Housing boom and burst as seen from the
Spanish Survey of Household Finances
Olympia Bover
Prepared for the 2nd Conference on Real Estate Market,
Bank of Greece,
Athina , December 2011
DG ECONOMICS, STATISTICS AND RESEARCH
PLAN FOR THIS TALK
1. Recent developments in the Spanish housing markets
Evidence from the Spanish Survey of Household Finance
2. Housing in Spanish household wealth
3. Evidence on self-assessed housing valuation by households
4. Housing wealth effects on consumption
5. House purchases and the dynamics of housing wealth
DG ECONOMICS, STATISTICS AND RESEARCH
2
1. RECENT DEVELOPMENTS
(I)
New home transactions
(level)
% GDP
10
30.000
360.000
25.000
300.000
20.000
240.000
8
15.000
180.000
7
10.000
120.000
6
5.000
60.000
5
0
4
0
2008M1
2009M1
2010M1
HOUSING INVESTMENT
9
2011M1
3
monthly level
twelve-month moving sum (right)
1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011
DG ECONOMICS, STATISTICS AND RESEARCH
3
1. RECENT DEVELOPMENTS
Housing approvals and finished dwellings
(twelve-month moving sum)
(II)
UNSOLD HOUSING
QUARTER-ON-QUARTER CHANGES (right scale)
ESTIMATED STOCK
1.000.000
Thousands
Thousands
800.000
600.000
400.000
200.000
0
2007M1 2008M1 2009M1 2010M1 2011M1
2004 2005 2006 2007 2008 2009 2010 2011
Housing approvals
Finished dwellings
DG ECONOMICS, STATISTICS AND RESEARCH
4
1. RECENT DEVELOPMENTS
(III)
House prices
(nominal growth rates)
6
4
HOUSE PRICES (real terms)
Ireland
USA (Case-Shiller)
France
Quarter on quarter growth rate
Year on year growth rate
2
0
110
-2
-4
Spain
UK (Nationwide)
Index= 100 cycle 100
-6
-8
90
-10
1
2
3
4
1
2008
2
3
4
1
2009
2
3
4
1
2010
2
3
4
2011
80
House prices
(cumulative adjustment from 2008 q1)
100
100
Real adjustment
Nominal adjustment
95
95
90
90
85
85
80
80
75
75
1
2
3
2008
4
1
2
3
2009
4
1
2
3
2010
4
1
2
3
2011
4
70
60
50
40
98 99 00 01 02 03 04 05 06 07 08 09 10 11
Nota: Datos de 2011 hasta Q2, excepto España y UK, hasta Q3. DG ECONOMICS, STATISTICS AND RESEARCH
5
1. RECENT DEVELOPMENTS
(IV)
PROVINCIAL BREAKDOWN OF ADJUSTMENT OF HOUSE PRICES
THE ADJUSTMENT OF HOUSING PRICES (In real terms)
COMPARISON BETWEEN THE CURRENT AND PREVIOUS CYCLES
1979
1991
-10,4
-12,2
-11,2
-8,1
-18,8
-9,7
120
-17,1
-9,6
-13,3
-14,9
2007
-13,5
-8,2
-20,7
-15,5
na
-13,9
-12,8
-16,3
-20,1
115
-16,5
-24,1
-4,1
-20,0
-15,3
-22,1
110
-19,4
-28,1
na
-23,7
-19,5
0,0
Real prices: peak year = 100
105
-5,3
-24,8
-19,0
-18,7
100
-11,3
-10,8
-15,2
-22,2
95
-14,6
-9,9
-22,8
-21,5
-12,3
90
-14,2
2010
85
80
-15,9
-4,0
2011 Q3
-21,4
-25,7
-4,2
CHANGE IN NOMINAL TERMS FROM PEAK
TO 2011 Q3 (%)
NATIONAL AVERAGE = 17,7%
MAXIMUM FALL-PERCENTILE 25
PERCENTILE 25-PERCENTILE 75
75
PERCENTILE 75-MINIMUM FALL
-13,7
70
-22,3
0
1
2
3
4
5
6
7
8
9
Years subsequent to peak in real prices
10
SOURCES: Ministerio de Fomento and Banco de España.
