House Price Cycles and the Real Economy

Report
House Price Cycles and
the Real Economy
DENIZ IGAN
IMF – RESEARCH
LIME WORKSHOP
BRUSSELS - DECEMBER 8, 2012
Disclaimer: Views expressed in the presentation and during the talk are those
of the presenter and should not be ascribed to the IMF.
Background Work
 The Changing Housing Cycle and the Implications for
Monetary Policy – Chapter 3 in WEO April 2008
 Housing, Credit, and Real Activity Cycles: Characteristics
and Comovement – JHE 2011 (revision of Three Cycles:
Housing, Credit, and Real Activity, IMF WP 09/231)
 Global Housing Cycles – IMF WP forthcoming
 Early Warning Exercise – conducted twice a year
Outline
 Research questions
 Relevant literature (briefly)
 Preview of results
 Methodology and data (briefly)
 Results
 Conclusions and policy implications
Research Questions
 How similar are the cycles: in duration,
amplitude, etc.?
 How do the cycles relate to the interest rate
changes?
 Are the cycles synchronized, or does one lead
the others?
 How important are global versus local factors?
 What are the policy implications?
Literature—Domestic Context
 Bank credit, house prices and aggregate demand
tend to move in tandem
 Financial accelerator effects through external
finance premium and collateral prices result in
procyclicality of bank credit

Bernanke and Gertler, 1989; Kiyotaki and Moore, 1997
 Liquidity and leverage of banks also contribute

Adrian and Shin (2008), Berger and Bouwman (2008)
 Liquidity of housing wealth and availability of
housing financing can also alter the relationship
through consumption and investment
Literature—International Context
 The degree of synchronization in real and financial
cycles increased

Kose, Prasad and Terrones, 2003; Imbs, 2004; Otrok and Terrones,
2005
 International comovement in house prices may
reflect comovement of interest rates

Especially with debt-financed ownership and adjustablerate mortgages
 Global liquidity may have also contributed

Belke and Orth, 2008
Literature—Monetary Policy
 Transmission through the credit channel and the
housing market
 Monetary policy, credit and asset prices


Benign neglect: Monetary policy is not an effective tool for
targeting asset prices, so better not to prick the bubble but
mop up afterwards (Bernanke, Gertler, and Gilchrist, 1999)
Leaning against the wind: Monetary policy needs to play a
more proactive role to prevent financial bubbles (Borio and
Lowe, 2002, and Borio, 2006)
Contribution of This Paper
 Multidirectional links
 Goodhart and Hofman (2008)
 Focus on output, bank credit,
house prices (as well
as residential investment) and interest rates
 Descriptive approach

GDFM as opposed to VAR
 Data properties
 Unit root tests
 Filtering as opposed to differencing
 Different horizons
 Short to medium term (6 to 16 quarters)
 Long term (16 to 32 quarters)
Preview of Results
 House price cycles lead credit and business
cycles over the long run
 Interest rates tend to lag other cycles at all
horizons
 Country cycles are largely driven by global
factors

the role of which has increased over time, especially for
credit and business cycles
 U.S. cycles tend to lead other countries’
respective cycles
Generalized Dynamic Factor Model
 Identifies a common component using a large
number of series
 Builds on the traditional factor models

Sargent and Sims (1977) and Geweke (1977)
 By allowing for serial correlation and weakly
cross-sectional correlation of idiosyncratic
components

Chamberlain (1983) and Chamberlain and Rothschild (1983)
 Recent examples



Giannone, Reichlin, and Sala (2002)
Forni and others (2005)
Eickmeier (2007)
(Approximate) GDFM
Yt  X t t
where Yt is a (N x T) vector stochastic stationary process with zero mean and
unit variance and Xt and Ξt are (N x T) vectors of common and idiosyncratic
components, respectively.
Xt can be written as:
Yt  CFt  t
where Ft is a (r x T) vector of common factors and C is a (N x r) matrix of factor
loadings.
The model has N>>T and r<<N.
The common factors assumed to follow an AR(1) process:
Ft  BFt 1  ut ,
with B and (r x r) matrix and ut a (r x T) vector of residuals.
Measures of Comovement
The dynamic correlation varies between -1 and 1. Formally,
where the numerator is the cospectrum between y1 and y2 processes at frequency
λ and S (y1) and S (y2) are the spectral density functions of the processes at
frequency λ defined over –π and π.
Coherence is intrinsically related to dynamic correlation given by:
Coherence is symmetric and a real number between 0 and 1.
Lead-Lag Relationship
The phase angle between processes helps identify
the lead-lag relationship
where q (y1 y2) is the quadrature spectrum. Only
when K (y1y2) ≠0, the phase angle converges in
distribution to a normal random variable.
Panel Data




