Housing Cycles and Macroeconomic Fluctuations: A Global Perspective Ambrogio Cesa-Bianchi1,2 2
1 Università Cattolica di Milano Inter–American Development Bank
24th July 2012
1 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Three lessons from the global financial crisis
2 / 34
I
Housing market fluctuations may induce business fluctuations in advanced (AEs) and emerging (EMEs) economies alike
I
Increased importance of emerging economies for global growth
I
Movements in AEs house prices may display high international comovement
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Main questions
I
This paper answers the following questions 1. Are international housing prices really correlated across countries? Is there a common factor driving a global housing cycle? 2. How are house price shocks transmitted to the real economy? I I
3 / 34
What is the impact of a US housing demand shock on domestic and foreign GDP? Do synchronized housing demand shocks reinforce each other? How do they compare to other asset price shocks?
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Main questions
I
This paper answers the following questions 1. Are international housing prices really correlated across countries? Is there a common factor driving a global housing cycle? 2. How are house price shocks transmitted to the real economy? I I
3 / 34
What is the impact of a US housing demand shock on domestic and foreign GDP? Do synchronized housing demand shocks reinforce each other? How do they compare to other asset price shocks?
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
How?
4 / 34
I
Global VAR model for the world economy
I
New house price data set and other macro–financial variables for 33 AEs and EMEs covering more than 90 percent of world GDP
I
Impulse response functions to identified housing demand shocks for the investigation of the spillover effects
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
How?
4 / 34
I
Global VAR model for the world economy
I
New house price data set and other macro–financial variables for 33 AEs and EMEs covering more than 90 percent of world GDP
I
Impulse response functions to identified housing demand shocks for the investigation of the spillover effects
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
How?
4 / 34
I
Global VAR model for the world economy
I
New house price data set and other macro–financial variables for 33 AEs and EMEs covering more than 90 percent of world GDP
I
Impulse response functions to identified housing demand shocks for the investigation of the spillover effects
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Filling the gap
I
Empirical models on the international transmission of housing shocks IMF (2004), Otrok and Terrones (2005), and Beltratti and Morana (2010), use dynamic factor models to understand international linkages between "global" housing factors and macroeconomic fluctuations ! EMEs are not considered!
I
Identification of shocks in the GVAR Literature has so far relied on generalized impulse response functions to non-identified disturbances ! Not adequate for analysis of financial shocks!
5 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Filling the gap
I
Empirical models on the international transmission of housing shocks IMF (2004), Otrok and Terrones (2005), and Beltratti and Morana (2010), use dynamic factor models to understand international linkages between "global" housing factors and macroeconomic fluctuations ! EMEs are not considered!
I
Identification of shocks in the GVAR Literature has so far relied on generalized impulse response functions to non-identified disturbances ! Not adequate for analysis of financial shocks!
5 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Preview of the results
6 / 34
I
Real house price returns can display high cross–country correlation, in particular when considering AEs and EMEs separately
I
US house price shocks have strong impact on AEs real economy but not on EMEs
I
Synchronized regional shocks to asset prices reinforce each other and have deeper and more long-lasting impact than country-specific shocks, especially in the case of house prices
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Preview of the results
6 / 34
I
Real house price returns can display high cross–country correlation, in particular when considering AEs and EMEs separately
I
US house price shocks have strong impact on AEs real economy but not on EMEs
