Transcript file

The Impact of Low Income Home Owners
on the Volatility of Housing Markets
Peter Westerheide
ZEW
European Real Estate Society Conference 2009
Stockholm
June 26, 2009
1
Outline
 Motivation
 Literature Review
 Data
 Methodology
 Empirical Results
 Conclusions and Outlook
2
Motivation
 Default of low income home owners (subprime borrowers) has
triggered the current crisis
 Usual focus is on the behavior of lenders, not of borrowers
• Lending without properly checking creditworthiness
• „Originate and distribute“
 Our Focus: What is the role of low income housing demand?
• Can we observe any destabilizing impact of low income
home ownership in the long run?
3
Motivation
 Arguments in favor of destabilizing impact:
Low income households
• face a higher income risk of unemployment
• have low liquid wealth and often no buffer stock to
compensate income fluctuations
• are usually highly leveraged and have high interest/debt
burdens
• cannot rely on bailout by relatives
4
Motivation
 Competing arguments: Low income households are
• less mobile (relative income position and regional location) ->
„trading up the ladder“ is less likely
• Wages are less variable than other components of income
therefore total income might be less variable
(as long as employed)
 Net effect depends on institutional framework, might be different
in the housing cycle
 Political target high home ownership rate: Is there a potential
tradeoff to the stability of housing markets?
5
Literature Review
Income
Mortgage
market
Housing
demand
Housing
prices
Volatility of Access to
income in
mortgage markets Structure of
different
Extent of low
income homeownership
depends on
Financial
leverage
demand and
income
completeness
marketsof markets
and
volatility of mortgage
volatility
(Chiuri/Japelli 2003, Bicacova/Siermienska 2007)
classes
Sensitivity
of prices
to income
Low income homeowners
increase
price
volatilityshocks depends on
Income
volatility
andare correlated
Level and
volatility
of income
leverage
(Lamont/Stein
1999, Benito 2006)
(Ortalo-Magné/Rady
2002)
probability
mortgage
(Diaz-Serrano
2004,ofDynan/Elmendorff/Sichel
2005, markets
Down
payment
constraints
affect the stability of housing
default 2008)2005)
(Ortalo-Magné/Rady
Jensen/Shore
Volatility of house prices depends on tax wedges (van den Noord 2005)
Income volatility is related to mortage default (Diaz-Serrano 2004)
6
Our approach
 Direct analysis of impact of low income home ownership on
house price volatility in a cross country comparison
 13 OECD countries, 1970-2006
 Panel Regression with Fixed Effects
 Pooled VAR
7
Data
 Most important problem:
Data on distribution of home ownership and income
• No micro data for long time horizons
• Assumption of fixed distribution not realistic
• But: home ownership rate might be used
as a proxy for share of low income households
• Empirical evidence supports this assumption
8
Data
Rate of Home Ownership
Stylized fact: home ownership rate and
income distribution
HHO Country
LHO Country
Income
9
Data
Ratio of home ownership 2nd/9th income decile
and average home ownership rate
R atio of H omeowners hip R ate
2nd/9th Inc ome D ec ile
6
5
GE
4
US
3
FIN
UK
2
IT
1
2
R = 0.7267
0
0%
10%
20%
30%
40%
50%
60%
70%
Averag e H ome Owners hip R ate
Own calculations, based on Bicacova /Siermienska (2007). Household head/spouse18-40 years old.
10
Data
Ratio of home ownership 3rd/9th income decile
and average home ownership rate
R atio of H om eowners hip R ate
3rd/9th Inc om e D ec ile
4.5
GE
4
3.5
3
US
2.5
FIN
2
UK
1.5
1
IT
0.5
R 2 = 0.8705
0
0%
10%
20%
30%
40%
50%
60%
70%
Averag e Hom e owners hip rate
Own calculations, based on Bicacova /Siermienska (2007). Household head/spouse18-40 years old.
