1
EMEAP SENIOR OFFICIALS MEETING, KUALA LUMPUR, 18 JANUARY 2011
THE AUSTRALIAN FINANCIAL SYSTEM
Graph 2
Banks’ Non-performing Assets
Domestic books, per cent of loans by type
%
%
Business*
3
3
2
2
Personal
1
1
Total
0
2004
Housing
2006
2008
2010
0
* Includes bill acceptances and debt securities, and other non-household loans
Source: APRA
For the housing portfolio – which accounts
for over half of banks’ on-balance sheet loans – the aggregate NPA ratio has steadied, to
be 0.7 per cent in September 2010. This remains very low by international standards
(Graph 3).
Graph 3
Non-performing Housing Loans
Per cent of loans*
%
%
8
8
US**
6
6
Spain
4
4
+
UK
2
2
Australia**
Canada**+
0
1990
1994
1998
2002
2006
0
2010
* Per cent of loans by value. Includes ‘impaired’ loans unless otherwise stated.
For Australia, only includes loans 90+ days in arrears prior to September 2003.
** Banks only.
+ Per cent of loans by number that are 90+ days in arrears.
Sources: APRA; Bank of Spain; Canadian Bankers’ Association; Council of
Mortgage Lenders; FDIC; RBA
3
Household sector net worth has retraced much of the deterioration experienced over 2008
and 2009 despite the recent moderation in house price appreciation. Australian dwelling
prices increased by 3 per cent in the first eleven months of 2010, compared to the 12 per
cent increase over 2009 (Graph 9).
Graph 9
Australian Dwelling Prices*
2005 = 100
Index
Index
RP Data-Rismark
140
140
130
130
120
120
APM
110
110
100
100
90
2005
2006
2007
2008
2009
* Weighted average of houses and apartments in capital cities
Sources: APM; RBA; RP Data-Rismark
Reserve Bank of Australia
11 January 2011
2010
90
2
DEVELOPMENTS IN AUSTRALIAN HOUSEHOLDS’ BORROWING CAPACITY
Households’ potential borrowing capacity declined by around 13 per cent between March 2009
and December 2010, according to lenders’ online loan calculators. While higher interest rates
are the main source of the reduction in potential borrowing capacity, there also appears to have
been some tightening in serviceability standards by a number of lenders.
Maximum Borrowing Capacity
Graph 1
A survey of lenders’ online loan calculators
suggests that the potential borrowing capacity
of Australian households declined by around
13 per cent between March 2009 and
December 2010 (Graph 1). Furthermore, the
decline in the maximum loan amount was
apparent across all lenders, albeit to different
degrees (Table 1). The range of maximum loan
sizes offered by lenders declined somewhat.
Maximum Loan Size
Per cent of gross income; single individual, loan of 20 years
%
%
March 2009
550
550
December 2007
500
500
June 2004
450
450
May 2008
400
400
December 2010
350
300
350
40
60
80
100
120
Gross household income ($’000)
140
300
* Median of 12 lenders.
Sources: company websites; RBA
Table 1: Change in Maximum Loan Size since March 2009
Single individual; gross income of $90 000; loan of 20 years
Total
RAMS
Mortgage Choice
WBC
CBA
STG
Suncorp
ANZ
CUA
HSBC
BOQ
AMP Bank
Median
Due to interest rates*
-18.4
-16.3
-15.4
-14.6
-13.5
-13.0
-12.7
-12.9
-12.1
-12.1
-5.8
-13.0
-11.9
-13.3
-13.3
-13.3
-13.3
-13.3
-13.3
-14.1
-13.3
-13.3
-13.3
-13.3
Due to tax rates
1.0
1.4
1.4
1.4
1.5
1.4
1.4
1.4
1.4
1.4
1.4
1.4
Residual
(includes serviceability/
interest-rate buffers)
-7.5
-4.4
-3.5
-2.7
-1.7
-1.1
-0.8
-0.2
-0.2
-0.2
6.1
-1.1
* Six lenders allow changes in the maximum loan size arising from changes in interest rates to be identified. The median
effect for these lenders is used as a proxy for the other lenders.
Sources: ATO; company websites; RBA.
Most of the decline in households’ borrowing capacity reflects the increase in mortgage interest
rates over this period. Movements in interest rates influence borrowing capacity as lenders
generally set borrowing limits based on rules regarding the maximum size of repayments
relative to either gross or net income. Repayments are, in turn, often calculated using
a percentage point buffer above the current variable interest rate.
