Transcript 投影片 1 - Property Finance

An International Comparative
Study on Housing
Price and Property Tax Liabilities
Sun-Tien Wu, Chun-Hao Yueh
The 17th ERES Annual Conference,
June 23-26, 2010, Milano, Italy
1
Contents of presentation
• Purpose of this study
• The conceptual framework of the study
• Empirical results
• Discussions
• Concluding Remarks
2
Motivation
The increasing house price trend in
U.S., U.K., Taiwan and H.K. is
concomitant with decreasing
interest rates and financial
deregulation since 1990’s up to the
eruption of financial catastrophe two
years ago. Does growing financial
integration contribute to a common
trend among the housing markets?
3
The house price index of
selected countries
350
300
250
200
150
100
50
1991Q1 1993Q1 1995Q1 1997Q1 1999Q1 2001Q1 2003Q1 2005Q1 2007Q1 2009Q1
UK
US
TW
CAN
KOR
HK
4
‧The figures above indicate that, the
selected countries share a common
upward trend of the housing price
since 1999.
‧However, each country has
respective high or low upward
trend and variation itself.
‧Question: what factors other than
the above common trend might
influence the housing price?
5
Purpose of the Study
‧To investigate whether there is a
common trend of housing price index
among the selected countries.
‧To find out other factors such as
property tax and interest rate which
might have influences on the housing
price index.
6
The Conceptual Framework
of the Study
• We follow Otrok and Terrones
(2005) to construct a dynamic
factor model (DFM) comprising
one common state variable –
world common trend, and other
independent country-specific
factors to capture the house price
trend of selected countries.
7
• The independent country-specific
factors include per-capita output,
mortgage interest rate and
effective tax rate which was
measured as the property tax
burden of each housing unit for
our selected countries.
8
The reason for including
property tax burden
• As well documented in housing
research literature, property values
are negatively affected by their
related tax liabilities, a phenomenon
that is often termed as capitalization
in the literature. (e.g. Palm and smith,
1998; Krantz et al., 1982; and Lea,
1982)
9
• The model is defined as follows:
i
HPti  c  i ft  byi Yti  binti INTti  btax
Taxti  ti for i  1...n and t  1...T ,
• where ti represents a idiosyncratic
component, which is distributed as
i.i.d N (0,  i2 )
10
• The common factor evolves as an
independent AR(p) process:
ft   ( L) ft  t
Following Otrok and Terrones（2005）,
we fix the variance of this innovation
to unity as a normalization of the
i.i.d .
model: t 
 N (0,1)
11
Figure 1: The original series
for selected OECD countries
House price index
GDP per capita(log)
260
1 0 .4 5
1 0 .0 0
240
1 0 .4 0
9 .9 2
220
1 0 .3 5
200
9 .8 4
1 0 .3 0
9 .7 6
180
1 0 .2 5
160
9 .6 8
1 0 .2 0
140
9 .6 0
1 0 .1 5
120
US
US
100
Canada
1 0 .1 0
80
9 .5 2
Canada
UK
UK
1 0 .0 5
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
Interest rate
14
9 .4 4
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
Effective tax rate
3 .0
US
US
Canada
Canada
12
UK
UK
2 .5
10
2 .0
8
6
1 .5
4
1 .0
2
0
0 .5
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
12
• The Figure shows that the trend for
house prices in U.S., U.K. and
Canada are pretty resembling to
each other.
• The movements of per capita GDP
and mortgage interest rate also look
similar.
• This indicates that there must be a
common trend among these selected
OECD countries.
13
• The effective property tax rate, on the
contrary, are quite different for those
countries, with U.S. and U.K. having
the highest and lowest value, while
Canada’s was in between.
• However, during our observation
period, the movement of this variable,
by and large exhibits a huntchbacked
pattern except for the last two years.
14
Estimation Result(I):
Model results for selected OECD countries
U.S.
restricted
model
Common Factor（
f t ）0.525***
(0.006)
U.K.
Canada
full
model
restricted
model
full
model
restricted
model
full
model
0.680***
(0.011)
0.707***
(0.010)
0.809***
(0.014)
0.533***
(0.007)
0.635***
(0.014)
GDP per capita（ Yt ）
110.360***
(1.385)
-1.086
(1.219)
-7.730***
(1.286)
Interest Rate（ Intt ）
-1.514***
(0.039)
-1.596***
(0.063)
-0.605***
(0.039)
Dewelling Tax（ Taxt ）
-2.381***
(0.176)
-34.261***
(1.567)
-25.692***
(1.099)
without country-specific factors
State space model
ft  0.905 ft 1  t
(0.017)
with country-specific factors
ft  0.863 ft 1  t
(0.063)
Note 1: Dependent variables are the house price index of selected countries.
Note 2: Constant terms are not reported.
Note 3:*, **, ***represents 10%, 5%, and 1% significance level, respectively.
