This Time Had Better Be Different: House Prices and the Banks Part 2

Flattr this!

Click here for this post in PDF

Fig­ure 1

In last week’s post I showed that there is a debt-financed, gov­ern­ment-spon­sored bub­ble in Aus­tralian house prices (click here and here for ear­li­er install­ments on the same top­ic). This week I’ll con­sid­er what the burst­ing of this bub­ble could mean for the banks that have financed it.

Betting the House

For two decades after the 1987 Stock Mar­ket Crash, banks have lived by the adage “as safe as hous­es”. Mort­gage lend­ing sur­passed busi­ness blend­ing in 1993, and ever since then it’s been on the up and up. Busi­ness lend­ing actu­al­ly fell dur­ing the 1990s reces­sion, and took off again only in 2006, when the Chi­na boom and the lever­aged-buy­out fren­zy began.

Fig­ure 2

Reg­u­lar read­ers will know that I place the respon­si­bil­i­ty for this increase in debt on the finan­cial sec­tor itself, not the bor­row­ers. The bank­ing sec­tor makes mon­ey by cre­at­ing debt and thus has an inher­ent desire to pump out as much as pos­si­ble. The eas­i­est way to do this is to entice the pub­lic into Ponzi Schemes, because then bor­row­ing can be de-cou­pled from income.

There’s a minor ver­i­fi­ca­tion of my per­spec­tive in this data, since the one seg­ment of debt that has­n’t risen com­pared to GDP is per­son­al debt—where the income of the bor­row­er is a seri­ous con­straint on how much debt the bor­row­er will take on. As much as banks have flogged cred­it cards, per­son­al debt has­n’t increased as a per­cent­age of GDP.

On the oth­er hand, mort­gage debt has risen sev­en­fold (com­pared to GDP) in the last two decades.

Fig­ure 3

The post-GFC peri­od in Aus­tralia has seen a fur­ther increase in the bank­ing sec­tor’s reliance on home loans—due to both the busi­ness sec­tor’s heavy delever­ag­ing in the wake of the cri­sis, and the gov­ern­men­t’s re-ignit­ing of the house price bub­ble via the First Home Ven­dors Boost in late 2008. Mort­gages now account for over 57 per­cent of the banks’ loan books, an all-time high.

Fig­ure 4

They also account for over 37% of total bank assets—again an all-time high, and up sub­stan­tial­ly from the GFC-induced low of 28.5% before the First Home Ven­dors Boost reversed the fall in mort­gage debt.

Fig­ure 5

So how exposed are the banks to a fall in house prices, and the increase in non-per­form­ing loans that could arise from this? There is no way of know­ing for sure before­hand, but cross-coun­try com­par­isons and his­to­ry can give a guide.

Bigger than Texas

A per­sis­tent refrain from the “no bub­ble” camp has been that Aus­tralia won’t suf­fer any­thing like a US down­turn from a house price crash, because Aus­tralian lend­ing has been much more respon­si­ble than Amer­i­can lend­ing was. I took a swipe at that in last week’s post, with a chart show­ing that Aus­trali­a’s mort­gage debt to GDP ratio exceeds the USA’s, and grew three times more rapid­ly than did Amer­i­can mort­gage debt since 1990 (see Fig­ure 13 of that post).

Sim­i­lar data, this time seen from the point of view of bank assets, is shown in the next two charts. Real estate loans are a high­er pro­por­tion of Aus­tralian bank loans than for US banks, and their rise in sig­nif­i­cance in Aus­tralia was far faster and sharp­er than for the USA.

Fig­ure 6

More sig­nif­i­cant­ly, real estate loans are a high­er pro­por­tion of bank assets in Aus­tralia than in the USA, and this applied through­out the Sub­prime Era in the USA. The cru­cial role of the First Home Ven­dors Boost in revers­ing the fall in the banks’ depen­dence on real estate loans is also strik­ing­ly appar­ent.

Fig­ure 7

Never mind the weight, feel the distribution

The “no bub­ble” case dis­miss­es this Aus­tralia-US com­par­i­son on two grounds:

  • most of Aus­trali­a’s hous­ing loans are to wealth­i­er house­holds, who are there­fore more like­ly to be able to ser­vice the debts so long as they remain employed; and
  • hous­ing loans here are full-recourse, so that home own­ers put pay­ing the mort­gage ahead of all oth­er con­sid­er­a­tions..