While the increase in interest rates has been the main factor behind the reduction in borrowing
capacity, there also appears to have been some tightening in lending standards since March
2009. In particular, most of the surveyed lenders have reduced their maximum loan-to-income
ratios by a greater amount than can be explained by the actual movement in interest rates and
changes in income tax.2 However, this ‘residual’ accounts for only a small change in
households’ borrowing capacity and it is possible that some of this may reflect an increase in
interest rate buffers
. Indeed, any tightening in serviceability
policies is roughly equivalent to an increase in interest rates of only 17 basis points.
In contrast to the other surveyed
lenders,
appears to have
loosened its serviceability policies.
This is consistent with the fact that
market share of housing loan
approvals has doubled since March
2009. However,
increased
market share has also been assisted by
its competitive pricing. Likewise,
market share moved in line with changes in its serviceability policies during this period
Debt-servicing ratio
The above analysis of maximum borrowing capacity can be recast to examine the maximum
debt-servicing ratios offered by lenders. While most lenders now use net income surplus models
– which require a borrower to have a minimum surplus of net income after taking into account
debt repayments and living expenses – the ratio of loan repayments to gross income is often
used as a simple metric of potential mortgage stress.
2
There are a number of ways financial institutions can tighten their lending standards that may not be reflected in the output
of online calculators; for example, tighter income checks, lower maximum Loan-to-Valuation-Ratios (LVRs), tighter criteria
on property type, or lower property valuations.
For a gross income of over $80 000, the maximum debt servicing ratio implied by banks’ online
calculators is currently around 50 per cent of gross income, somewhat higher than in March
2009 (Graph 3). This is partly because higher interest rates have resulted in higher maximum
repayments (the higher level of interest rates more than offsets the fall in maximum loan size). It
also reflects the lower tax paid on incomes of over $35 000 p.a. for the current fiscal year. That
is, for the same level of gross income, an individual’s net income, and therefore debt servicing
capacity, will be higher this year compared to last year.
Abstracting from tax effects, the maximum net debt servicing ratio (loan repayments as a share
of net income) is slightly higher than in March 2009 for net household incomes above $60 000
(Graph 4). Given that we would have expected a considerable rise in the net debt servicing ratio,
as a result of the rise in interest rates, these results are consistent with a general tightening in
lending standards.
Graph 3
Graph 4
Maximum Debt-servicing Ratio*
Maximum Debt-servicing Ratio*
Loan repayments as a share of gross income; single individual, 20 years
%
%
December 2007
Loan repayments as a share of net income; single individual, 20 years
%
%
75
May 2008
50
December 2007
50
70
45
45
June 2004
March 2009
40
40
December 2010
35
35
30
30
30
50
70
90
110
130
Gross household income ($’000)
* Assuming constant repayments
Sources: ATO; company websites; RBA
Cameron Deans and Lisa Zhou
Institutional Markets Section
Domestic Markets Department
20 January 2011
150
May 2008
75
December 2010
70
June 2004
65
65
March 2009
60
60
55
55
50
50
45
45
30
40
50
60
70
80
90
Net household income ($’000)
* Assuming constant repayments
Sources: ATO; company websites; RBA
100
110
4
From:
To:
Cc:
Subject:
Date:
CHAN, Iris
STEWART, Chris
BROADBENT, John; GORAJEK, Adam
RE: Note FS: APRA Credit Conditions Survey - December Quarter 2010 [SEC=UNCLASSIFIED]
Tuesday, 1 February 2011 09:53:36
Hi Chris,
I think you’re right about conditions for residential property development not having changed
much over the past two quarters. Obviously we don’t have any concrete information about
how tight conditions are in an absolute sense, but at least no bank has explicitly said that they
eased conditions for the segment in the December quarter.
Cheers,
Iris
From: STEWART, Chris
Sent: Monday, 31 January 2011 22:30
To: CHAN, Iris
Cc: BROADBENT, John
Subject: FW: Note FS: APRA Credit Conditions Survey - December Quarter 2010
[SEC=UNCLASSIFIED]
Hi Iris,
My reading of the September Quarter results – this afternoon as background for the SMP –
was that there really wasn’t much evidence suggesting that conditions were changing in any
material way from being relatively restrictive.