18 industrial countries
1981 Q1 to 2006 Q4
1,283 series
Real activity indicators





Consumption
Investment, including residential
International trade
Confidence indicators
Portfolio and FDI flows
 Financial variables



Credit to the private sector and other monetary aggregates
Short-term and long-term interest rates
House prices and stock prices
 Balance sheet data


Household savings and assets (in particular, housing stock)
Capital stock of business sector
Unit Root Tests
 ERS (Elliott, Rothenberg and Stock, 1996)
 Generalized least squares
 More powerful than standard Dickey-Fuller test
 KPSS (Kwiatowski, Phillips, Schmidt and Shin,
1992)
Cross check
 Stationarity as H0 instead of unit root

 Constant and deterministic trend
 Lags chosen based on the Schwarz information
criterion
Results of Unit Root Tests
 House prices are sometimes I(2)
 France, Ireland, the Netherlands, New Zealand,
Sweden and the United States
 Credit series are also I(2) in some cases
 Japan and Spain
 Over-differencing versus under-differencing
Unit Root Tests (Details)
Table 1. Unit Root Tests
ERS
Australia
Austria
I(1)
I(1)
Output
Decision in
Cases of
Conflicting
KPSS
Evidence
I(1)
ERS
I(1)
I(1)
2
I(1)
1, 5
I(2)
Residential Investment
Decision in
Cases of
Conflicting
ERS KPSS
Evidence
I(0)
I(1)
I(1)
I(1)2
ERS
Credit
Decision in
Cases of
Conflicting
KPSS
Evidence
I(1)
I(1)
I(1)
I(0)
I(1)
I(1)
I(2)
I(1)
I(1)
I(1)
I(1)
I(1)
I(1)
Belgium
I(0)
I(1)
Canada
I(1)
I(1)
I(1)
I(1)
I(1)
I(1)
I(1)
I(1)
Denmark
Finland
France
I(1)
I(0)
I(1)
I(1)
I(1)
I(1)
I(1)
I(1)
I(2)
I(1)
I(1)
I(2)
I(1)
I(1)
I(1)
I(1)
I(1)
I(1)
I(1)
I(2)
I(2)
I(1)
I(2)
I(2)
Germany
I(1)
I(1)
I(1)
I(1)
Ireland
I(1)
I(2)
Italy
I(1)
I(1)
Japan
I(1)
I(2)
Netherlands
I(1)
I(1)
I(1)
House Prices
Decision in
Cases of
Conflicting
KPSS
Evidence
I(1)
1
I(1)
I(1)1
I(1)
I(2)
I(0)
I(2)
I(1)
I(1)
I(1)
I(2)
I(1)1, 5
I(2)1, 5
I(1)
I(1)
I(1)
I(2)
1, 5
I(1)
I(1)
I(1)
I(1)
1
I(1)
I(1)
I(1)
I(1)
I(1)
I(1)
I(2)
I(2)
1
I(1)
I(1)
2
I(1)
I(1)
I(0)
I(2)
I(2)
I(2)
I(1)
I(2)
I(2)1
I(0)
I(1)
I(1)
I(1)
I(1)1, 5
I(2)
I(0)
1, 5
I(2)
1, 5
I(0)
I(1)
New Zealand
I(1)
I(1)
I(1)
I(2)
I(2)
I(0)
I(1)
Norway
I(1)
I(1)
I(1)
I(2)
I(2)1, 5
I(1)
I(1)
Spain
I(1)
I(1)
I(1)
I(1)
I(1)
I(1)
Switzerland
I(1)
I(1)
I(1)
I(1)
I(1)
I(2)
United Kingdom
United States
I(1)
I(1)
I(1)
I(1)
I(1)
I(2)
I(1)
I(2)
I(1)
I(1)
I(1)
I(1)
I(1)
I(1)
I(1)4
I(2)5
Source: Authors' estimates.
Notes: The table reports the results of two unit root tests: ERS (Elliott, Rothenberg, and Stock, 1996) and KPSS (Kwiatkowski, Phillips, Schmidt,
and Shin, 1992). All tests were done including a constant and a trend. The number of lags was chosen using the Schwarz criterion and ensuring
that no serial correlation is left in the residuals. Highlighting in the table identifies the cases where the two tests rendered conflticting results. In