Blox­ham made the for­mer claim in his recent piece:

How­ev­er, there are oth­er rea­sons why lev­els of house­hold debt should not be a large con­cern. The key one is that 75 per cent of all house­hold debt in Aus­tralia is held by the top two-fifths of income earn­ers. (Paul Blox­ham , The Aus­tralian hous­ing bub­ble fur­phy, Busi­ness Spec­ta­tor March 18 2011)

Alan Kohler recount­ed an inter­est­ing con­ver­sa­tion with “one of Aus­trali­a’s top retail bankers” a cou­ple of years ago on the lat­ter point:

There is some ‘mort­gage stress’ in the north­ern sub­urbs of Mel­bourne, the west­ern sub­urbs of Syd­ney and some parts of Bris­bane, but while all the banks are brac­ing them­selves for it and increas­ing gen­er­al pro­vi­sions, there is no sign yet of the defaults that are bring­ing the US bank­ing sys­tem to its knees.

We often see graphs show­ing that Aus­trali­a’s ratios of house­hold debt to GDP and debt to house­hold income had gone up more than in the Unit­ed States. So, while the US is deep into a mort­gage-based finan­cial cri­sis, it is sure­ly a cause for cel­e­bra­tion that Aus­tralia has not seen even the slight­est uptick in arrears.

Please explain,” I said to my din­ner com­pan­ion. Obvi­ous­ly, low unem­ploy­ment and robust nation­al income, includ­ing strong retail sales until recent­ly, have been the most impor­tant part of it. But on the oth­er hand, the US econ­o­my was doing okay until the mort­gage bust hap­pened; it was the sub-prime cri­sis that bust­ed the US econ­o­my, not the oth­er way around.

Apart from that it is down to two things, he says: with­in the banks, “sales” did not gain ascen­dan­cy over “cred­it” in Aus­tralia to the extent that it did in the US; and US mort­gages are non-recourse where­as banks in Aus­tralia can have full recourse to the bor­row­ers’ oth­er assets, which means bor­row­ers are less inclined to just walk away. (Alan Kohler, “Healthy by default”, Busi­ness Spec­ta­tor August 21, 2008; emphases added)

Kris Sayce gave a good come­back to Blox­ham’s “most of the debt is held by those who can afford it” line when he not­ed that “two-fifths of income earn­ers is quite a large pool of peo­ple”:

In fact, it’s near­ly half the income earn­ers. Is that num­ber any dif­fer­ent to any oth­er econ­o­my? You’d nat­u­ral­ly think the high­er income earn­ers would have most of the debt because they’re the ones more like­ly to want it, need it or be offered it.

So with about 11.4 mil­lion Aus­tralians employed, that makes for about 4.6 mil­lion Aus­tralians hold­ing over $1.125 tril­lion of house­hold debt – remem­ber total house­hold debt is about $1.5 tril­lion. That comes to about $244,565 per per­son.

Per­haps we’re not very bright. But we’re strug­gling to see how that makes the pop­ping of the hous­ing bub­ble a “vir­tu­al impos­si­bil­i­ty.” (Kris Sayce, “Are Falling House Prices “Vir­tu­al­ly Impos­si­ble”?”, Mon­ey Morn­ing 18 March 2011)

The best come­backs to Alan Kohler’s din­ner com­pan­ion may well be time itself. Impaired assets (See Note [1]) did hit an all-time low of 4.1% of Bank Tier 1 Assets and 0.2% of total assets in Jan­u­ary 2008, but by the time Kohler and his banker sat down to din­ner, impair­ment was on the rise again. Impaired assets have since reached a plateau of 25% of Tier 1 cap­i­tal and 1.25% of total assets—and this has occurred while house prices were still ris­ing. Despite the pres­sure that full-recourse lend­ing puts on bor­row­ers, this is com­pa­ra­ble to the lev­el of impaired assets in US banks before house prices col­lapsed when the Sub­Prime Boom turned into the Sub­Prime Cri­sis (see Table 2 on page 10 of this paper).

Fig­ure 8

Since real estate loans are worth rough­ly 7 times bank Tier 1 capital—up from only 2 times in 1990—it would­n’t take much of an increase in non-per­form­ing hous­ing loans to push Aus­tralian banks to the lev­el of impair­ment expe­ri­enced by Amer­i­can banks in 2007 and 2008.


Fig­ure 9

The lev­el and impor­tance of non-recourse lend­ing in the US is also exag­ger­at­ed. While some major States have it, many do not—and one of the worst per­form­ing states in and since the Sub­prime Cri­sis was Flori­da, which has full recourse lend­ing.

Final­ly, the “nev­er mind the weight, feel the dis­tri­b­u­tion” defence of the absolute mort­gage debt lev­el has a neg­a­tive impli­ca­tion for the Aus­tralian econ­o­my: if debt is more broad­ly dis­trib­uted in Aus­tralia than in the USA, then the neg­a­tive effects of debt ser­vice on con­sump­tion lev­els are like­ly to be greater here than in Amer­i­ca. This is espe­cial­ly so since mort­gage rates today are 50% high­er here than in the USA. Inter­est pay­ments on mort­gage debt in Aus­tralia now rep­re­sent 6.7% of GDP, twice as much as in the USA. It’s lit­tle won­der that Aus­trali­a’s retail­ers are cry­ing poor.