Regards
Chris
From: CHAN, Iris
Sent: Monday, 31 January 2011 7:20 PM
To: Notes policy groups
Subject: Note FS: APRA Credit Conditions Survey - December Quarter 2010 [SEC=UNCLASSIFIED]
In the December credit conditions survey, banks appeared to be looking for market share in
housing loans in an environment of muted credit demand. Nonetheless, no
significant relaxation of lending standards appeared to have occurred thus far. Banks reported
that margins
decreased for housing loans following a two-year period of reported margin compression.
Iris Chan
Australian Financial System
Financial Stability Department
Reserve Bank of Australia
P: +61 2 9551 8542
F: +61 2 9551 8052
CONFIDENTIAL
APRA CREDIT CONDITIONS SURVEY – DECEMBER QUARTER 2010
In the December credit conditions survey, banks appeared to be looking for market
share in
housing loans in an environment of muted credit demand.
Nonetheless, no significant relaxation of lending standards appeared to have occurred
thus far. Banks reported that margins
increased for housing loans following a two-year period of
reported margin compression.
CONFIDENTIAL
Residential housing lending
there appeared to be some
tightening in non-price lending standards,
Half of the sample
also reported a reduction in non-standard
loans as a share of new mortgage lending
Some of these
changes may have been done to prepare for the responsible lending requirements of the
new National Consumer Credit Protection regime, which took effect for all credit
providers on 1 January 2011. Media reports have suggested that banks are now
requiring both their branch and broker channels to ask additional questions of potential
borrowers to determine the suitability of a credit product, and borrowers to provide
more verification when applying for low-doc loans.6 The number of policy overrides
was reported to be little changed.
6 For example, see Searle (2010), ‘Credit Checks Stiffened’, Australian Financial Review, 10 January 2011,
p.39.
CONFIDENTIAL
All banks
reported
that
they
expect
higher
delinquencies over the coming quarter for
both owner-occupier and investor housing, a
trend that some respondents attributed to
seasonal impact from the Christmas/new
year period.
Banks also expect the demand for housing
credit to decline, partly because of the floods and partly because borrowers expect
interest rates to rise further.
Iris Chan
Australian Financial System, Financial Stability
31 January 2011
5
DOMESTIC MARKETS REVIEW: JANUARY 2011
21
Domestic Markets Department
The major banks have been active in
directing mortgage brokers to undertake a
more rigorous approach to assessing the
suitability of the borrower as, while this
obligation has been imposed on brokers
since mid 2010, there have been reports of
poor compliance.
Low doc housing lending continues under
the new credit regulations
Several
lenders
have
also
reportedly
overhauled their low-doc lending practices
following
the
aforementioned
regulatory
changes. Reaction to the new rules has
varied. While some lenders have cited
uncertainty
(regarding
the
standard
of
borrower assessment that is required under
the new regulations) as a reason for not reentering the low-doc market, other lenders
do not see any need to overhaul their
existing practices.
6
LOMAS, Phil
From:
Sent:
To:
Cc:
Subject:
Attachments:
CHAN, Iris
Wednesday, 9 February 2011 16:36
DONOVAN, Bernadette
GORAJEK, Adam
NPA sneak peek [SEC=UNCLASSIFIED]
image001.png; image002.png; image003.png; image004.png
1
Banks' Asset Quality
$b Non- performing
housing assets Domestic books
Non-performing
business assets*
Specific provisions
$b
20 20
15 15
Impaired
10 10
Business
Past- due
5
5 Housing
0 0
2007
2009
2011
2008
2010
2007
2009
2011
* Includes bill acceptances and securities, and other non -household loans
Sources: APRA Banks' Non- performing Assets
%
Domestic books
Per cent of on-balance sheet
Per cent of all on-balance
loans by type
sheet loans 4 %
4
Business*
3
3 2 Personal
2
Total
1 1
Housing
0 0
2007
2009
2011
2008
2010
* Includes lending to non-ADI financial businesses, bill acceptances and debt
* securities, and other non -household loans
Source: APRA
2
7
HILDA analysis of flood areas
A higher proportion of flood affected households in Queensland qualify as vulnerable under our
usual definition (DSR>50, LVR>80 or 90%) compared to Australia as a whole.
