the swedish housing market: trends and risks · •my report discusses both issues. here i focus on...
TRANSCRIPT
The Swedish housing market: Trends and risks
Presentation by
Peter Birch Sørensen
at the Swedish Fiscal Policy Council conference on
Studier i finanspolitik
in Stockholm on June 10, 2013
The issues
• Are Swedish house prices currently seriously overvalued?
• If so, what are the risks to financial and economic stability in Sweden?
• My report discusses both issues. Here I focus on the first one
Agenda
• Swedish house price developments in historical and international perspective
• Can real house prices continue to trend upwards? • Can recent Swedish house price developments be
explained by econometric house price models? • Are current Swedish house prices out of line with rents and
disposable incomes? • Are actual house prices out of line with fundamental house
prices?
The evolution of Swedish house prices
Real house prices in Sweden
0
50
100
150
200
250
52 56 60 64 68 72 76 80 84 88 92 96 00 04 08 12
Quarterly real house prices in various countries
0
20
40
60
80
100
120
140
1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
UK
USA
Ireland
Sweden
Can real house prices continue to trend upwards?
Potential reasons for a secular increase in real house prices
• Rising real land prices
• Rising real construction costs
Agricultural land prices and consumer prices in Sweden
0
50
100
150
200
250
300
350
400
450
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Price of agricultural land Consumer prices
Construction costs and consumer prices in Sweden
0
50
100
150
200
250
300
350
400
450
1980M01 1985M01 1990M01 1995M01 2000M01 2005M01 2010M01
Construction costs Consumer prices
Theoretical priors
In the long run, we have
• A stationary user cost of housing
• A constant budget share of housing costs?
What are the assumptions needed for these priors to hold?
Some notation
income elasticity of housing demand
price (user cost) elasticity of housing demand
price elasticity of housing supply
budget share of housing
real house price
Y
R
S
B
P
Implications of simple housing market model (when user costs are stationary)
Scenario 1 1 (Brusewitz (1998)) and 0 and 0
Scenario 2 1 (Englund et al. (1995)) and 0, sign of uncertain
Scenario 3
Y R S
Y R S
dB dP
dP dB
1 and 0 and 0
Scenario 4 1 and 0
Y R S
Y R S
dB dP
dB dP
Further implications
Assumptions (Scenario 1):
• Constant budget share of housing
• Stationary user cost of housing
• Secular rise in real house prices
Implication:
• The real housing stock must grow at a lower rate than real income
The ratio of the real housing stock to real disposable income in Sweden
0
20
40
60
80
100
120
1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
A caveat: measurement problems
• Measured construction costs and observed house prices may partly reflect increases in the quality of housing
Implications:
• The statistics may underestimate the increase in the (quality-adjusted) housing stock
• The recorded increase in real house prices may overstate the true price increase
An econometric model of Swedish house prices
A recent model of Swedish house prices
• Claussen (2012) estimates an error-correction model of real quarterly house prices
Fraction of increase in real house prices from 1996 to 2011 that can be explained by
• Increase in real disposable income: 62%
• Fall in mortgage interest rates: 26%
• Rise in household financial wealth: 8%
• Unexplained residual: 4%
Claussen (2012): Actual and predicted real house prices in Sweden
Some worries
The Claussen model may not be well suited for forecasting, because
• The real interest rate is not treated as a stationary variable
• The equilibrium real house price is estimated to grow by 1.4% for every 1% increase in real income. Implication: The real housing stock must fall in the long run unless the budget share of housing costs is forever increasing
Are current Swedish house prices out of line with rents and incomes?
Issues
• Has owner-occupation become too expensive relative to renting?
• Has owner-occupation become too expensive relative to incomes?
The cost of owner-occupied housing
• Imputed rent = (user cost)x(real house price) • User cost = real after-tax interest rate + property tax + maintenance cost – expected real capital gain + premium for risk and credit constraints In the following, the user cost is approximated by the
after-tax real 5-year mortgage interest rate plus a constant
Ratio of imputed to actual rents and ratio of house prices to rents in Sweden
0
20
40
60
80
100
120
140
160
180
200
1980 1985 1990 1995 2000 2005 2010
House price/rent Imputed rent/rent
Average over entire period = 100.
Ratio of imputed rent to disposable income and ratio of house prices to disposable income in Sweden
Average over entire period = 100.
0
20
40
60
80
100
120
140
160
180
1980 1985 1990 1995 2000 2005 2010
House price/income Imputed rent/income
Fundamental versus actual house
prices in Sweden
(Background study: Bergman and Sørensen (2013))
Issues
• How do we define the ”fundamental” house price?
• How have fundamental house prices in Sweden evolved?
• Do actual house prices converge on fundamental house prices?