DG ECONOMICS, STATISTICS AND RESEARCH
6
2. DISTRIBUTION OF HOUSEHOLD ASSETS IN
SPAIN FROM EFF 2008
Percentage of households
owning asset (% )
Distribution of the value of households assets (% )
Main
Other real estate
Business and
other real assets
Bank accounts
Other financial
assets
Main residence
Other real estate
54.8
24.8
9.2
5.3
5.9
82.7
36.1
Less than 20
77.7
13.3
1.0
4.3
3.7
78.1
20.1
Between 20 and 40
67.5
20.7
4.7
4.3
2.8
78.7
29.7
Between 40 and 60
64.5
20.9
7.1
4.4
3.1
84.1
33.0
Between 60 and 80
56.3
26.3
6.9
5.4
5.1
83.1
42.2
Between 80 and 90
55.6
23.4
8.3
5.8
6.9
87.7
49.1
Between 90 and 100
35.8
31.6
16.5
6.2
9.9
91.3
61.8
residence
All households
Income percentile
DG ECONOMICS, STATISTICS AND RESEARCH
7
3. EVIDENCE ON SELF-ASSESSED HOUSING
VALUATION BY HOUSEHOLDS
ƒ EFF information contains information on self-assessed values of the
main residence (and its change across successive waves)
Change in the self-assessed value of the main residence :
• 2009Q1 vs. 2005Q4 -7% in real terms (in contrast to no variation according
to appraisal based aggregate HP statistics)
• 2005Q4 vs. 2002Q4 +68% in real terms (in contrast to +45% according to
appraisal based aggregate HP statistics)
> > looks as if self-assessments tend to amplify/overshoot trends
ƒ
ƒ
Potential reasons
• appraisals (long term concept) vs. self-assessed value (short term)
• not only houses that are sold but stock
• different house types sold in periods of increasing prices
ƒ Importance of change in self-assessed value
> > This is the housing value that matters for wealth effects on consumption!
DG ECONOMICS, STATISTICS AND RESEARCH
8
4. HOUSING WEALTH EFFECTS ON CONSUMPTION
(I)
ƒ
How much consumption will increase as a result of an exogenous
increase in wealth
ƒ
Problems with estimations of wealth effects based on aggregate data
. instability over long periods due to changes in credit market
. potential reverse causality
ƒ
Estimating wealth effects at the micro level is important because of
heterogeneity; they are expected to vary with age, composition of
household wealth…
Aim to estimate a (non structural form) causal effect that is a useful input
for policy
→ Exogenous variation in wealth data comes from house price (HP) variation
by locality and inheritance indicators; also include a rich set of controls
ƒ
DG ECONOMICS, STATISTICS AND RESEARCH
9
4. HOUSING WEALTH EFFECTS ON CONSUMPTION
(II)
ƒ
ƒ
ƒ
ƒ
ƒ
Under certainty, ∆ (owner occupied) house prices may have no effects on
consumption if moving costs are large and the borrowing possibilities for
the liquidity constrained are limited (even in a model without bequests)
The user cost of owner occupied housing rises with HP but increase in
spending may occur if
(i) downsizing or (ii) reverse mortgages
But actual downsizing or reverse mortgages in Spain are uncommon
However, possibility of downsizing in the future if needed may be
sufficient for middle-aged homeowners to reduce the need for other
precautionary savings
Precautionary savings model has specific predictions about mpc
mpc out of housing wealth varies by age
mpc declining with wealth
If precautionary motives strong, not only drop in consumption but also
not enough precautionary savings in the future
DG ECONOMICS, STATISTICS AND RESEARCH
10
4. HOUSING WEALTH EFFECTS ON CONSUMPTION
(III)
ƒ
Evidence mpc out of housing wealth ≈ 0.03
Real estate wealth of Spanish households end 2008: 5,500 to 6,000 billion €
Decrease 10% HP ie decrease 550 billion € real estate wealth x 0.03
→ means a decrease in 16.5 billion € in consumption
ƒ
Partial equilibrium effects: insofar as changes in house prices were
associated with changes in employment or interest rates, the general
equilibrium effect could be more pronounced
ƒ
These figures mask important differences across groups of households
. prime age households around 0.06
. vary from 0.035 at the bottom of the housing wealth distribution to 0.004 at
the very top
DG ECONOMICS, STATISTICS AND RESEARCH
11
4. HOUSING WEALTH EFFECTS ON CONSUMPTION
(IV)
ƒ
Given high % of home-owners and owners of other real estate, HP
fluctuations may significantly affect aggregate expenditure even with
relatively low average mpc
ƒ
The estimated non-linearity and age pattern of wealth effects confirm our
a priori that our estimated wealth effects reflect precautionary saving
motives
ƒ
When these effects reflect changes in precautionary savings:
. severe busts may seriously affect retirement plans and expectations of
middle age households in so far as they are counting on housing equity gains
in the face of personal negative shocks when old.