Of course, the RBA could always reduce the debt repay­ment pres­sure by reduc­ing the cash rate. But with the mar­gin between the cash rate and mort­gages now being about 3%, it would need to reduce the cash rate to 1.5% to reduce the debt repay­ment bur­den in Aus­tralia to the same lev­el as Amer­i­ca’s.



Fig­ure 10

So if Amer­i­ca’s con­sumers are debt-con­strained in their spend­ing, Aus­tralian con­sumers are even more so—with neg­a­tive impli­ca­tions for employ­ment in the retail sec­tor.

Com­pared to the USA there­fore, there is no rea­son to expect that Aus­tralian banks will fare bet­ter from a sus­tained fall in house prices. What about the com­par­i­son with past finan­cial crises in Aus­tralia?

This time really is different

There are at least three ways in which what­ev­er might hap­pen in the near future will dif­fer from the past:

  • On the atten­u­at­ing side, deposit insur­ance, which was only implic­it or lim­it­ed in the past, is much more estab­lished now; and
  • If the banks face insol­ven­cy, the Gov­ern­ment and Reserve Bank will bail them out as the US Gov­ern­ment and Fed­er­al Reserve did—though let’s hope with­out also bail­ing out the man­age­ment, share­hold­ers and bond­hold­ers, as in the USA (OK, so call me an opti­mist! And if you haven’t seen Inside Job yet, see it);

On the neg­a­tive side, how­ev­er, we have the Big Tri­fec­ta:

  • The bub­bles in debt, hous­ing and bank stocks are far big­ger this time than any pre­vi­ous event—including the Mel­bourne Land Boom and Bust that trig­gered the 1890s Depres­sion.

I’ll make some sta­tis­ti­cal com­par­isons over the very long term, but the main focus here is on sev­er­al peri­ods when house prices fell sub­stan­tial­ly in real terms after a pre­ced­ing boom, and what hap­pened to bank shares when house prices fell:

  • The 1880s-1890s, when the Mel­bourne Land Boom bust­ed and caused the 1890s Depres­sion;
  • The 1920s till ear­ly 1930s, when the Roar­ing Twen­ties gave way to the Great Depres­sion;
  • The ear­ly to mid-1970s, when a spec­u­la­tive bub­ble in Syd­ney real estate caused a rapid accel­er­a­tion in pri­vate debt, and a tem­po­rary fall in pri­vate debt com­pared to GDP due to ram­pant infla­tion;
  • The late 1980s to ear­ly 1990s, when the Stock Mar­ket Crash was fol­lowed by a spec­u­la­tive bub­ble in real estate—stoked by the sec­ond incar­na­tion of the First Home Ven­dors Boost; and
  • From 1997 till now.

I chose the first four peri­ods for two rea­sons: they were times when house prices fell in real (and on the first two occa­sions, also nom­i­nal) terms, and bank share prices suf­fered a sub­stan­tial fall; and they also stand out as peri­ods when an accel­er­a­tion in debt caused a boom that gave way to a delever­age-dri­ven slump, when pri­vate debt reached either a long term or short term peak (com­pared to GDP) and fell after­wards. They are obvi­ous in the graph of Aus­trali­a’s long term pri­vate debt to GDP ratio.

Fig­ure 11

They also turn up as sig­nif­i­cant spikes in the Cred­it Impulse (Big­gs, May­er et al. 2010)—the accel­er­a­tion of debt (divid­ed by GDP) which deter­mines the con­tri­bu­tion that debt makes to changes in aggre­gate demand (See Note [2]).

The Cred­it Impulse data also lets us dis­tin­guish the pre-WWII more lais­sez-faire peri­od from the “reg­u­lat­ed” one that fol­lowed it: cred­it was much more volatile in the pre-WWII peri­od, but the trend val­ue of the Cred­it Impulse was only slight­ly above zero at 0.1%.

Fig­ure 12

The Post-WWII peri­od had much less volatil­i­ty in the debt-financed com­po­nent of changes in aggre­gate demand, but the over­all trend was far high­er at 0.6%. This could be part of the expla­na­tion as to why Post-WWII eco­nom­ic per­for­mance has been less volatile than pre-WWII, but it also indi­cates that ris­ing debt has played more of a role in dri­ving demand in the post-War peri­od than before.

Omi­nous­ly too, even though the post-WWII peri­od in gen­er­al has been less volatile, the neg­a­tive impact of the Cred­it Impulse in this down­turn was far greater than in either the 1890s or the 1930s.

Fig­ure 13

One final fac­tor that also sep­a­rates the pre-WWII data from post-WWII is the rate of infla­tion. The 1890s and 1930s debt bub­bles burst at a time of low infla­tion, and rapid­ly gave way to defla­tion. This actu­al­ly drove the debt ratio high­er in the first instance, as the fall in prices exceed­ed the fall in debt. But ulti­mate­ly those debts were reduced in a time of low infla­tion.