Flood affected areas
All Australia
Households With Low Equity and High Repayments
Households With Low Equity and High Repayments
Per cent of households with owner-occupier debt*
Per cent of households with owner-occupier debt*
%
8
%
%
8
8
DSR>30, LVR>80
%
8
DSR>30, LVR>80
6
6
DSR>30, LVR>90
4
4
4
4
DSR>50, LVR>80
2
DSR>50, LVR>90
0
0
2001 2002 2003 2004 2005 2006 2007 2008 2009
* Includes the first and second mortgages secured against the property
Source: HILDA Release 9.0
6
DSR>30, LVR>90
DSR>50, LVR>80
2
6
2
2
DSR>50, LVR>90
0
0
2001 2002 2003 2004 2005 2006 2007 2008 2009
* Includes the first and second mortgages secured against the property
Source: HILDA Release 9.0
This is caused predominantly by higher LVRs in the flood affected regions compared to Australian as
a whole. DSRs across the two regions were similar.
%
DSRs and LVRs
DSRs and LVRs
Indebted owner-occupier households
Indebted owner-occupier households
LVR
DSR
%
%
60
30
75th percentile
30
75th percentile
60
75th percentile
75th percentile
Median
Median
20
40
20
40
Median
Median
25th percentile
10
20
10
0
0
25th percentile
0
2001
2004
2007
2001
2004
2007
25th percentile
0
2004
40
0<DSR<30
2004
2007
%
%
40
40
30
30
20
20
10
10
0
%
40
0<DSR<30
30<DSR<50
30
60-80
80-100
DSR>50
20
10
0
40-60
Indebted owner-occupier households, 2009
DSR>50
30<DSR<50
20-40
2001
DSR Distribution by LVR
Indebted owner-occupier households, 2009
0-20
2007
Source: HILDA Release 9.0
DSR Distribution by LVR
30
20
25th percentile
2001
Source: HILDA Release 9.0
%
%
LVR
DSR
10
0
100+
0
0-20
Source: HILDA Release 9.0
20
20-40
40-60
60-80
80-100
100+
Source: HILDA Release 9.0
Loan-to-valuation Ratios
Loan-to-valuation Ratios
Indebted owner-occupier households by income quintile
Indebted owner-occupier households by income quintile
%
2001
2008
70
60
75th percentile
75th percentile
50
Median
40
Median
30
20
25th percentile
25th percentile
10
0
%
%
70
70
60
60
50
50
40
40
30
30
20
20
10
10
0
2
3
4
5
Total
2
3
4
5
Household disposable income quintile
Source: HILDA Release 9.0
Total
2001
%
2008
70
75th percentile
75th percentile
60
50
Median
40
Median
30
25th percentile
20
25th percentile
10
0
0
2
3
4
5
Total
2
3
4
5
Total
Household disposable income quintile
Source: HILDA Release 9.0
Shows thatin flood affected areas in 2009 more HHs increased their LVR because of negative house
price movements than in previous years, with debt being less important. Also, more HHs
experienced an increase in LVRs in flood affected areas than in previous years. These trends are not
evident at the aggregate level.
%
LVR Movements
LVR Movements
By reason, per cent of indebted owner-occupiers*
By reason, per cent of indebted owner-occupiers*
LVR down/equity up
LVR up/equity down
House price
movements
60
Debt & price
movements
40
Debt
movements
20
0
%
%
60
60
40
40
20
20
0
0
2002 2004 2006 2008 2002 2004 2006 2008
Renters and other
20
%
%
40
40
30
Home bought without
a mortgage
Debt & price
movements
Debt
movements
40
20
0
Per cent of all households
Owner-occupiers with original
mortgage
Indebted owner-occupiers*
30
60
Households' Debt Repayment Status
Per cent of all households
40
House price
movements
%
* Movements in LVRs between two adjoining survey years.
Source: HILDA Release 9.0
Households' Debt Repayment Status
All households
LVR up/equity down
2002 2004 2006 2008 2002 2004 2006 2008
* Movements in LVRs between two adjoining survey years.
Source: HILDA Release 9.0
%
LVR down/equity up
All households
Owner-occupiers with original
mortgage
40
Indebted owner-occupiers*
30
%
30
Renters and other
Ahead of schedule
20
Home bought without
a mortgage
20
Mortgage paid off
10
About on schedule
10
Ahead of schedule
Mortgage paid off
10
About on schedule
20
10
Behind schedule
0
2001
0
2004
2007
2001
2004
* Owner-occupiers with any mortgage debt
Source: HILDA Release 9.0
2007
Behind schedule
0
2001
0
2004
2007
2001
2004
2007
* Owner-occupiers with any mortgage debt
Source: HILDA Release 9.0
Surprisingly, a greater proportion flood affected households are ahead if schedule, although this
mainly reflects the fact that a greater proportion of flood affected households are indebted owneroccupiers.
