al changes mean this time is dif­fer­ent” case was valid, then mort­gages and house prices should have been high­er rel­a­tive to incomes in the 1960s than today (let along the 1970s!) because inter­est rates and infla­tion were much low­er then than now.

And were they? Here we have to do some detec­tive work, to com­bine the very brief ABS time series on house prices (which starts in 2002) with longer term house price esti­mates for Syd­ney and Mel­bourne put togeth­er by Nigel Sta­ple­don of UNSW (which starts in 1880 and ends in 2005). Though put togeth­er with very dif­fer­ent method­olo­gies, the over­lap is good for the 3 years they share in common—especially for Mel­bourne.

Fig­ure 7: Two meth­ods for esti­mat­ing house prices with com­pa­ra­ble results

It’s also pos­si­ble to derive an implied ABS medi­an house price for Syd­ney and Mel­bourne by com­bin­ing the ABS’s medi­an house price index data—which goes back to 1986—with its price data from 2002 on. Sta­ple­don’s data also fits this series very well—again, espe­cial­ly for Mel­bourne.

Fig­ure 8: They’re also con­sis­tent over the last 25 years when com­bined with ABS Index data

Giv­en this close cor­re­spon­dence, I’m will­ing to use Sta­ple­don’s data as a rea­son­able guide to what medi­an house prices were before the ABS began col­lect­ing house price index data.

Fig­ure 9: Esti­mat­ed medi­an prices for Syd­ney and Mel­bourne using Sta­ple­don’s data till 1986 and ABS after­wards

Now we can com­bine this data with ABS and RBA data on dis­pos­able incomes, pop­u­la­tion and the num­ber of dwellings to see how the ratio of house prices to dis­pos­able income has fared over time.

Fig­ure 10: A tripling of house prices com­pared to incomes over the last 50 years

There are var­i­ous prob­lems with this com­par­i­son:

  • It com­pares medi­an house prices to aver­age incomes, and there­fore under­states the medi­an to medi­an (or aver­age to aver­age) com­par­i­son by about 25 per­cent;
  • The ABS Index only cov­ers free-stand­ing hous­es, thus over­stat­ing (prob­a­bly also by about 25 per­cent) the medi­an price lev­el by omit­ting cheap­er apart­ments;
  • It does­n’t account for dif­fer­ences in aver­age dis­pos­able incomes by city, thus over­stat­ing the ratio for Syd­ney and Mel­bourne, but under­stat­ing it for the oth­er cities.

But over­all it’s a rea­son­able guide to some­thing we des­per­ate­ly need more infor­ma­tion on, and the over-time com­par­isons are valid. An aver­age-income house­hold could have pur­chased the medi­an house in Syd­ney with less than 2 years of dis­pos­able income in 1960; it now takes over 6 years—and at the peak, it took 8 years.

What’s more, the ser­vic­ing cost of this debt was low­er in the 1960s than it has been for the last decade, because mort­gage rates were 30% low­er back then.

So much for Stevens’ claim that “the price to income for Aus­tralia … is about four and half and it has not moved much either way for ten years”. The myopic focus of “no bub­ble” com­men­ta­tors on the last 10 years of data ignores a bub­ble that, since 1985, has dou­bled the rel­a­tive cost of buy­ing a house. Since the ear­ly 1960s, when the old­est Baby Boomers were buy­ing their first prop­er­ties, it has tripled the cost.

To restore the house price to income ratio that applied in 1985, before this bub­ble real­ly took off, house prices would have to fall by 50 per­cent com­pared to incomes.

The final refuge of bub­ble deniers is a claim that I’ve heard much less of in recent years—after the US Bub­ble clear­ly burst in 2006—but which is still worth address­ing: that house prices always rise faster than con­sumer prices over the long term. The best empir­i­cal retort to this is the price index com­piled for Ams­ter­dam’s most expen­sive canal from 1628—just before the Tulip Craze began—till 1973. There were lengthy peri­ods where prices gen­er­al­ly went down in real terms, and equal­ly lengthy peri­ods where they went up. It was pos­si­ble to be born when a long term slump began, and die at a mature age believ­ing that house prices always fall; and dit­to for believ­ing, from your own expe­ri­ence, that they always rise. But over the very long term, there is no trend.

Fig­ure 11: Ams­ter­dam prices; booms and bust over 350 years, but no trend

Driving Miss Bubble

There were two main dri­vers of this bub­ble: a finan­cial sec­tor that makes mon­ey by cre­at­ing debt, and a gov­ern­ment sec­tor that has (to some extent unwit­ting­ly) used asset price manip­u­la­tion as a cheap means to stim­u­late the econ­o­my.

The impact of the gov­ern­ment is obvi­ous when you over­lay the First Home Own­ers Grant over the house price to income data.