such cases, unit root tests were also done excluding a constant and trend, and graphical evidence was examined particularly closely. Output is the
real GDP as calculated by the IMF's International Financial Statistics , house prices are expressed in real terms by deflating the nominal house
price indices by CPI, credit is bank credit to the private sector deflated by CPI.
1. The charts do suggest that stationarity is only achieved at the specified differenced series.
2. The ERS test barely passes the confidence level.
3. Schwarz criterion suggests taking 3 lags and there is no SC either with 1 or 2 lags. Three lags suggest I(1).
4. The KPSS test barely passes the confidence level. In addition, observation of the series suggests that it contains a unit root.
5. When the ERS test on first differences does reject the I(1) null, a relatively higher error of type I is assumed for KPSS test (1 percent).
Band-Pass Filter
 Corbae and Ouliaris (2006) ideal band-pass filter
 Passes through the components of time series with
periodic fluctuations between 6 and 32 quarters (in
line with widely used minor-major cycle lengths)
 Consistent
 No finite sampling error
 Not subject to end-point problems
Advantages of Filtering over Differencing
Figure 1 - United States: Spectra of Real GDP Filtered
(Y axis: spectrum; X axis: periodicities in quarters)
1,4
0,00025
1,2
0,0002
1
0,00015
0,8
0,6
0,0001
0,4
0,00005
0,2
0
In
fin
ite
64
,0
32
,0
21
,3
16
,0
12
,8
10
,7
9,
1
8,
0
7,
1
6,
4
5,
8
5,
3
4,
9
4,
6
4,
3
4,
0
3,
8
3,
6
3,
4
3,
2
3,
0
2,
9
2,
8
2,
7
2,
6
2,
5
2,
4
2,
3
2,
2
2,
1
2,
1
0
GDP Ideal Band Pass, Corbae-Ouliaris Filter
GDP First Differenced (RHS)
Wrong data transformation may introduce a downward bias in
the degree of economic integration and an upward bias in the
efficiency of uncoordinated macroeconomic policies.
Table 2. Variance Shares
Average
M aximum
M inimum
Standard deviation
Coefficient of variation
Variance share exceeding 20 p ercent
Differenced Series
Filtered Series
0.07
0.18
0.01
0.04
0.62
0.03
0.31
0.70
0.01
0.17
0.55
0.68
Source: Authors' estimates.
Notes: The table rep orts the variance shares of common comp onents for two alternative
way s of data treatment - first differencing and filtering using the ideal band p ass filter.
The series included in this illustrative exercise are outp ut, house p rices, credit, short- and
long-term interest rates. The last row shows the number of series, in p ercent of the total
number of series, for which the variance share exp lained by the common comp onents of
the included series exceeds 20 p ercent.
Characteristics of Cycles
 Credit and house price cycles tend to be slightly
more protracted on average than business cycles...

but peaks and troughs generally are not too far.
 Credit and house price cycles have larger swings
than real activity cycles).
 In some countries, there are significant differences
across these characteristics…