I
Synchronized regional shocks to asset prices reinforce each other and have deeper and more long-lasting impact than country-specific shocks, especially in the case of house prices
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Preview of the results
6 / 34
I
Real house price returns can display high cross–country correlation, in particular when considering AEs and EMEs separately
I
US house price shocks have strong impact on AEs real economy but not on EMEs
I
Synchronized regional shocks to asset prices reinforce each other and have deeper and more long-lasting impact than country-specific shocks, especially in the case of house prices
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Outline
1. Are Housing Cycles Really Correlated? Three Facts 2. GVAR Model 3. Identification Strategy 4. Results 5. Conclusion
7 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Fact 1. Taking into account EMEs can make a difference Figure: Real House Price Indices (Median across all series within each group; constant prices; index; sample period: 1990-Q1 to 2009-Q4) 140
120 110
120
100 100 90 80 80 60
8 / 34
Global HP AEs HP EMs HP
70
Global HP US HP China HP
40 90 92 94 96 98 00 02 04 06 08 10
60 90 92 94 96 98 00 02 04 06 08 10
(a) Global, AEs, and EMEs
(b) Global, US, and China
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Fact 2. Synchronization of house prices varies significantly over time Figure: Average Moving Pair-wise Correlation of Annual House Price Inflation (Constant prices; annual growth rates; sample period: 1990-Q1 to 2009-Q4) 0.5
0.5
0.5
0.5
0.4
0.4
0.4
0.4
0.3
0.3
0.3
0.3
0.2
0.2
0.2
0.2
0.1
0.1
0.1
0.1
0
0
−0.1 94
−0.1 97
00
03
06
09
(a) AEs
9 / 34
0
0
−0.1 94
−0.1 97
00
03
06
(b) EMEs
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
09
Fact 3. Both global and group–specific factors are likely to be important Figure: Principal Component Analysis on Annual House Price Inflation (Explained variance of the principal components; constant prices; annual growth rates; sample period: 1990-Q1 to 2009-Q4)
Variance Explained (%)
(a) ALL
(c) EMEs
60
60
50
50
50
40
40
40
30
30
30
20
20
20
10
10
10
0
10 / 34
(b) AEs
60
1
2
3
0
1
A. Cesa-Bianchi
2
3
0
1
2
3
Housing Cycles and Macroeconomic Fluctuations
Outline
1. Are Housing Cycles Really Correlated? Three Facts 2. GVAR Model 3. Identification Strategy 4. Results 5. Conclusion
11 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
GVAR model - Pesaran et al., (2004, 2007, 2010) Two steps
I
Step 1. Estimation of N country specific models Augmented VAR: domestic variables are related to country–specific foreign (weak exogenous) variables Estimate each model allowing for cointegration
I
Step 2. Solution of the global model Collect all the endogenous variables in a global vector Get a reduced form VAR–style model
12 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
GVAR model - Pesaran et al., (2004, 2007, 2010) Two steps
I
Step 1. Estimation of N country specific models Augmented VAR: domestic variables are related to country–specific foreign (weak exogenous) variables Estimate each model allowing for cointegration
I
Step 2. Solution of the global model Collect all the endogenous variables in a global vector Get a reduced form VAR–style model
12 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
GVAR model Step 1. Country specific VARX model I
Country specific VARX(1, 1) model for the ith economy xi,t = Φi xi,t
I
1
+ Λ0i xi,t + Λ0i xi,t
1
+ uit ,
Country specific foreign variables xi,t defined as cross–section averages with fixed weights N
xi,t =
∑ Wij xj,t = Wi xt
j=0
0 , x0 , ..., x0 )0 is the vector of all endogenous variables xt = (x0,t 1,t N,t
Wi = (Wi0 , Wi1 , ..., WiN ) is the ki
13 / 34
A. Cesa-Bianchi
k matrix of fixed weights
Housing Cycles and Macroeconomic Fluctuations
GVAR model Step 1. Country specific VARX model I
Country specific VARX(1, 1) model for the ith economy xi,t = Φi xi,t
I
1
+ Λ0i xi,t + Λ0i xi,t
1
+ uit ,
Country specific foreign variables xi,t defined as cross–section averages with fixed weights N
xi,t =
∑ Wij xj,t = Wi xt
j=0
0 , x0 , ..., x0 )0 is the vector of all endogenous variables xt = (x0,t 1,t N,t
Wi = (Wi0 , Wi1 , ..., WiN ) is the ki
13 / 34
A. Cesa-Bianchi
k matrix of fixed weights
Housing Cycles and Macroeconomic Fluctuations
GVAR model Step 2. The global model - Combining the VARX
I
Define a ki
k selection matrix Si such that xi,t = Si xt