Measures of Housing Arrears
Measures of Housing Arrears
Per cent by number*
Per cent by number; not seasonally adjusted
%
%
5
5
4
4
4
4
3
3
3
3
2
2
2
2
1
1
1
1
0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
0
0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
0
%
5
Behind schedule*
* Per cent of owner-occupier households, indebted with their 1st mortgage.
** Prime & non-conforming loans securitised by all lenders; excludes selfsecuritisations.
Sources: HILDA Release 8.0; Perpetual
%
* Per cent of owner-occupier households, indebted with their 1st mortgage.
Sources: HILDA Release 8.0;
Owner-occupier Housing Debt
Owner-occupier Housing Debt
2009, average
$000
300
2009, average
Original mortgage
Loans from friends, family, solicitors or
community organisations
Other mortgage debt*
Per cent of total debt in
income quintile (RHS)
200
100
0
%
$000
60
300
40
200
20
100
0
1
2
3
4
5
5
Behind schedule
Total
Income quintile
* Second mortgage, home equity and refinanced loans
Source: HILDA Release 9.0
%
Original mortgage
Loans from friends, family, solicitors or
community organisations
Other mortgage debt*
60
Per cent of total debt in
income quintile (RHS)
40
20
0
0
1
2
3
4
5
Total
Income quintile
* Second mortgage, home equity and refinanced loans
Source: HILDA Release 9.0
Owner-occupier Housing Debt 2008
Owner-occupier Housing Debt 2008
As a per cent of disposable income
As a per cent of disposable income
%
%
Original mortgage
Loans from friends, family, solicitors or
community organisations
Other mortgage debt*
800
800
%
%
Original mortgage
Loans from friends, family, solicitors or
community organisations
Other mortgage debt*
600
600
600
600
400
400
400
400
200
200 200
200
0
0
1
2
3
4
5
Total
0
0
1
2
Income quintile
* Second mortgage, home equity and refinanced loans
Source: HILDA Release 8.0
Rank
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
4
5
Total
Income quintile
* Second mortgage, home equity and refinanced loans
Source: HILDA Release 8.0
Queensland Housing Loan Arrears by Region
As at December 2010
Per cent of population
Region
in flood affected postcodes
Gold Coast East
100.00
Gold Coast Bal
100.00
Far North - North West
Mackay - Central West
Ipswich City
100.00
Sunshine Coast
5.05
Caboolture Shire
23.33
Wide Bay-Burnett
99.84
Logan City
87.44
Redcliffe City
100.00
West Moreton
99.43
Fitzroy
Inner Brisbane
100.00
Darling Downs - South West
89.60
Southeast Inner Brisbane
100.00
Northern QLD
Redland Shire
100.00
Southeast Outer Brisbane
100.00
Pine Rivers Shire
97.92
Northwest Outer Brisbane
100.00
Northwest Inner Brisbane
98.99
Australia
Sources: Perpetual; RBA
3
90+ days
arrears rate
0.85
0.72
0.71
0.66
0.59
0.58
0.56
0.54
0.54
0.51
0.50
0.44
0.43
0.41
0.36
0.33
0.24
0.23
0.22
0.22
0.17
0.44
8
ASIC SUMMER SCHOOL TALK :
SPECULATIVE BOOMS , BUBBLES AND BUSTS – HOW MUCH CAN WE KNOW ?
I would start by mentioning that nothing I have to say is relevant to near-term policy decisions.
Credit growth is slow; housing prices have levelled off; commercial property prices are at a
trough; and other markets are well off their peaks. The household saving ratio has returned to
levels last seen in the late 1980s. None of that sounds like a speculative boom to me.
History shows that asset price busts cost the most when they occur together with a banking
crisis. The recent bust in the United States was centred on households and housing markets. It
is actually quite unusual for households to be the instigators of the crisis in this way. Housing
price busts are more often a symptom or knock-on effect of financial instability, not the thing
that started it off. But property in general is usually implicated in banking crises. Commercial
real estate – offices, shopping centres, as well as loans for property development – are often
the source of problems for the financial system. They have certainly dominated the loan losses
of Irish, British and the smaller US banks lately. The decline in commercial property prices has in
fact been larger than for housing in almost every industrialised country (Graph 1).