The 1970s episode, on the oth­er hand, was char­ac­ter­ized by ram­pant inflation—and the debt ratio fell because ris­ing prices reduced the effec­tive debt bur­den. Where­as the falls in real house prices in the 1890s and the 1930s there­fore meant that nom­i­nal prices were falling even faster, the 1970s fall in real house prices main­ly reflect­ed con­sumer price infla­tion out­strip­ping house price growth. The 1990 bub­ble also burst when infla­tion was still sub­stan­tial, though far low­er than it was in the mid-1970s.

Today’s infla­tion sto­ry has more in com­mon with the pre-WWII world than the 1970s. Our cur­rent bub­ble is burst­ing in a low-infla­tion envi­ron­ment.


Fig­ure 14

Now let’s see what his­to­ry tells us about the impact of falling house prices on bank shares.

The 1880s-1890s

This was the bank bust to end all bank busts—just like WWI was the War to end all wars. Bank shares increased by over 75% in real terms as spec­u­la­tive lend­ing financed a land bub­ble in Mel­bourne that increased real house prices by 33% (Stapledon’s index com­bines Syd­ney and Mel­bourne, so this fig­ure under­states the degree of rise and fall in Mel­bourne prices). The role of debt in dri­ving this bub­ble and the sub­se­quent Depres­sion is unmis­tak­able: pri­vate debt rose from under 30% of GDP in 1872 to over 100% in 1892, and then unwound over the next 3 decades to a low of 40% in 1925.

The turn­around in debt and the col­lapse in house prices pre­cip­i­tat­ed a 50% fall in bank shares in less than six months as house prices start­ed to fall back to below the pre-boom lev­el.

Fig­ure 15

The excel­lent RBA Research paper “Two Depres­sions, One Bank­ing Col­lapse” by Chay Fish­er & Christo­pher Kent (RDP1999-06) argues fair­ly con­vinc­ing­ly that the 1890s Depres­sion was a more severe Depres­sion for Aus­tralia than the Great one—mainly because there were more bank fail­ures in the 1890s than in the 1930s. The sever­i­ty of the 1890s fall in bank shares may relate to the high­er lev­el of debt in 1890 than in the 1930s—a peak of 104 per­cent of GDP in 1892 ver­sus only 76 per­cent in 1932 (the peak this time round was 157 per­cent in March 2008).

The cor­re­la­tion of the two series in absolute terms is obvi­ous (the cor­re­la­tion coef­fi­cient is 0.8), and the changes in the two series are also strong­ly cor­re­lat­ed (0.42).


Fig­ure 16

The 1920s-1940s

The 1920s began with the end of the great delever­ag­ing that had com­menced in 1892. Real house prices rose by about 25 per­cent in the first two years—though main­ly because of defla­tion in con­sumer prices—and then fluc­tu­at­ed down for the next four years before a minor boom. But the main debt-financed bub­ble in the 1920s was in the Stock Mar­ket.

Fig­ure 17

There was how­ev­er still a crash in bank shares after house prices turned south in ear­ly 1929. It was not as severe as in 1893, and of course coin­cid­ed with a col­lapse in the gen­er­al stock mar­ket (I can’t give com­pa­ra­ble fig­ures because of the dif­fer­ent meth­ods used to com­pile the two indices—see the Appen­dix). But still there was a fall of 24% in bank shares over 7 months at its steep­est, and a 39% fall from peak to trough—preceded by a 25% fall in house prices.

Fig­ure 18

Bank shares also tracked house prices over the 20 years from the Roar­ing Twen­ties boom to the begin­ning of WWII: the cor­re­la­tion was 0.44 for the indices, and 0.47 for the change in the indices.

Fig­ure 19

The 1970s

The 1970s bub­ble was the last gasp of the long peri­od of robust yet tran­quil growth that had char­ac­ter­ized the ear­ly post-WWII peri­od. The pecu­liar macro­eco­nom­ics of the time—the start of “Stagflation”—clouds the house price bub­ble pic­ture some­what (I dis­cuss this in the Appen­dix), but there still was a big house price bub­ble then, and a big hit to bank shares when it end­ed.