Si xt Gi xt where Gi = Si I
= Φi Si xt 1 + Λi0 Wi xt + Λi1 Wi xt = Hi xt 1 + uit ,
1
+ uit ,
Λi0 Wi and Hi = Φi Si + Λi1 Wi
and stack each country-specific model for i = 0, 1, ..., N Gxt = Hxt
1
+ ut ,
0 )0 , and H = (H 0 , H 0 , ..., H 0 )0 where G = (G00 , G10 , ..., GN 0 1 N
14 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
GVAR model Step 2. The global model - Combining the VARX
I
Define a ki
k selection matrix Si such that xi,t = Si xt
Si xt Gi xt where Gi = Si I
= Φi Si xt 1 + Λi0 Wi xt + Λi1 Wi xt = Hi xt 1 + uit ,
1
+ uit ,
Λi0 Wi and Hi = Φi Si + Λi1 Wi
and stack each country-specific model for i = 0, 1, ..., N Gxt = Hxt
1
+ ut ,
0 )0 , and H = (H 0 , H 0 , ..., H 0 )0 where G = (G00 , G10 , ..., GN 0 1 N
14 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
GVAR model Model Setup I I
Estimation period: 1983Q1 to 2009Q4 33 country–specific VARX models Table: Countries and Regions in the GVAR Code Advanced Economies Australia Austria Belgium Canada Finland France Germany Italy Japan
15 / 34
Netherlands Norway New Zealand Spain Sweden Switzerland UK US
A. Cesa-Bianchi
Emerging Economies Argentina Brazil China Chile India Indonesia Korea Malaysia Mexico
Peru Philippines South Africa Saudi Arabia Singapore Thailand Turkey
Housing Cycles and Macroeconomic Fluctuations
GVAR model VARX* Specification I
Country-specific models include the following endogenous and foreign variables Table: Variables Specification of the Country-specific VARX* Models Non-US Models
16 / 34
US Model
Domestic
Foreign
Domestic
Foreign
yi πi qi hpi ρSi ρLi (e p)i -
yi πi qi hpi ρSi ρLi po
yUS π US qUS hpUS ρSUS ρUS po
yUS π US hpUS (e p)US -
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Outline
1. Are Housing Cycles Really Correlated? Three Facts 2. GVAR Model 3. Identification Strategy 4. Results 5. Conclusion
17 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Identification of shocks in the GVAR
I
Literature has largely relied on Generalized Impulse Response Functions (Koop et al., 1996) GIRFs are obtained by ordering the shocked variable first in a recursive VAR ! Improper for asset price shocks
I
Proposed identification procedure consist of two steps A set of orthogonal country–specific shocks is derived following Sims (1980) The identified shocks are coherently introduced in the GVAR model
18 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Identification of shocks in the GVAR
I
Literature has largely relied on Generalized Impulse Response Functions (Koop et al., 1996) GIRFs are obtained by ordering the shocked variable first in a recursive VAR ! Improper for asset price shocks
I
Proposed identification procedure consist of two steps A set of orthogonal country–specific shocks is derived following Sims (1980) The identified shocks are coherently introduced in the GVAR model
18 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Identification of shocks in the GVAR Step1. Within–Country Identification I
Housing demand shock Real house price increase Nominal short-term interest rate does not fall ! To rule out expansionary monetary policy shocks No contemporaneous impact on GDP and CPI inflation ! To rule out more fundamental expansionary shocks
I
Operationally, the identification is achieved with a standard Cholesky decomposition
I
For the ordering, I closely follow the literature (Musso et al., 2012, Aspachs and Rabanal, 2011) xit = yi0 , π i0 , rSi , hpi0 , rLi 0 , (e
19 / 34
A. Cesa-Bianchi
p)i0 , qi0
0
Housing Cycles and Macroeconomic Fluctuations
Identification of shocks in the GVAR Step1. Within–Country Identification I
Housing demand shock Real house price increase Nominal short-term interest rate does not fall ! To rule out expansionary monetary policy shocks No contemporaneous impact on GDP and CPI inflation ! To rule out more fundamental expansionary shocks
I
Operationally, the identification is achieved with a standard Cholesky decomposition
I
For the ordering, I closely follow the literature (Musso et al., 2012, Aspachs and Rabanal, 2011) xit = yi0 , π i0 , rSi , hpi0 , rLi 0 , (e
19 / 34
A. Cesa-Bianchi
p)i0 , qi0
0
Housing Cycles and Macroeconomic Fluctuations
Identification of shocks in the GVAR Step2. GVAR Identification
I
Let P0 be lower triangular Cholesky factor of the residuals covariance matrix of country 0, then the GVAR model can be written as Gxt = Hxt 1 + PG vt . where vt = (PG ) 2
P0 60 6 PG = 6 . 4 .. 0
20 / 34
0 Ik1 .. 0
.