Graph 1
Property Price Indicators
31 December 2000 = 100
Index
Commercial real estate
Residential real estate
Index
Spain
200
200
Australia
US
150
150
UK
Ireland
100
100
Sweden
50
2002
2006
2010
2002
2006
2010
50
Sources: APM; Bloomberg; BIS; Jones Lang LaSalle; Thomson Reuters
Why are property markets so often implicated in financial instability in this way? The answer
lies in the combination of their use as collateral for lending, and the dynamics of the physical
property markets. Lenders would rather lend against collateral than unsecured. And borrowers
like to magnify their capital gains on property through leverage. But if prices fall, leverage
magnifies the capital losses. Some borrowers become distressed and a fire sale ensues, causing
further distress.
Equity markets can also boom and bust. But they are less likely to spark off a vicious circle of
distress sales, because buyers in that market are generally less leveraged.
The inherent dynamics of the physical market for property also contribute to the boom-bust
cycles. It takes time to build a building, more so for large, complex commercial developments
than detached housing. And new construction is always going to be a small fraction of the
existing stock. So the supply of both commercial real estate and housing is always going to be
sluggish. And if it is sluggish, the market can end up oscillating in repeated boom-bust cycles.
These are the inherent dynamics of the system. In this sense, and unlike equity markets,
property markets can boom and bust even without a credit-fuelled speculative bubble.
1
What are the fundamentals?
It should be obvious that we would want to know if an expansion in property markets is
sustainable, or a boom that will inevitably bust, whether it is truly a bubble or not. The question
is how we could tell. Of course, hindsight is a beautiful thing and now many people think that of
course you can tell if there is a bubble! I think we need to be more modest about our
understanding of the economy than that.
My first point is that it is not sensible just to look at some historical average for prices, or some
other simple ratio, and assume that they will revert to that level. The fundamentals do not stay
constant over long periods. For example, Australia has had a much lower inflation rate over the
past twenty years than it did over the previous twenty. So, nominal interest rates are lower.
That means borrowers can service a bigger mortgage with the same repayment and on the
same income (Graph 2).
Graph 2
Maximum Loan-to-income Ratio*
Assuming 3% real interest rate on 25-year loan
Ratio
Ratio
8
8
Repayment = 50% income
6
6
Repayment = 40% income
4
4
Repayment = 30% income
2
0
2
0
1
2
3
4
5
6
7
Inflation (%)
8
9
10
11
0
12
Source: RBA
Another example is that credit supply is no longer artificially restricted the way it was in the
1970s and early 1980s, before the financial sector was deregulated. Some good credit risks
amongst households were not able to get mortgages back then; now they can. So it is not
correct to assume that prices will revert to mean, either in real terms or relative to income.
There are other fundamentals than income.
Another reason why you shouldn’t assume that prices revert exactly to historical averages is
that most models of fundamentals are really models of a single person or household. If you use
them, you are implicitly assuming that the so-called ‘representative agent’ assumption is true.
But in reality, people are different; their circumstances – their fundamentals – can and do
change. Distributional issues will therefore matter a great deal. For example, household debtincome ratios and the leverage on the housing stock in Australia have risen over the past
decade. But survey data show that much of that occurred because more households in older
age groups still have a small mortgage (Graph 3). Fewer of them have completely paid their
mortgage off. This seems a less risky outcome than if it had been that younger, recent homebuyers had even bigger mortgages than they currently do.
2
Graph 3
Indebted Owner-occupiers by Age Group
%
%
Share of households
60
60
2001
30
%
30
%
Median LVR
60
60
30
30
2009
0
0
15-24
25-34
35-44
45-54
55-64
65+
Source: HILDA Release 9.0
Fundamentals also differ across countries. Housing supply can be more or less flexible; the tax
treatment of housing can differ. And housing is more expensive in bigger cities, even relative to
city-level income (Gabaix 1999). National averages are therefore higher if everyone lives in big
cities (Ellis and Andrews 2001). So you should not expect either rental returns or price-income
ratios to be the same across countries. It is as mistaken to think that prices will revert to some
cross-country average as it is to think they will necessarily revert to an historical average. You
would end up concluding that movements in housing prices during the pre-crisis boom phase
implied that the US had less of a bubble than Canada (Graph 4)!