Countries that allow MEW and refinancing tend to have
larger amplitude and longer duration in house price cycles
relative to the business cycle if the share of variable rate
mortgages is large (e.g., New Zealand versus Germany).
Characteristics of Cycles
Procyclicality
 Financial accelerator model is not empirically
supported in all countries—correlations vary in
significance and sign.
 Collateral is a driver of procyclicality more than
lending is—stronger correlation between output
and house prices than between output and bank
credit.
Procyclicality
Table 5. Correlation Coefficients for Total Cyclical Components
Output-House Prices
Output-Credit
Output-Residential
Investment
Credit-House Prices
Credit-Residential
Investment
House PricesResidential Investment
Australia
Austria
Belgium
Canada
Denmark
Finland
France
Germany
Ireland
Italy
Japan
Netherlands
New Zealand
Norway
Spain
Switzerland
United Kingdom
United States
0.55
0.74
0.72
0.70
0.46
0.89
0.59
0.28
0.53
0.37
0.45
-0.14
0.51
0.54
0.74
0.36
0.85
0.13
0.69
0.82
0.79
0.63
0.62
0.40
0.73
-0.31
0.81
0.19
0.46
0.69
0.17
-0.09
0.61
0.11
0.69
0.59
0.65
-0.04
0.67
0.50
0.38
0.87
0.70
0.50
0.50
0.45
0.43
0.44
0.79
0.20
0.75
0.26
0.73
0.50
0.36
0.75
0.85
0.77
0.60
0.53
0.33
-0.02
0.54
0.25
0.73
-0.03
0.44
-0.10
0.25
0.17
0.74
0.37
0.26
0.02
0.67
0.30
0.48
0.67
0.28
-0.36
0.37
0.44
0.35
0.59
0.08
-0.20
0.24
-0.06
0.45
0.21
0.76
-0.40
0.83
0.70
0.58
0.93
0.51
0.28
0.14
0.50
-0.07
0.32
0.65
0.59
0.76
0.45
0.78
0.51
Mean
0.52
0.48
0.52
0.42
0.27
0.49
Source: Authors' estimates.
Notes: The table reports the correlation coefficients for the total cyclical component of output (real GDP), house prices, residential investment, and credit series, over the
entire sample period. For each pair listed in the column title, entries in the table show the correlation between the series over the whole sample period. Correlation
coefficients larger than 0.40 are deemed to be significant and are highlighted.
Common Components
 A large share of common components drive the three
cycles.
 The share of the common component in GDP and in credit
has increased over time.
 The importance of common shocks and/or the speed of
transmission of shocks has increased over time.
Common Components
GDP
Common component
Total cyclical movement
0.05
0.03
GERMANY
UNITED STATES
0.04
0.02
0.03
0.02
0.01
0.01
0.00
0.00
-0.01
-0.01
-0.02
-0.03
-0.02
-0.04
-0.05
1981Q1
1985Q1
1989Q1
1993Q1
1997Q1
2001Q1
2005Q1
0.03
-0.03
1981Q1
1985Q1
1989Q1
1993Q1
1997Q1
2001Q1
2005Q1
1997Q1
2001Q1
2005Q1
0.04
UNITED KINGDOM
JAPAN
0.03
0.02
0.02
0.01
0.01
0.00
0.00
-0.01
-0.01
-0.02
-0.03
1981Q1
-0.02
1985Q1
1989Q1
1993Q1
1997Q1
2001Q1
2005Q1
-0.03
1981Q1
1985Q1
1989Q1
1993Q1
Common Components
Credit
Common component
Total cyclical movement
0.05
0.10
UNITED STATES
0.04
GERMANY
0.08
0.06
0.03
0.04
0.02
0.02
0.01
0.00
0.00
-0.02
-0.01