This was Aus­trali­a’s first real­ly big debt-financed spec­u­la­tive bub­ble, which most com­men­ta­tors and econ­o­mists seem to have for­got­ten entire­ly. Its fla­vor is well cap­tured in the intro­duc­tion to Syd­ney Boom, Syd­ney Bust:

Syd­ney had nev­er expe­ri­enced a prop­er­ty boom on the scale of that between 1968 and 1974. It involved a fren­zy of buy­ing, sell­ing and build­ing which reshaped the cen­tral busi­ness dis­trict, great­ly increased the sup­ply of indus­tri­al and retail­ing space, and accel­er­at­ed the expan­sion of the city’s fringe. Its vis­i­ble lega­cy of emp­ty offices and stunt­ed sub­di­vi­sions was matched by a host of finan­cial casu­al­ties which incor­po­rat­ed an unknown, but very large, con­tin­gent of small investors, togeth­er with the spec­tac­u­lar demise of a num­ber of devel­op­ment and con­struc­tion com­pa­nies and finan­cial insti­tu­tions. The boom was the most sig­nif­i­cant finan­cial hap­pen­ing of the 1970s and the shock waves from the inevitable crash were felt right up to 1980. It was an extra­or­di­nary event for Syd­ney, and for Australia.(Daly 1982, p. 1)

House prices rose 40 per­cent in real terms from 1967 till 1974, and then fell 16 per­cent from 1974 till 1980. Bank shares went through a roller-coast­er ride, fol­low­ing Posei­don up and down from 1967 till 1970, and then ris­ing sharply as the debt-bub­ble took off in 1972, with a 31 per­cent rise between late 1972 and ear­ly 1973. But from there it was all down­hill, with bank shares falling 35 per­cent across 1973 while house prices were still ris­ing.

But when house prices start­ed to fall, bank shares real­ly tanked, falling 54 per­cent in just sev­en month dur­ing 1974.

Fig­ure 20

How­ev­er, the extreme volatil­i­ty of both asset and com­mod­i­ty prices, and the impact of two share bub­bles and busts—the Posei­don Bub­ble of the late 1960s and the ear­ly 1970s boom and bust—eliminated the cor­re­la­tion of bank share prices to house prices that applied in the 1890s and 1930s: the cor­re­la­tion of the indices was ‑0.46 and of changes in the indices was ‑0.01.

Fig­ure 21

The 1980s

The reces­sion we had to have” remains unfor­get­table. That plunge began with Aus­trali­a’s sec­ond big post-WWII spec­u­la­tive bub­ble, as Bond, Skase, Con­nell and a seem­ing­ly lim­it­less cast of white-shoe brigaders estab­lished the local Ivan Boesky “Greed is Good” church—with banks eager­ly throw­ing mon­ey and debt into its tithing box.

It would have all end­ed with the Stock Mar­ket Crash of 1987, were it not for the gov­ern­ment res­cues (both here and in the USA) that enabled the spec­u­la­tors and the banks to regroup and throw their paper weight into real estate.

Fig­ure 22

Hav­ing plunged 30 per­cent in one month (Octo­ber, of course), bank shares rock­et­ed up again, climb­ing a stag­ger­ing 54 per­cent in 11 months to reach a new peak in Octo­ber 1988, as spec­u­la­tors and the sec­ond incar­na­tion of the First Home Ven­dors Grant drove house prices up 37 per­cent over just one and a half years. Bank shares bounced around for a while, but once the decline in house prices set in, bank shares again tanked—falling 40 per­cent over 11 months in 1990.

Fig­ure 23

The pos­i­tive cor­re­la­tions between the indices and their rates of change which had been swamped by the high infla­tion of the ear­ly 1970s returned: the cor­re­la­tion of the indices was 0.45 and the cor­re­la­tion of their rates of change was 0.42.

Fig­ure 24

Which brings us to today.

From 1997 till today

I have argued else­where that the cur­rent bub­ble began in 1997, but the debt-finance that final­ly set it off began far earlier—in 1990. The fact that unem­ploy­ment was explod­ing from under 6 per­cent in ear­ly 1990 to almost 11 per­cent in ear­ly 1994 was not, it seems, a rea­son to be restrained in lend­ing to the house­hold sec­tor. It was far more impor­tant to expand the mar­ket­ing of debt, and since the busi­ness sec­tor could no longer be per­suad­ed to take more on, the vir­gin field of the house­hold sec­tor had to be explored. Mort­gage debt, which had flat­lined at about 16 per­cent of GDP since records were first kept, took off, increas­ing by 50 per­cent dur­ing the 1990s reces­sion (from 1990 till the start of 1994), and ulti­mate­ly ris­ing by 360 per­cent over the two decades—from 19 per­cent of GDP to 88 percent—with the final fling of the First Home Ven­dors Boost giv­ing it that final push into the stratos­phere.

Fig­ure 25

By 1997 the sheer pres­sure of ris­ing mort­gage finance brought to an end a peri­od of flatlin­ing house prices, and the bub­bles in both house prices and bank shares took off in earnest.

The rise in bank shares far out­weighed the increase in the over­all share index (the two indices are now com­pa­ra­ble, where­as for the longer series they were com­piled in dif­fer­ent ways). Bank shares rose 230 per­cent from 1997 till their peak in 2007, ver­sus a rise of only 110 per­cent in the over­all mar­ket index.