Graph 4
House Prices and Household Debt
Percentage point change in ratios to household income*
(2000 to 2006)
% pts
% pts
 House prices
 Household debt
200
200
150
150
100
100
50
50
0
US
Australia Canada
Spain
NZ
UK
0
* Household income is after tax, before interest payments
Sources: BIS; national sources
I would also argue that it is not as simple as just estimating a model of fundamentals, and
attributing the deviation between the model and actual data as the ‘bubble’ component.
As I mentioned, the models are usually of a single household. And we need to take variations in
credit constraints into account; most models of fundamentals don’t do that.
I have another, more philosophical, objection to this approach. It assumes that the model is
exactly right. But we know that ‘all models are wrong’; there are always simplifications. How do
you know that the difference between data and model is really a bubble? It could just be that
your model isn’t very good, or that your data don’t measure exactly what your theory requires.
The required fundamentals are things like expected future interest rates or expected future
incomes growth. The estimate of fundamental levels for prices will only be as good as the
estimates of these inputs.
3
In my view, just asking whether prices are in line with fundamentals misses the point. It is a
wholly static question. Of course if there were a bubble, that would be a concern. Speculative
booms and bubbles can collapse in on themselves from their own dynamics.
But we don’t face a simple decision of: bubble – worry; no bubble – don’t worry.
Even if asset prices are in line with estimated fundamentals, there are times when we should
not be complacent. As I mentioned in my speech last May,1 every speculative boom starts with
something real, something fundamental. Behind the dot-com bubble, there really was new
technology and strong productivity. And before the Asian crisis, Asia really was developing
fast.2
So if you see unusually strong fundamentals, and price rises to match, you don’t need to be
alarmed. But you should definitely be alert! That asset market need not go from that
fundamentals-based phase to a more speculative one, but its conditions are such that it could.
Likewise you should be alert to the fundamentals’ dynamics, and the risk that they might
reverse. That doesn’t mean that policymakers should try to undo the effects of strong
fundamentals on asset markets or the real economy. But it does suggest that they should take
steps to ensure that the financial system would be resilient if those fundamentals did reverse.
That explains our interest in the lending standards that apply to the financing of property and
other assets.
You should be even more alert if you detect increasing speculative activity. Remember, bubbles
are speculative booms; the motivations of the asset buyers matter. We might not be able to see
the speculation directly, but we can dust for its fingerprints. For example, prudential
supervisors notice changes in lenders’ behaviour, such as when their business models and
product offerings have become more aggressive. Policymakers can and do share those
observations and concerns, even if they can’t make them public. We can also look for signs of
borrowers using loan products that enable speculative purchases, like investor housing loans or
deposit bonds. For example, it turns out that it was interest-only loans – not subprime loans –
that predicted which US cities would have boom-bust cycles in housing prices (Barlevy and
Fisher 2010). And in ASIC’s realm, we can see if there are signs of speculative fervour in the
growth of the ‘property seminar’ industry. This was a big deal during the 2002–03 boom, and
we said so at the time in the Bank’s Submission to the Productivity Commission.
We can also use household surveys and other disaggregated data, to check for concentrations
of risk. For both housing and commercial property, it is the marginal borrower, not the median
borrower, where the risks lie. And the median household is not the median borrower. As I
mentioned before, much of the recent increase in housing debt in Australia was driven by lowrisk older households with small mortgages. It would be different if households with low and
variable incomes and little equity in the home had contributed the most. That is what
happened in the United States during the boom in the subprime market.
1
http://www.rba.gov.au/speeches/2010/sp-so-180510.html
I struggle to name the fundamental behind the US subprime mortgage boom, but long-term interest rates and thus US
mortgage rates did fall to low levels, and the GSEs that dominated the US prime mortgage market were facing new constraints
on their activities.
2
4
What policy responses are possible?
Whether it’s a speculative boom or a shift in fundamentals that only might become one, the
question is how to respond. If the fundamentals are strong, then conventional macroeconomic
policies will be pushing in the right direction anyway. You don’t have to target asset prices. You
just have to recognise that they tell you something about what is happening in the real
economy. A good example of this connection was, again, the boom in housing prices in 2002–
2003. The real economy was also very strong at the time, especially consumption growth. Even
if you had thought that the growth in housing prices was based on fundamentals, it was clear
what macro policies should be doing.