1u t
0 0 .. . IkN
3
7 7 7, 5
and 2
3 v0 6 u1 7 6 7 vt = 6 . 7 , 4 .. 5 uN
A. Cesa-Bianchi
2
I
6 Σu1 v0 6 Σv = 6 . 4 ..
ΣuN v0
Σv0 u1 Σu1t .. ΣuN u1
Housing Cycles and Macroeconomic Fluctuations
.
3 Σv0 uN Σu1 uN 7 7 .. 7 . 5 ΣuNt
Outline
1. Are Housing Cycles Really Correlated? Three Facts 2. GVAR Model 3. Identification Strategy 4. Results 5. Conclusion
21 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Relevant shocks
I
Consider one country specific shocks and two regional shocks A US housing demand shock A synchronized housing demand shock originated in AEs A synchronized equity price shock originated in AEs
22 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
The US housing demand shock leads to an expansion in the US economy
Figure: US House Price Shock - Transmission to the US Economy USA GDP
USA INFLATION
USA SHORT INT. RATE 0.15
0.15
1
0.1 0.5
0.1
0.05 0.05 0
0 4
8
12
16
4
USA HOUSE PRICE
8
12
16
2 1.5 1 0.5 8
12
16
8
16
4 2 0 −2 4
8
12
16
4
8
US House Price
23 / 34
12
USA EQUITY
0.08 0.06 0.04 0.02 0 −0.02 −0.04
2.5
4
4
USA LONG INT. RATE
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
12
16
US housing demand shock has significant spillovers on AEs...
Figure: US Housing Demand Shock - Transmission to AEs Real GDP USA GDP
CANADA GDP
UK GDP
1
1
1 0.5
0.5 0.5 0
0 4
8
12
16
0
4
JAPAN GDP
8
12
16
4
GERMANY GDP
8
12
16
12
16
ITALY GDP
2
1.5 1.5 1
1
1
0.5
0.5 0
0 4
8
12
16
0 4
8
12
16
4
8
US House Price
24 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
...while EMEs response is heterogeneous
Figure: US Housing Demand Shock - Transmission to EMEs Real GDP CHINA GDP
INDIA GDP
TURKEY GDP
0.4 1 0.5 0.2
0.5
0
0 0 −0.5
−0.5 4
8
12
16
4
BRAZIL GDP
8
12
16
4
MEXICO GDP
8
12
16
SOUTH AFRICA GDP
1.5 1 1
0.5
0.5
0
0.5
0 −0.5
−0.5 4
8
12
16
0 4
8
12
16
4
8
US House Price
25 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
12
16
Synchronized AEs house price shocks have deeper impact than equity price... Figure: AEs Regional House Price and Equity Price Shock - Transmission to AEs Real GDP USA GDP
CANADA GDP
UK GDP
0.6
0.6
0.4
0.4
0.2
0.2
0.4 0.2 0
0
0 4
8
12
16
4
JAPAN GDP
8
12
16
4
GERMANY GDP
1
1
0.5
0.5
0
0
8
12
16
12
16
ITALY GDP 0.6 0.4 0.2
4
8
12
16
0 4
AE House Price Shock
26 / 34
A. Cesa-Bianchi
8
12
16
4
8
AE Equity Price Shock
Housing Cycles and Macroeconomic Fluctuations
...while EMEs evidence again is mixed
Figure: AEs Regional House Price and Equity Price Shock - Transmission to EMEs Real GDP CHINA GDP
INDIA GDP
0.4 0.2
0.1
0
0
−0.2 4
8
12
16
−0.1
4
BRAZIL GDP
8
12
16
0.4
0.6
0.2
0.4
12
16
16
0.2
−0.2 8
12
0.4
0
−0.2
8
SOUTH AFRICA GDP
0.2
4
4
MEXICO GDP
0
0 4
AE House Price Shock
27 / 34
TURKEY GDP 0.6 0.4 0.2 0 −0.2 −0.4
0.2
A. Cesa-Bianchi
8
12
16
4
8
AE Equity Price Shock
Housing Cycles and Macroeconomic Fluctuations
12
16
Conclusions
I
This paper investigates international housing cycles in advanced and emerging economies
I
Main findings: 1. House price returns can be highly correlated at global level, but even more at group–specific level 2. A US housing demand shock is transmitted to most AEs but has no impact on four large EMEs 3. Synchronized AEs housing shocks reinforce each other and have deeper and more long-lasting impact than country-specific shocks
28 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Conclusions
I
This paper investigates international housing cycles in advanced and emerging economies
I
Main findings: 1. House price returns can be highly correlated at global level, but even more at group–specific level 2. A US housing demand shock is transmitted to most AEs but has no impact on four large EMEs 3. Synchronized AEs housing shocks reinforce each other and have deeper and more long-lasting impact than country-specific shocks