Fig­ure 12: The FHOG lifts prices and stokes the debt-dri­ven econ­o­my

Sta­tis­ti­cal­ly, its impact sticks out like a sore thumb as well. Between 1951 and when the FHOG was first intro­duced (in 1983), the aver­age quar­ter­ly change in real house prices was 0.07 percent—or effec­tive­ly zero. After it, the aver­age quar­ter­ly change was just shy of 1%. When the Scheme was in oper­a­tion (it was not in oper­a­tion dur­ing the 1990s), the rise was 2% per quar­ter; on the two occa­sions when the Grant was dou­bled, real house prices rose by 3 per­cent per quar­ter.

Nor­mal Stats Before FHOS After FHOS All Data Dur­ing FHOS Between FHOS peri­ods When FHOS dou­bled
Mean 0.07% 0.95% 0.47% 2.17% 0.26% 3.10%
Min -5.53% -3.73% -5.53% -2.26% -2.26% -0.92%
Max 3.91% 7.86% 7.86% 7.86% 2.95% 4.93%
Std. Dev. 1.73% 2.17% 1.99% 2.71% 1.27% 1.83%
Count 131 108 240 25 50 7

On all but one occa­sion, the Grant was used as a macro­eco­nom­ic tool—a cheap way of boost­ing the econ­o­my dur­ing a down­turn, whether actu­al or feared (the one oth­er time—when Howard revived it in 2000—it was as a “tem­po­rary” sup­port to the build­ing indus­try when the GST was intro­duced; that tem­po­rary sup­port has now last­ed almost 12 years).

The Grant works because the rel­a­tive­ly small gov­ern­ment grant is lev­ered not once, but at least twice. First­ly the First Home Buy­er’s bor­row­ing capac­i­ty is boost­ed by the lender’s Loan to Val­u­a­tion Ratio—so $7,000 to the bor­row­ers becomes some­thing north of $50,000 for the ven­dors with today’s sky-high LVRs. Then the ven­dors use the addi­tion­al cash they received as increased deposits for their next pur­chase.

The banks are hap­py to fund this process, because they make mon­ey by cre­at­ing debt, and are there­fore always look­ing for avenues by which it can be cre­at­ed. When bor­row­ing is based upon expect­ed future income, or even aimed at fund­ing con­sump­tion today, cre­at­ing addi­tion­al debt is hard. But if bor­row­ers can be per­suad­ed that there’s mon­ey to be made by bor­row­ing mon­ey and spec­u­lat­ing on asset prices, there is—for a while—an easy means to cre­ate more debt.

Ever since 1990, that’s been the secret to both the house bub­ble and the prof­itabil­i­ty of Aus­tralian banks. They’ve made their mon­ey by financ­ing Aus­trali­a’s prop­er­ty bub­ble; they start­ed to do so the moment the pre­vi­ous spec­u­la­tive bubble—the one that gave us Alan Bond and Christo­pher Skase—died out; and, though the spin is that the USA had irre­spon­si­ble lend­ing while Aus­trali­a’s lenders were pru­dent, mort­gage debt grew three times more rapid­ly in Aus­tralia than in the USA, and reached a peak 18 per­cent high­er than the USA’s.

Fig­ure 13: Aus­tralian banks have been respon­si­ble lenders? Com­pared to whom?

The growth of this debt is what real­ly drove house prices high­er, and now that our mort­gage debt to GDP ratio is start­ing to turn, so too are our house prices—just as in the USA.

Fig­ure 14: Ris­ing debt drove the US bub­ble up, and slow­ing debt caused it to burst

The only thing that delayed this process in Aus­tralia was the last gasp of the First Home Ven­dors Scheme under Rudd, which turned a nascent decline in Aus­trali­a’s mort­gage debt to GDP ratio into a final fling of the debt bub­ble. Had the trend con­tin­ued, the mort­gage debt to GDP ratio would have fall­en about 2 per­cent. Instead it rose over 6 per­cent, inject­ing about $100 bil­lion of addi­tion­al debt-financed spend­ing into the Aus­tralian econ­o­my. It was a major fac­tor in Aus­trali­a’s appar­ent­ly good per­for­mance dur­ing the finan­cial cri­sis, but as one my blog­gers remarked, it worked by “kick­ing the can down the road”.

Fig­ure 15: The same dynam­ic is play­ing out in Aus­tralia, though delayed by the FHVB

We all know what hap­pened to the US finance sec­tor after the US house price bub­ble burst. In the next post I’ll con­sid­er what could hap­pen to Aus­tralian banks as our bub­ble ends.

Bookmark the permalink.

About Steve Keen

I am Professor of Economics and Head of Economics, History and Politics at Kingston University London, and a long time critic of conventional economic thought. As well as attacking mainstream thought in Debunking Economics, I am also developing an alternative dynamic approach to economic modelling. The key issue I am tackling here is the prospect for a debt-deflation on the back of the enormous private debts accumulated globally, and our very low rate of inflation.