The increase in house prices also dwarfed any pre­vi­ous bub­ble: an increase of over 120 per­cent over fif­teen years.

Bank shares and house prices both tanked when the GFC hit: house prices fell 9 per­cent and bank shares fell 61 per­cent. But thank­ful­ly the cav­al­ry rode to the rescue—in the shape of the First Home Ven­dors Boost—and both house prices and bank shares took off again. House prices rose 17 per­cent while bank shares rose 60 per­cent (ver­sus a 45 per­cent rise in the mar­ket) before falling 12 per­cent after the expiry of the FHVB.

Fig­ure 26

The cor­re­la­tion between bank shares and house prices is again pos­i­tive: 0.51 for the indices and a low 0.1 for the change in indices over the whole peri­od, but 0.46 since 2005.

Fig­ure 27

So now we are on the edge of the burst­ing anoth­er house price bub­ble. What could the future bring?

When the bubble pops…

There are sev­er­al con­sis­tent pat­terns that can be seen in the past data.

First­ly, house prices and bank shares are cor­re­lat­ed. There was one aberration—the 1970s—but that was marked by pecu­liar dynam­ics aris­ing from the his­tor­i­cal­ly high infla­tion at the time. Gen­er­al­ly, bank shares go up when house prices rise, and fall when the fall. Part­ly, this is the gen­er­al cor­re­la­tion of asset prices with each oth­er, but part­ly also it’s the causal rela­tion­ship between bank lend­ing, house prices, and bank prof­its: banks make mon­ey by cre­at­ing debt, ris­ing mort­gage debt caus­es house prices to rise, and ris­ing house prices set off the Ponzi Scheme that encour­ages more mort­gage bor­row­ing. The bub­ble bursts when the entry price to the Ponzi Scheme becomes pro­hib­i­tive, or when ear­ly entrants try to take their prof­its and run.

Sec­ond­ly, the fall in the bank share price is nor­mal­ly very steep, and it occurs short­ly after house prices have passed their peaks. Hold­ing bank shares when house prices are falling is a good way to lose money—and con­verse­ly, if you get the tim­ing right, bet­ting against them can be prof­itable. That’s why Jere­my Grantham—and many oth­er hedge fund man­agers from around the world—are pay­ing close atten­tion to Aus­tralian house prices.

Third­ly, house prices and bank shares are dri­ven by ris­ing debt, and when debt starts to fall, not only do house prices and bank shares fall, the econ­o­my also nor­mal­ly falls into a very deep reces­sion or Depres­sion. This is the cru­cial role of delever­ag­ing in caus­ing eco­nom­ic down­turns, includ­ing the seri­ous ones where debt falls not just dur­ing a short cycle pri­or to anoth­er upward trend, but in an extend­ed sec­u­lar decline.

There is also one cau­tion­ary note about the cur­rent bub­ble: though his­to­ry would imply that there is a very large down­side to bank shares now, it’s also obvi­ous that bank shares fell a great deal in 2007-09, so that much of the down­side may already have been fac­tored in.

How­ev­er, on every met­ric: on the ratio of debt to GDP, on how much that ratio rose from the start of the bub­ble to its end, on how big the house price bub­ble was, and on how much bank shares rose, this bub­ble dwarfs them all.

Debt to GDP

The 1997 debt to GDP ratio start­ed high­er than all but the 1890s bub­ble end­ed, and the bub­ble went on long after all the oth­ers had popped.

Fig­ure 28

Though the actu­al debt to GDP ratio today dwarfs all its pre­de­ces­sors, in terms of the growth of debt from the begin­ning of the bub­ble, it has one rival: the 1920s.

Fig­ure 29

How­ev­er this is part­ly because of defla­tion dur­ing the ear­ly Great Depres­sion: defla­tion ruled from 1930 till 1934, and the debt to GDP ratio rose not because of ris­ing debt, but falling prices. Though the increase in debt in the final throes of the Roar­ing Twen­ties was faster than we expe­ri­enced, over the whole boom debt grew as quick­ly now as then, and it has kept grow­ing for four years longer than in the 1920s. Even though the ratio is falling now, it’s because debt is now ris­ing more slow­ly than nom­i­nal GDP: we still haven’t expe­ri­enced delever­ag­ing yet (unlike the USA).

House Prices

The rise in prices dur­ing this bub­ble again has no equal in the his­tor­i­cal record.

Fig­ure 30

Bank Shares

Bank shares are also in a class of their own in this bub­ble, even after the sharp fall from 2007 till 2009. In terms of how high bank share prices climbed, this bub­ble tow­ers over all that have gone before, and even what is left of this bub­ble is still only matched by the biggest of the pre­ced­ing bub­bles, the 1890s and the 1970s.

Fig­ure 31

It’s a long way from the top if you’ve sold your soul

Bank lend­ing drove house prices sky high, and the prof­its banks made from this Ponzi Scheme dragged their share prices up with the bub­ble (and hand­some­ly lined the pock­ets of their man­agers).