. it is those holding less housing wealth that may see their retirement plans
and expectations more affected (due to nonlinearity of effects)
DG ECONOMICS, STATISTICS AND RESEARCH
12
5. HOUSE PURCHASES AND THE DYNAMICS OF
HOUSING WEALTH
(I)
ƒ
Develop empirical models of purchase behaviour of main and secondary
housing by Spanish households to identify determinants of house
purchases and housing wealth
ƒ
Particular attention to the relationship between expected house prices
and purchases
ƒ
Exploit (current and retrospective) information from the first two waves of
the Spanish Survey of Household Finances (panel component only)
Attention to transactions involving secondary housing since these
account for more than one third of housing transactions
→ heterogeneity across countries: Greece, Portugal, Spain vs. Netherlands
(30-35% of housing stock is secondary housing vs. 1%)
ƒ
DG ECONOMICS, STATISTICS AND RESEARCH
13
5. HOUSE PURCHASES AND THE DYNAMICS OF
HOUSING WEALTH
(II)
MAIN DETERMINANTS OF PURCHASE PROBABILITY
ƒ Strong effect of expected housing return measured by Et(∆lnHPt,t+5)- it
For each household in the sample we calculate the probability of purchase in
2005 using the estimated effects. We then replace its expected return (0.088
on average) by the one in 2009 (0.015 on average). This predicts a 61% fall
in the number of purchases between 2005 and 2009. Interestingly, according
to the National Stats Office the aggregate number of housing transactions fell
by 58%
ƒ Income, University education
ƒ Being married (but no effect of children)
→ Therefore, changes in expectations of rates of return are able to predict
the fall in aggregate purchases between e.g. 2005 and 2009
USE S-s RULE MODELS TO DISENTANGLE THE EFFECTS BETWEEN
¾ effects on the desired share of housing in household wealth
¾ effect on the width of the inaction range
DG ECONOMICS, STATISTICS AND RESEARCH
14
5. HOUSE PURCHASES AND THE DYNAMICS OF
HOUSING WEALTH
(III)
ƒ
Effects on the target i.e. on proportion of wealth in housing households
aim to hold
• Positive and well determined effect of return on housing
→ the size of the effect implies that the increase in the rate of return that
occurred between 2002 and 2001 (around 2%) increased the target housing
wealth ratio in 2.2 percentage points
• Income and wealth have a negative effect: diversification for high income
and wealth households (housing as a necessity in terms of life-cycle wealth)
• Negative effect of being self-employed (probably due to wealth invested in
own business)
• No life-cycle effects
ƒ
Effects on the inaction range i.e. on time to reach target
• Negative effect of income and wealth pointing to the importance of fixed
transaction costs
• Housing returns do not affect the inaction range
DG ECONOMICS, STATISTICS AND RESEARCH
15
5. HOUSE PURCHASES AND THE DYNAMICS OF
HOUSING WEALTH
(IV)
¾ Expected housing returns affect desired housing wealth target but not the
inaction range. This implies that when expected returns drop there is a
reduction in the desired amount of housing wealth not just a delay in
purchases
¾ Estimated association between housing demand and recent house price
growth is revealing about how price frenzies may develop
DG ECONOMICS, STATISTICS AND RESEARCH
16
SOME LINKS
Bulletin Article with main results from EFF2008
http://www.bde.es/webbde/SES/Secciones/Publicaciones/InformesBoletines
Revistas/BoletinEconomico/11/Jul/Files/art3e.pdf
Wealth effects on consumption
http://www.olympiabover.name/Housing%20wealth_June07.pdf
Housing purchases and the dynamics of housing wealth
http://www.olympiabover.name/Housing-demand_October2010.pdf
DG ECONOMICS, STATISTICS AND RESEARCH
17
THANKS FOR YOUR ATTENTION
DG ECONOMICS, STATISTICS AND RESEARCH