-0.04
-0.02
-0.06
-0.03
-0.08
-0.04
1981Q1
1985Q1
1989Q1
1993Q1
1997Q1
2001Q1
2005Q1
0.06
-0.10
1981Q1
1985Q1
1989Q1
1993Q1
1997Q1
2001Q1
2005Q1
1997Q1
2001Q1
2005Q1
0.04
UNITED KINGDOM
JAPAN
0.03
0.04
0.02
0.02
0.01
0.00
0.00
-0.01
-0.02
-0.02
-0.03
-0.04
-0.04
-0.06
1981Q1
1985Q1
1989Q1
1993Q1
1997Q1
2001Q1
2005Q1
-0.05
1981Q1
1985Q1
1989Q1
1993Q1
Common Components
House prices
Common component
Total cyclical movement
0.03
0.02
UNITED STATES
GERMANY
0.02
0.02
0.01
0.01
0.01
0.00
0.00
-0.01
-0.01
-0.01
-0.02
-0.02
-0.02
-0.03
-0.03
1981Q1
1985Q1
1989Q1
1993Q1
1997Q1
2001Q1
2005Q1
0.20
-0.03
1981Q1
1985Q1
1989Q1
1993Q1
1997Q1
2001Q1
2005Q1
1997Q1
2001Q1
2005Q1
0.08
UNITED KINGDOM
JAPAN
0.06
0.15
0.04
0.10
0.02
0.05
0.00
-0.02
0.00
-0.04
-0.05
-0.10
1981Q1
-0.06
1985Q1
1989Q1
1993Q1
1997Q1
2001Q1
2005Q1
-0.08
1981Q1
1985Q1
1989Q1
1993Q1
Evolution of Common Components
Table 7. Evolution of Cyclical Movements Driven by Common Components
1980s
Output
1990s
House Prices
1980s
1990s
2000s
Residential Investment
1980s
1990s
2000s
2000s
Australia
Austria
Belgium
Canada
Denmark
Finland
France
Germany
Ireland
Italy
Japan
Netherlands
New Zealand
Norway
Spain
Switzerland
United Kingdom
United States
0.82
0.30
0.90
0.85
0.08
0.79
0.43
0.67
0.76
0.89
0.72
0.72
0.16
0.56
0.60
0.84
0.90
0.78
0.84
0.72
0.89
0.92
0.63
0.81
0.82
0.66
0.85
0.79
0.71
0.84
0.90
0.18
0.88
0.73
0.92
0.56
0.64
0.91
0.97
0.91
0.66
0.95
0.95
0.81
0.93
0.91
0.93
0.96
0.90
0.50
0.95
0.95
0.70
0.90
0.63
0.59
0.90
0.89
0.68
0.65
0.20
0.33
0.57
0.26
0.64
0.21
0.80
0.62
0.70
0.81
0.84
0.85
0.60
0.90
0.78
0.62
0.59
0.82
0.62
0.65
0.76
0.79
0.65
0.85
0.84
0.87
0.85
0.86
0.74
0.62
0.35
0.54
0.92
0.74
0.87
0.82
0.69
-0.02
0.90
0.92
0.21
0.11
0.53
0.53
0.83
0.59
0.78
0.42
0.45
0.84
0.88
0.77
0.70
0.65
0.86
0.79
0.73
0.68
0.76
0.79
-0.16
0.83
0.75
0.62
0.66
0.19
0.74
0.09
0.84
0.35
0.71
0.85
0.90
0.18
0.81
0.56
0.61
0.31
0.64
0.89
0.75
0.58
0.54
0.67
Mean
Median
0.65
0.74
0.77
0.83
0.87
0.92
0.60
0.65
0.71
0.77
0.58
0.64
0.63
0.74
0.59
0.65
1980s
Credit
1990s
2000s
0.62
0.81
0.90
0.46
0.69
0.94
0.94
0.49
0.70
0.66
0.82
0.86
0.84
0.31
0.86
0.57
0.51
0.73
0.90
0.49
0.81
0.89
0.55
0.68
0.40
0.38
0.91
0.67
0.31
0.63
0.78
0.39
0.21
0.40
0.86
0.80
0.85
0.91
0.91
0.87
0.60
0.79
0.25
0.57
0.84
0.35
0.67
0.87
0.67
0.34
0.80
0.27
0.71
0.51
0.87
0.89
0.89
0.92
0.90
0.91
0.82
0.53
0.94
0.88
0.69
0.96
0.84
0.67
0.83
0.59
0.91
0.87
0.68
0.72
0.60
0.60
0.66
0.69
0.84
0.88
Source: Authors' estimates.
Notes: The table reports the correlation coefficients for the total cyclical components (including the common and idiosyncratic component, as shown in Figure
2) and the common components (shown in Figure 4).
Leads and Lags
 During the minor cycle, house prices lead output and
credit only in a few cases.
 In the long run, there is some support for financial
accelerator mechanism, but which channel?