It’s great fun while it lasts, but all Ponzi Schemes end for the sim­ple rea­son that they must: they aren’t “mak­ing mon­ey”, but sim­ply shuf­fling it—and grow­ing debt. When new entrants can’t be enticed to join the game, the shuf­fling stops and the Scheme col­laps­es under the weight of accu­mu­lat­ed debt. There are very good odds that, when this Ponzi Scheme col­laps­es and house prices fall, bank shares will go down with them.



Between 1954 and 1974, unem­ploy­ment aver­aged 1.9 per­cent, and it only once exceed­ed 3 per­cent (in 1961, when a gov­ern­ment-ini­ti­at­ed cred­it squeeze caused a reces­sion that almost result­ed in the defeat of Aus­trali­a’s then Lib­er­al gov­ern­ment, which ruled from 1949 till 1972). Infla­tion from 1954 till 1973 aver­aged 3 per­cent, and then rose dra­mat­i­cal­ly between 1973 and 1974 as unem­ploy­ment fell.

This fit­ted the belief of con­ven­tion­al “Key­ne­sian” econ­o­mists of the time that there was a trade-off between infla­tion and unem­ploy­ment: one cost of a low­er unem­ploy­ment rate, they argued, was a high­er rate of infla­tion.

But then the so-called “stagfla­tion­ary” break­down occurred: unem­ploy­ment and infla­tion both rose in 1974. Neo­clas­si­cal econ­o­mists blamed this on “Key­ne­sian” eco­nom­ic pol­i­cy, which they argued caused peo­ple’s expec­ta­tions of infla­tion to rise—thus result­ing in demands for high­er wages—and OPEC’s oil price hike.

Fig­ure 32

The lat­ter argu­ment is eas­i­ly refut­ed by check­ing the data: infla­tion took off well before OPEC’s price hike.

Fig­ure 33

The for­mer has some cre­dence as an expla­na­tion for the take-off in the infla­tion rate—workers were fac­tor­ing in both the bar­gain­ing pow­er of low unem­ploy­ment and a lagged response to ris­ing infla­tion into their wage demands.

Fig­ure 34

The Neo­clas­si­cal expla­na­tion for why this rise in infla­tion also coin­cid­ed with ris­ing unem­ploy­ment was “Key­ne­sian” pol­i­cy had kept unem­ploy­ment below its “Nat­ur­al” rate, and it was mere­ly return­ing to this lev­el. This was plau­si­ble enough to swing the pol­i­cy pen­du­lum towards Neo­clas­si­cal think­ing back then, but it looks a lot less plau­si­ble with the ben­e­fit of hind­sight.

Fig­ure 35

Though infla­tion fell fair­ly rapid­ly, and unem­ploy­ment ulti­mate­ly fell after sev­er­al cycles of ris­ing unem­ploy­ment, over the entire “Neo­clas­si­cal” peri­od both infla­tion and unem­ploy­ment were high­er than they were under the “Key­ne­sian” peri­od. So rather than infla­tion going down and unem­ploy­ment going up, as neo­clas­si­cal econ­o­mists expect­ed, both rose—with unem­ploy­ment ris­ing sub­stan­tial­ly. On empir­i­cal grounds alone, the neo­clas­si­cal peri­od was a fail­ure, even before the GFC hit.

Table 1

Pol­i­cy dom­i­nance Key­ne­sian Neo­clas­si­cal
Years 1955–1976 1976-Now
Aver­age Infla­tion 4.5 5.4
Aver­age Unem­ploy­ment 2.1 7

There was a far bet­ter expla­na­tion of the 1970s expe­ri­ence lurk­ing in data ignored by neo­clas­si­cal eco­nom­ics: the lev­el and rate of growth of pri­vate debt. As you can see from Fig­ure 32, pri­vate debt, which had been con­stant (rel­a­tive to GDP) since the end of WWII, began to take off in 1964, and went through a rapid accel­er­a­tion from 1972 till 1974, before falling rapid­ly.

The debt-financed demand for con­struc­tion dur­ing that bub­ble added to the already tight labor mar­ket, and helped dri­ve wages high­er in both a clas­sic wage-price spi­ral and a his­toric increase in labor’s share of nation­al income—which has been unwound for­ev­er since.

Fig­ure 36

Infla­tion, high­er unem­ploy­ment that weak­ened labor’s bar­gain­ing pow­er, anti-union pub­lic pol­i­cy and an approach to wage-set­ting pol­i­cy that empha­sized cost of liv­ing adjust­ments but ignored shar­ing pro­duc­tiv­i­ty gains, all con­tributed to that unwind­ing.