But it might not make sense to tighten macroeconomic policies more than the real economy
requires. So if you are concerned that the boom might be partly speculative, you might want to
look for other ways to respond. Many people in the international policy community are looking
at ‘macroprudential’ policy. I don’t have time to discuss these today. But broadly speaking, it
means using prudential tools to lean against credit-fuelled asset speculation.
There are some policies that can help limit how far a speculative boom can go, even though
they weren’t designed with that purpose in mind. A good example of this is consumer
protection regulation around the provision of credit. Australia’s National Consumer Credit Code
requires that the lender must be able to verify and show that the consumer can be reasonably
expected to repay the loan from their own resources, without having to sell the collateral. If
not, the loan can be modified or even set aside by the courts. This provides a powerful
incentive for lenders to extend credit responsibly. They should be less prone to do the assetbased and ‘no-doc’ lending that caused so many problems in the United States in the lead-up to
the crisis. By limiting lending in this way, the NCCC helps limit the finance available for risky and
speculative purposes, at least if a household is doing the borrowing.
Tax policy is another example of a type of policy that isn’t obviously about financial stability,
but can be designed to limit speculative behaviour. Of course, tax systems are designed with
many other goals in mind, not just their effects on asset markets and financial stability. But it is
useful to have open eyes about the incentives tax systems create or don’t create for speculative
behaviour. If different countries have different tax systems, the same developments in asset
markets might have different implications for their financial stability.
Policymakers with a range of mandates have to pay attention to asset prices and the financing
of those markets. But there is no single indicator or smoking gun that tells you it’s a bubble. Nor
can you find an easy rule of thumb to know if it’s a boom that might not be speculative now,
but could become so.
In the end, there is no substitute for careful analysis from a range of perspectives, focussed on
distributions and risks, not just macro-level data and simple ratios. There is also no substitute
for policymakers who can use that analysis judiciously, neither dismissing every boom as based
on fundamentals, nor seeing bubbles in every wobble of a price index. And if they do see
something that concerns them, those policymakers need to have the mandate and the
willingness to respond. Sometimes that might mean responding to a boom when it occurs,
perhaps with a prudential instrument. But perhaps just as often, it might require ensuring that
the broader institutional environment does not overly reward leveraged speculation.
5
References
Barlevy G and JDM Fisher (2010), ‘Mortgage Choices and Housing Speculation’, Federal Reserve
Bank of Chicago Working Paper 2010-13.
Ellis L and D Andrews (2001), ‘City Sizes, Housing Costs, and Wealth’, Reserve Bank of Australia
Research Discussion Paper 2001-08.
Gabaix X (1999), ‘Zipf’s Law and the Growth of Cities’, American Economic Review, 89(2), pp
129–132.
6
Speculative booms, bubbles and
busts – how much can we know?
Luci Ellis
Head of Financial Stability Department
Reserve Bank of Australia
Presentation to ASIC Summer School
22 February 2011
Property Price Indicators
31 December 2000 = 100
Index
Commercial real estate
Residential real estate
Index
Spain
200
200
Australia
US
150
150
UK
Ireland
100
100
Sweden
50
2002
2006
2010
2002
2006
2010
Sources: APM; Bloomberg; BIS; Jones Lang LaSalle; Thomson Reuters
50
Maximum Loan-to-income Ratio*
Assuming 3% real interest rate on 25-year loan
Ratio
Ratio
8
8
Repayment = 50% income
6
6
Repayment = 40% income
4
4
Repayment = 30% income
2
2
0
0
12
0
1
2
Source: RBA
3
4
5
6
7
Inflation (%)
8
9
10
11
Indebted Owner-occupiers by Age Group
%
%
Share of households
60
60
2001
30
%
30
%
Median LVR
60
60
30
30
2009
0
0
15-24
25-34
Source: HILDA Release 9.0
35-44
45-54
55-64
65+
House Prices and Household Debt
Percentage point change in ratios to household income*
(2000 to 2006)
% pts
% pts
 House prices
 Household debt
200
200
150
150
100
100
50
50
0
0
US
Australia Canada
Spain
NZ
* Household income is after tax, before interest payments
Sources: BIS; national sources
UK