Transcript The international transmission of house price shocks Are there

The international transmission of
house price shocks
Are there contagion effects?
Olivier de Bandt (BdF)
Karim Barhoumi (BdF)
Catherine Bruneau (Paris X and BdF)
Transmission/contagion
• Transmission: reaction of house prices to
fundamentals in « normal times »,
including all available information
– Fundamentals are correlated
– Arbitrage behaviour across markets smooth out
idiosyncracies
• Contagion: 2 definitions:
– Amplitude of reaction differs in « crisis periods », with
possible non linearities
– Pandemic model: « from local to global and global to
local »
Data
• House prices from OECD and national
sources
• OECD quarterly national accounts : GDP,
inflation, short and long term interest rates,
housing investment
• 15 countries: AUS, CAN, CHE, FIN, FRA, DEU, IRE,
ITA, JPN, NLD, NOR, NZL, ESP, UK, US
Methods
• Single linear equations to estimate the link
between ‘local’ and ‘global’ levels,
including house prices
• Linear Favar models and causality tests
– to take into account endogeneity
– But need to accomodate the high persistence
of variables
• Crisis dummies and STAR models to
assess possible non linearities
Main Findings
• Contagion from US house prices, which
appear to be exogenous
• Spreading to the rest of the world,
according to the « pandemic view » of
contagion : common house prices
« Granger cause » domestic house prices
in Favar models
Plan
I – A closer look at the data
II- Empirical results
III- Conclusion
l- A closer look at the data (1/4)
Common SW’s house prices in OECD countries
I- A closer look at the data (2/4)
Using a larger database : fac1 correlated with
interest rates
I- A closer look at the data (3/4)
Using a larger database : fac2 correlated with GDP
growth)
I- A closer look at the data (4/4)
Using a larger database : fac3 corr. with OGAP
II- Empirical results
• 1- single one period ahead equation with
global house price factor
• 2- single one period ahead equation with
crisis dummy
• 3-single one period ahead equation with
other global factors
• 4- single non linear (LSTAR) with all global
factors
• 5- causality tests in Favar models
1- Single one period ahead linear equations
=> international housing factor is significant
in many countries: AUS, ESP, UK
2- robustness to Financial Crisis
periods (Reinhart & Rogoff, 2008)
Tab2: Robustness to crisis periods
3- Tab3: Sensitiveness to global factors in single
one period ahead equation
4-LSTAR models: contemporaneous impact of the
threshold variable in the two regimes
5-Causality in favar models of
reduced order: US house prices are
exogeneous and affect Common house
prices
Causality in favar models of
reduced order: Other domestic house
prices are affected by Common house
prices
Causality from systems : Other
domestic house prices are affected by
Common house prices
III-Conclusion
• Evidence in favour of international
transmission
• Evidence in favour of « pandemic model »
with contagion from USA to rest of
countries