balance sheet improvements -> credit -> house prices
or

house prices -> creditworthiness -> credit
 Short-term interest rates never lead house prices.
 House prices lead output, which in turn leads credit.
 U.S. cycles lead the corresponding cycles in the long
run in most cases, and U.S. credit cycles lead only in
the short run.
Leads and Lags (Details - 1)
Table 6. Leads and Lags between Cycles within Countries
Credit-Output
Credit-House Prices
Credit-Residential Investment
Output-House Prices
Output-Residential Investment
6-16 quarters 16-32 quarters 6-16 quarters 16-32 quarters 6-16 quarters 16-32 quarters 6-16 quarters 16-32 quarters 6-16 quarters 16-32 quarters
Australia
Austria
Belgium
Canada
Denmark
Finland
France
Germany
Ireland
Italy
Japan
Netherlands
New Zealand
Norway
Spain
Switzerland
United Kingdom
United States
lag
lead
lag
lag
lead
lag
contemp.
lead
lead
lead
lag
contemp.
lag
lead
lag
contemp.
lead
lead
lag
lag
contemp.
lag
lead
lag
lag
lead
lead
lag
lead
lead
lag
lead
lag
lead
lag
contemp.
contemp.
lead
lag
lag
lead
lag
lag
contemp.
lead
lead
contemp.
lag
contemp.
lead
lag
lag
lag
lag
lag
lead
lag
lag
lag
lag
lag
contemp.
lead
lag
lead
lag
lag
lead
lag
lead
lag
lag
lag
lag
lag
lag
contemp.
lag
lag
lead
lag
lag
lag
lag
lag
lead
lag
lead
lag
lag
contemp.
lag
lag
lag
lag
lag
lag
lead
lag
lead
lag
contemp.
lag
contemp.
contemp.
lag
lag
lag
contemp.
contemp.
lag
lag
lag
lag
lag
contemp.
lead
contemp.
lead
lag
lead
lag
contemp.
contemp.
contemp.
lag
lag
lead
lag
contemp.
lag
lag
lag
lag
lead
lead
lead
lag
lead
lead
lag
lag
contemp.
lag
lag
lag
lag
lag
lag
lead
lag
lead
lag
lead
lag
lag
contemp.
lead
lag
lag
lag
lag
contemp.
lag
contemp.
lag
lag
lag
contemp.
lag
lag
lead
contemp.
contemp.
contemp.
lead
lead
lag
lag
lag
Lags
Contemporaneous
Leads
39
17
44
50
11
39
50
22
28
67
6
28
78
6
17
67
22
11
44
39
17
56
11
33
72
6
22
50
33
17
Source: Authors' estimates.
Notes: The table reports the lead-lag relationship between pairs of series that are the focus of the analysis. Interest rates are nominal short-term rates. For each pair listed in the column title,
entries in the table indicate whether the first variable leads or lags the second variable, or whether the relationship is contemporaneous, on average, over the frequency band. The numbers in
the bottom indicate the percentage of countries with a given type of relation. Leads and lags are calculated using the approach suggested by Croux, Forni and Reichlin (2001).
Leads and Lags (Details – 2)
Table 8. Lead-Lag Relations between the United States and Other Countries
Output
House prices
Residential Investment
Credit
Short-term interest rates
8-16 quarters 16-32 quarters 8-16 quarters 16-32 quarters 8-16 quarters 16-32 quarters 8-16 quarters 16-32 quarters 8-16 quarters 16-32 quarters
Australia
Austria
Belgium
Canada
Denmark
Finland
France
Germany
Ireland
Italy
Japan
Netherlands
New Zealand
Norway
Spain
Switzerland
United Kingdom
lag
lead
lead
lag
lead
lead
lead
lead
lead
lead
lead
lead
lag
contemp.
lead
lead
contemp.
lag
lead
lead
contemp.
lead
lead
lead
lead
lead
lead
lead
lead
contemp.
lead
lead
lead
contemp.
lead
lead
lead
lead
lead
lead
lead
lag
lead
lead
lag
lead
lead
lag
lead
lead
lead
contemp.
contemp.
lead
lead
contemp.
lead
lead
lag
contemp.
lead
lead
lag
lead
lag
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lag
lead
contemp.
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lag
lead
lead
lead
lead
contemp.
contemp.
contemp.
contemp.
contemp.
lead
lead
contemp.
lead
contemp.
lead
lead
contemp.
contemp.
lead
contemp.
contemp.
lead
lead
lead
lead
contemp.
lead
lead
lag
lead
lead
lead
lead
lead
lag