15
• Estimation begins with the restricted
model where the model is simplified
that only unobservable common trend is
considered:
HPt  c   ft  
i
i
i
t
for i  1...n and t  1...T .
• The results show that the common trend
is highly persistent and significant.
Besides, the estimated  i for each
selected OECD countries is also
significantly positive.
16
• This result mimics the finding of Chirinko
et al. (2004) and that of Otrok and
Terrones (2005) and may be interpreted
as the fact that the U.S., U.K. and Canada
have relative high level of financial
openness so that there exists an evident
common trend even though real estates
are quintessential non-tradable goods.
17
• Then, we add the country- specific
factors to the selected countries that
might have affected each country’s
house price index besides their
common trend.
• The result shows that house prices in
selected OECD countries moved
significant negatively with mortgage
interest rate and property tax liabilities,
implying that the capitalization effect
works.
18
The actual and fitted house price
for selected OECD countries
US
UK
260
220
240
200
220
180
200
160
180
140
160
120
140
real
120
real
100
state
state
state and others
100
state and others
80
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
Canada
170
160
150
140
130
120
110
real
100
state
state and others
90
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
19
• The figure shows the common trends of
the house price in U.S., U.K. and Canada
do not make significant difference
whether the country-specific factors are
included in the estimation or not.
• This implies that country-specific factors
in those countries might have been
affected by their international
counterparts, as Otrok and Terrones
(2005) asserted.
20
Figure 2: The original series for
selected East Asia countries
House price index
350
GDP per capita(log)
11.6
14.9
14.8
300
11.4
14.7
250
11.2
200
11.0
14.6
14.5
14.4
150
10.8
14.3
14.2
100
Taiwan
HK
Korea
50
10.6
Taiwan
HK
10.4
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
Interest rate
17.5
Taiwan
HK
Korea
15.0
14.1
Korea
14.0
1991 1993 1995 1997 1999 2001 2003 2005 2007 2009
Effectiv e tax rate
0.40
Taiwan
HK
0.35
0.30
12.5
0.25
10.0
0.20
7.5
0.15
5.0
0.10
2.5
0.05
0.0
0.00
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
21
Observation points
-case of East Asia country
• House price moved quite differently
for East Asia Countries.
• The discernible common trend for
house prices is Less obvious than the
case of the selected OECD countries.
22
• Unlike the OECD case, the different
movement patterns here implies that
house price in these countries had
been affected mainly by domestic
factors.
• Income and interest rate series,
however, exhibit somewhat similar
trend as OECD countries, although
different in patterns.
23
Estimation Result(II):
Model results for selected East-asia countries
H.K.
restricted
model
Common Factor（
f t ）1.977***
(0.021)
Taiwan
Korea
full
model
restricted
model
full
model
restricted
model
full
model
0.733***
(0.004)
1.899***
(0.018)
0.732***
(0.004)
1.682***
(0.017)
-0.869***
(0.003)
GDP per capita（ Yt ）
-3.663***
(0.420)
1.382***
(1.463)
17.773***
(0.309)
Interest Rate（ Intt ）
-0.372***
(0.011)
-0.692***
(0.064)
-0.488***
(0.017)
Dewelling Tax（ Taxt ）
-1.688***
(0.546)
-0.995*
(0.586)
without country-specific factors
State space model
ft  0.721 ft 1  t
(0.007)
with country-specific factors
ft  0.962 ft 1  t
(0.049)
Note 1: Dependent variables are the house price index of selected countries.
Note 2: Constant terms are not reported.
Note 3:*, **, ***represents 10%, 5%, and 1% significance level, respectively.
24
Points to note:
1. Coefficients for the common factor are all
positive and significant,
implying that house price in each country had
been affected by their common trend.
2.The extent of influence of the trend here is
much lower than the OECD case.
3. Like the OECF Case, the influences of interest
rate and tax burden are obvious and
consistent with our expectation.
25
Points to note:
4. The influence of per capita GDP on
house price is undecided--positive in
Taiwan and Korea but negative in H.K.
5. Result looks better when countryspecific factors are included in the
estimation.
26
The actual and fitted house price
for selected East Asia countries
HK
Taiwan
350
175
300
150
250
125
200
100
150
real
real
s tate
s tate
s tate and others
100
s tate and others
75
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
Korea
200
175
150
125
100
75
real
s tate
s tate and others
50
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
27
Three points to make
1. Fitted values deviate more from
the actual values than the OECD
case.
2. Fitted value failed to resemble the
actual value in Korea.
3. Inclusion of country –specific
factors improves the predictions in
these Asian countries.
28
Concluding Remarks
1.
The common trend do affect house price
movements in both cases.
2.
The empirical result of this study attests more
or less that of Chirinko et al.(2004)and Otrok
and Terrones(2005) and succeeds only limited
in attaining its object due to the ambiguous
performance of the income variable.
3.
Wish the problem of the income variable can be
solved in future.
29
Thank You
30