The fundamental house price
• Starting point: Formula for imputed rent
Implication when consumers are forward-looking:
• Fundamental house price = discounted value of rationally expected future imputed rents
(Note: discount rate = user cost excluding expected capital gain)
”Rent” model of fundamental house price
Assumption:
• Imputed rents on owner-occupied housing are proportional to the rents paid for rental housing
Implication:
• Expected future imputed rents depend on expected future rents for rental housing
”Supply and demand (S-D)” model of fundamental house price
Assumptions:
• Imputed rents on owner-occupied housing adjust so as to equilibriate housing demand and housing supply
• Housing demand varies positively with real disposable income and negatively with imputed rents
Implication:
• Expected future imputed rents depend on expected future incomes and the expected future housing stocks
Empirical implementation • Expectations of future imputed rents and future user costs
are based on a VAR model of the historical interaction between incomes, rents, user costs, housing investment and actual house prices
• Income and price elasticities of housing demand are chosen
exogenously on the basis of prior knowledge • Estimation procedure imposes the restriction that the
average levels of actual and fundamental house prices are identical over the estimation period
• Estimates are based on quarterly data for Sweden from
1986:1 to 2012:1
Actual and estimated fundamental real house prices (rent model)
pta
pt
1990 1995 2000 2005 2010
4.25
4.50
4.75
5.00
pta
pt
gapt = p
ta - p
t
1990 1995 2000 2005 2010
-0.25
0.00
0.25
gapt = p
ta - p
t
Estimated price gap 2012:1 = 18%
Actual and estimated fundamental real house prices (S-D model, εY =1, εR =0.5)
pta
pt
1990 1995 2000 2005 2010
4.25
4.50
4.75
5.00
pta
pt
gapt = p
ta - p
t
1990 1995 2000 2005 2010
-0.25
0.00
0.25
gapt = p
ta - p
t
Estimated price gap 2012:1 = 12%
Sensitivity of house price gap to price elasticity of housing demand (εY =1)
gapt = p
ta - p
t (e
R = 0.5 e
Y=1)
gapt = p
ta - p
t (e
R = 1 e
Y=1)
1990 1995 2000 2005 2010
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
gapt = p
ta - p
t (e
R = 0.5 e
Y=1)
gapt = p
ta - p
t (e
R = 1 e
Y=1)
Sensitivity of house price gap to income elasticity of housing demand (εR =0.5)
gapt = p
ta - p
t (e
R = 0.5 e
Y=1)
gapt = p
ta - p
t (e
R = 0.5 e
Y=0.5)
1990 1995 2000 2005 2010
-0.4
-0.2
0.0
0.2
0.4
gapt = p
ta - p
t (e
R = 0.5 e
Y=1)
gapt = p
ta - p
t (e
R = 0.5 e
Y=0.5)
House price dynamics Hypotheses: • Actual house prices do not affect fundamental house prices • Fundamental house prices help to predict actual house
prices • A positive gap between actual and fundamental house
prices causes a future drop in actual prices, and vice versa • Findings: All hypotheses accepted by a Vector Error
Correction model of the interaction between the actual and the estimated fundamental house prices
Estimated half-life of the house price gap (years)
Note: The half-life is the number of years it takes before half of the house price gap is closed, following a shock to the housing market
Some other factors affecting
fundamental house prices in Sweden
Factors that may have boosted fundamental house prices
• Financial innovations
• The recent decrease in property taxes
• Migration towards the large cities
The supply side of the housing market: Swedish housing investment is still relatively low
0
0,01
0,02
0,03
0,04
0,05
0,06
0,07
0,08
0,09
2000 2002 2004 2006 2008 2010 2012
Rat
io o
f h
ou
sin
g in
vest
me
nt
to G
DP
Sweden
UK
USA
Denmark
Norway
But Swedish construction activity has in fact responded to rising house prices
Sweden (0.91)
Denmark (1.24)
UK (1.22)
USA (0.46) Spain (0.31)
France (0.21)
Norway (0.63)
Belgium (0.75)
0
0,05
0,1
0,15
0,2
0,25
0,3
0 0,05 0,1 0,15 0,2 0,25 0,3
Ch
ange
in t
he
(lo
g o
f th
e)
rati
o o
f h
ou
sin
g in
vest
me
nt
to G
DP
Change in (log of) real house price
Conclusions on the supply side
• By international standards, the supply side of the Swedish housing market does not seem to be inflexible
• The relatively low ratio of housing investment to GDP may reduce the negative impact on the Swedish economy if house prices fall
Summary of main points
• A secular increase in real house prices is possible if the income and price elasticities of housing demand are both equal to one
• The most recent econometric model of Swedish house prices implies an implausibly high income effect on house prices from a long run perspective
• Current Swedish house prices seem somewhat overvalued when judged against the level of rents, but now when judged against disposable incomes
Summary of main points (ct’nd)
• Swedish real house prices appeared to be 12-18 percent above their fundamental level in early 2012
• However, this estimate does not account for the likely boost to fundamental house prices from financial innovations, property tax reductions and internal migration
• The supply side of the Swedish housing market seems to be reasonably flexible by international standards