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
contemp.
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
lead
contemp.
lead
lead
lead
lead
Lags
Contemporaneous
Leads
18
12
71
6
18
76
18
0
82
18
24
59
6
6
88
6
0
94
0
65
35
12
6
82
0
0
100
0
12
88
Source: Authors' estimates.
Notes: For each pair of countries and the variable, entries in the table indicate whether the cyclical component in the United States leads or lags the cyclical component in the
respective country listed in the first column, or whether the relationship is contemporaneous, on average, over the frequency band. The numbers in the bottom indicate the
percentage of countries with a given type of relation. Leads and lags are calculated using the approach suggested by Croux, Forni and Reichlin (2001).
Conclusions
 House price cycles lead other cycles in
the long run
 Interest rates tend to lag other cycles at
all horizons
 Global factors are important for all
cycles, especially for credit and real
activity in the latter part of the sample
period
 U.S. leads other countries in all cycles
Similar conclusions from other research
 A six-variable VAR: real GDP, private consumption,
residential investment, CPI inflation, the nominal (short
term) interest rate and real house prices. 20 countries, sample
from 1986-2009.
 House price shocks are identified with a Cholesky
decomposition. In practice they look like “housing demand
shocks” because they lead to a strong comovement between
real house prices and residential investment. Alternative
identification through sign restrictions.
 VAR delivers average responses, but there could be
asymmetries between house price booms and busts.
Impact on Real GDP
of a 10% decline in real house prices
4
3
1.5
1
France
2
4
Spain
0.5
4
3
Australia
3
2
2
1
0
1
0
1
-0.5
0
-1
0
-1
-2
-4
-2
0 1 2 3 4 5 6 7 8 9 10 11 12
4
2
United States
1
2
3
4
5
6
7
8
9 10 11 12
2.0
1
0
1
2
3
4
5
6
7
8
9 10 11 12
7
8
9 10 11 12
3
Netherlands
2
New Zealand
1
1.5
2
1
-3
0
2.5
4
3
-2
-2
0 1 2 3 4 5 6 7 8 9 10 11 12
5
United Kingdom
-1
-1
-1.5
-3
3
Korea
0
1.0
-1
0.5
-2
0
0
-1
-2
-1
-3
0.0
-4
-3
-2
-0.5
-4
-3
-5
0 1 2 3 4 5 6 7 8 9 10 11 12
-5
-1.0
0 1 2 3 4 5 6 7 8 9 10 11 12
-6
0 1 2 3 4 5 6 7 8 9 10 11 12
0
1
2
3
4
5
6
Mortgage market characteristics matter
Average estimated impact
on GDP by VAR analysis
(in percent) Countries included in the calculation of average impact
Penalty-fee prepayment and refinancing
available
not available
-2.77 Australia, Denmark, Japan, United Kingdom, United States
Austria, Belgium, Canada, Finland, France, Germany, Greece, Italy,
-1.71
Netherlands, Norway, Spain, Sweden
Option to withdraw mortgage equity
available
not available
Typical loan-to-value ratio at origination
less than or equal to 60 percent
between 60 and 79 percent
more than or equal to 80 percent
Australia, Canada, Denmark, Finland, Greece, Ireland, Japan, Netherlands,
Norway, Sweden, United Kingdom, United States
-1.61 Austria, Belgium, France, Germany, Italy, Spain
-2.27
-0.06 Austria, Italy
-1.67 Canada, Finland, France, Germany, Norway, Spain, United Kingdom
Australia, Belgium, Denmark, Ireland, Japan, Netherlands, Sweden,
-2.21
United States
Policy Implications
 Can a uniform policy prescription of taking into
account asset prices in monetary policy making be
made? It seems not, statistical properties of house
prices vary across countries. In addition, some
shocks to house price inflation are persistent.
 The U.S. business, house price, and interest rate
cycles tend to lead the respective cycles in other
countries over all horizons and more so recently.
Questions domestically-focused economic policies.

similar documents