28 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Conclusions
I
This paper investigates international housing cycles in advanced and emerging economies
I
Main findings: 1. House price returns can be highly correlated at global level, but even more at group–specific level 2. A US housing demand shock is transmitted to most AEs but has no impact on four large EMEs 3. Synchronized AEs housing shocks reinforce each other and have deeper and more long-lasting impact than country-specific shocks
28 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Conclusions
I
This paper investigates international housing cycles in advanced and emerging economies
I
Main findings: 1. House price returns can be highly correlated at global level, but even more at group–specific level 2. A US housing demand shock is transmitted to most AEs but has no impact on four large EMEs 3. Synchronized AEs housing shocks reinforce each other and have deeper and more long-lasting impact than country-specific shocks
28 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Implications
I
Regionalization hypothesis advanced by Hirata, Kose and Otrok (2011) Business cycle synchronization has increased among AEs and among EMEs separately while the relative importance of the global factor has declined Some EMEs have become more resilient to shocks originated in AEs
I
Decreased importance of US shocks in the global economy (Yeyati and Williams, 2012) Rather than decoupling from the world economy, many EMEs shifted their loading from the US into other EMEs
29 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Implications
I
Regionalization hypothesis advanced by Hirata, Kose and Otrok (2011) Business cycle synchronization has increased among AEs and among EMEs separately while the relative importance of the global factor has declined Some EMEs have become more resilient to shocks originated in AEs
I
Decreased importance of US shocks in the global economy (Yeyati and Williams, 2012) Rather than decoupling from the world economy, many EMEs shifted their loading from the US into other EMEs
29 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Thank you
30 / 34
A. Cesa-Bianchi
Housing Cycles and Macroeconomic Fluctuations
Additional results Evolution of trade flows
Figure: Evolution of China’s Trade Flows 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
31 / 34
83
86 Euro Area
89 Japan
92 US
A. Cesa-Bianchi
95 98 Rest of Asia
01 Other AEs
04 07 Other EMEs
Housing Cycles and Macroeconomic Fluctuations
10
Additional results Evolution of trade flows
Figure: Evolution of Brazil’s Trade Flows 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
32 / 34
83
86 Euro Area
89 Japan
92 US
A. Cesa-Bianchi
95 98 Rest of Asia
01 Other AEs
04 07 Other EMEs
Housing Cycles and Macroeconomic Fluctuations
10
Robustness Alternative Orderings of US Variables
Figure: US Housing Demand Shock USA GDP
CANADA GDP
UK GDP
1
1
1
0.5
0.5 0.5
0
0
0 4
8
12
16
−0.5 4
JAPAN GDP
8
12
16
4
GERMANY GDP
8
12
16
FRANCE GDP
2 1 1
1 0 −1
4
8
12
0.5
0
0
−1
−0.5
16
4
ITALY GDP
8
12
16
4
SPAIN GDP
8
12
16
SWITZERLAND GDP
1.5 1
1
0.5
0.5
0
0
0
−0.5
−0.5 4
33 / 34
1
8
12
16
4
A. Cesa-Bianchi
8
12
16
−1
4
8
Housing Cycles and Macroeconomic Fluctuations
12
16
Robustness Sample VS Block Diagonal Covariance Matrix
Figure: US Housing Demand Shock USA GDP
CANADA GDP
1.5
UK GDP
2
1.5
1
0.5
1
1
0.5 0
0 0
−0.5 4
8
12
16
−0.5 4
JAPAN GDP
8
12
16
4
GERMANY GDP
3
3
2
2
1
1
12
16
2 1
0
0
8
FRANCE GDP
0
−1 4
8
12
16
4
ITALY GDP
12
16
0 8
12
16
12
16
1.5 1 0.5 0 −0.5
1
0
8
SWITZERLAND GDP
2
1
4
4
SPAIN GDP
2
34 / 34
8
4
A. Cesa-Bianchi
8
12
16
4
8
Housing Cycles and Macroeconomic Fluctuations
12
16