11
Data
 Proxy Homeownership Rate
• No annual data (except UK and US)
• Survey based, 4 – 10 years
• Common definition: Share of owner occupied
dwellings in all dwellings
• Estimation of long term trends necessary
 Assumption: high inertia
 Short term fluctuations mainly reflect measurement error
12
Data
 House Prices
• OECD Database, Girouard et al. 2006
• Heterogeneous data, focussing mostly on
used family homes
 Other data OECD and UN, gaps filled by interpolation
based on national data:
• GDP per capita, long term interest rate, CPI, unemployment
rate, debt/GDP-ratio
13
Descriptive Evidence
Increase of Ownership Ratio 1970-2006
(Percentage Points)
25.0
IT
UK
NL
20.0
ES
15.0
FR
IRL
10.0
SE
GE
CH
FIN
CA
5.0
US
R² = 0.3876
JP
0.0
0.0%
2.0%
4.0%
6.0%
8.0%
10.0%
12.0%
Standard Deviation: Annual Percentage Change of Real House Prices
14
Descriptive Evidence
90.0
Average Ownership Ratio 1970-2006
(%)
80.0
ES
IRL
70.0
US
FIN
CA
JP
60.0
FR
50.0
IT
UK
SE
NL
GE
40.0
CH
30.0
20.0
10.0
0.0
0.0%
R2 = 0.1857
2.0%
4.0%
6.0%
8.0%
10.0%
12.0%
Standard Deviation of Annual Percentage Change of Real House Prices
15
Methodology
 Multivariate Regression: Volatility of house prices
• Volatility of Real Income Growth
(proxied by Real GDP per Capita)
Income
• Unemployment rate
• Interest Rate
• Debt-to-GDP ratio
Credit Market
• House Price growth
• Volatility of Inflation Rate (CPI)
Prices
• Home ownership rate
16
Methodology
 Time Period 1970 – 2006
• Problem: Measure volatility in sub periods that are
long enough to show substantial variation…
• …but short enough to have sufficient data points in
the time series dimension
• Tests with several period lengths
• Finally: 11 overlapping periods of 7 years
17
Econometric Results
Results of Regression for Nominal House Price Volatility
Dependent Variable:  Volatility of Nominal House Prices
P>|t|
95 % confidence
interval
-0,12
0,904
-0,450
0,398
0,364
4,01
0,000
0,737
2,184
0,709
0,309
2,29
0,024
0,095
1,322
-0,327
0,244
-1,34
0,183
-0, 81
0,156
 Debt to GDP Ratio
0,041
0,034
1,20
0,234
0,026
0,108
 Growth Rate of Nominal
House Prices
0,261
0,081
3,20
0,002
0,099
0,422
 Ownership ratio
0,780
0,344
2,27
0,025
0,098
1,463
-0,005
0,005
-1,02
0,311
-0,016
0,005
Coeff.
Std. Error
t-value
-0,026
0,214
 Volatility of GDP
1,460
 Interest Rate
 Volatility of CPI
 Unemployment Rate
Const.
R2 = 0,366
117 Observations
(9 periods/13 countries)
Fixed effects regression, adjusted for serial correlation.
18
Econometric Results
Results of Regression for Real House Price Volatility
Dependent Variable;  Volatility of Real House Prices
95 % confidence
interval
Coeff.
Std. Error
t-value
P>|t|
-0,079
0,181
-0,44
0,663
-0,439
0,280
 Volatility of GDP
1,896
0,321
5,91
0,000
1,259
2,532
 Real Interest Rate
0,122
0,193
0,63
0,530
-0,261
0,505
 Unemployment Rate
-0,194
0,216
-0,9
0,372
-0,623
0,235
 Debt to GDP Ratio
0,052
0,027
1,93
0,057
0,002
0,105
 Growth Rate of Real
House Prices
0,174
0,067
2,62
0,010
0,042
0,306
 Ownership ratio
0,504
0,278
1,82
0,072
0,046
1,057
-0,013
0,004
-3,28
0,001
-0,020
-0,005
 Volatility of CPI
Const.
R2 = 0,332
117 Observations
(9 periods/13 countries)
Fixed effects regression, adjusted for serial correlation.
19
Methodology
 Specification of a VAR with annual values 1970-2006
•
Are house prices more sensitive to shocks in HHO
vs. LHO countries?
•
Variables: real house prices, real interest rates,
real GPD per capita, ownership ratio
•
2 lags
20
Econometric Results
Reaction to House Prices
Reaction to GDP
4
3
3.5
2.5
3
2
2.5
2
1.5
1.5
1
GDP OR < 60
1
0.5
RHP OR < 60
0.5
GDP OR >= 60
0
RHP OR >= 60
0
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Reaction to Interest Rates
0.01
0
-0.01
LI OR < 60
-0.02
LI OR >=60
-0.03
-0.04
-0.05
-0.06
-0.07
1
2
3
4
5
6
7
8
9
10
21
Conclusion
 Ownership rate is proxy for share of low income home
owners
 Some evidence for a positive correlation of home
ownership rate and volatility of house prices
 Confirms theoretical findings
22
Outlook
 Inclusion of better credit market indicator (mortgage
market index)
 Account for age structures/demographic structures
 Test with US micro data for different regions?
 Refine estimation of VAR
…
23
Thank you for your attention!!!
Contact:
Peter Westerheide
++49 621 1235 146
[email protected]
24