The share market indices

The bank share index used in this post was com­piled by com­bin­ing 3 data sources. Work­ing back­wards in time, these were:

  • The S&P’s ASX 200 Finan­cials Index (AXFJ) from May 2001 till now;
  • A com­pos­ite formed from the prices for the 4 major bank share prices that match­es the val­ue of the Finan­cials Index from 2000 till May 2001; and
  • Data from the Glob­al Finan­cial Data­base from 1875 till 2000, which in turn con­sists of three series:
    • Secu­ri­ty Prices and Yields, 1875–1955,” Syd­ney Stock Exchange Offi­cial Gazette, July 14, 1958, pp 257–258 (1875–1936), togeth­er with D. McL. Lam­ber­ton, Share Price Indices in Aus­tralia, Syd­ney: Law Book Co., 1958; and
    • The Aus­tralian Stock Exchange Indices, Syd­ney: AASE, 1980; and
    • Aus­tralian Stock Exchange Lim­it­ed, ASX Indices & Yields, Syd­ney: ASX, 1995 (updat­ed till 2000)

From a perusal of the GFD doc­u­men­ta­tion and a com­par­i­son of the Bank­ing and Finance index to the broad­er mar­ket index, it appears that the bank index is a straight price index pre-1980, where­as the GFD’s data for the over­all mar­ket is an accu­mu­la­tion index till 1980 and a price index after that. These incon­sis­ten­cies make it impos­si­ble to com­pare the two over the very long term, but the move­ments in each at dif­fer­ent time peri­ods can be com­pared (and the com­par­i­son is also fine from 1980 on).

Fig­ure 37


[1] The notes to Table B05 state that “‘Impaired assets’ refers to the aggre­gate of a report­ing bank’s non-accru­al and restruc­tured expo­sures, both on- and off-bal­ance sheet, plus any assets acquired through the enforce­ment of secu­ri­ty con­di­tions. Off-bal­ance sheet expo­sures include, inter alia, com­mit­ments to pro­vide funds that can­not be can­celled or revoked and the cred­it equiv­a­lent amounts of inter­est rate, for­eign exchange and oth­er mar­ket-relat­ed instru­ments.”

[2] One of the many issues that dis­tin­guish­es my approach to eco­nom­ics from neo­clas­si­cal econ­o­mists is my focus on the role that changes in debt play in aggre­gate demand. Neo­clas­si­cal econ­o­mists wrong­ly ignore the role of aggre­gate lev­el of debt because they see debt as sim­ply a trans­fer of spend­ing pow­er from one agent to another—so that there is no change in aggre­gate spend­ing pow­er if debt ris­es.  This is the rea­son that Bernanke gave for ignor­ing Fisher’s “debt defla­tion” the­o­ry of the Great Depres­sion (Fish­er 1933):

Fish­er’s idea was less influ­en­tial in aca­d­e­m­ic cir­cles, though, because of the coun­ter­ar­gu­ment that debt-defla­tion rep­re­sent­ed no more than a redis­tri­b­u­tion from one group (debtors) to anoth­er (cred­i­tors). Absent implau­si­bly large dif­fer­ences in mar­gin­al spend­ing propen­si­ties among the groups, it was sug­gest­ed, pure redis­tri­b­u­tions should have no sig­nif­i­cant macro-eco­nom­ic effects… (Bernanke 2000, p. 24)

And it’s the explic­it assump­tion that Krug­man uses in his recent paper on the Great Reces­sion:

Ignor­ing the for­eign com­po­nent, or look­ing at the world as a whole, the over­all lev­el of debt makes no dif­fer­ence to aggre­gate net worth — one per­son­’s lia­bil­i­ty is anoth­er per­son­’s asset. (Krug­man and Eggerts­son 2010, p. 3)

This shows their igno­rance of the capac­i­ty for the bank­ing sec­tor to cre­ate spend­ing pow­er “out of noth­ing”, and thus cre­ate spend­ing pow­er in the process. I cov­er this top­ic in detail in these posts ( and


Bernanke, B. S. (2000). Essays on the Great Depres­sion. Prince­ton, Prince­ton Uni­ver­si­ty Press.

Big­gs, M., T. May­er, et al. (2010). “Cred­it and Eco­nom­ic Recov­ery: Demys­ti­fy­ing Phoenix Mir­a­cles.” SSRN eLi­brary.

Daly, M. T. (1982). Syd­ney Boom, Syd­ney Bust. Syd­ney, George Allen and Unwin.

Fish­er, I. (1933). “The Debt-Defla­tion The­o­ry of Great Depres­sions.” Econo­met­ri­ca
1(4): 337–357.

Krug­man, P. and G. B. Eggerts­son (2010). Debt, Delever­ag­ing, and the Liq­uid­i­ty Trap: A Fish­er-Min­sky-Koo approach [2nd draft 2/14/2011]. New York, Fed­er­al Reserve Bank of New York & Prince­ton Uni­ver­si­ty.

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.