the risk profile of housing corporations. European Real Estate Society Conference. Stockholm, June 24-27, 2009. Bert Kramer. Tessa Kuijl, Marc Francke.
ortec-finance.com
The impact of house price index specification levels on the risk profile of housing corporations European Real Estate Society Conference Stockholm, June 24-27, 2009 Bert Kramer Tessa Kuijl, Marc Francke Ortec Finance Research Center
Page 1
ortec-finance.com
Contents Introduction House price indices Impact on sales revenue Impact on solvency Conclusions
Page 2
ortec-finance.com
Introduction - I
Dutch housing corporations under political pressure: Subject to corporation tax (> €500 mln/year); Compulsory urban renewal levy (€75 mln/year); Increase energy efficiency of housing stock; Increase production; Negative publicity; Tightened supervision;
In times of : Falling house prices due to little demand on the owner-occupied market. Liquidity problems for some housing corporations.
Page 3
ortec-finance.com
Introduction - II
Page 4
Increasing field of tension between guaranteeing financial continuity and social objectives.
Asset Liability Management (ALM) becomes increasingly important to obtain insight into financial risk connected to specific social and financing strategies.
ortec-finance.com
ALM for housing corporations Balance sheet Real estate
Simulation Capital Loans
Policy: Rent, maintenance, sales, investments, demolition/newly built, treasury Uncertainty: construction costs inflation, interest rates, house prices
Risk analysis
Page 5
ortec-finance.com
House price indices - I
Used (a.o.) to calculate risk and return on sales programs
Up till now: Dutch national house price index used
Renes and Jokovi (2008): not one house market, but several regional markets in the Netherlands
Recently, lower level house price indices have become available
Francke et al (2009): statistical properties differ per house type and region
Research question:
Analyze impact of specification level of house price index on risk-return profile of housing corporations
Page 6
ortec-finance.com
House price indices - II
Page 7
Indices from OrtaX local linear trend repeat sales model
National index compared with indices based on COROP region and house type (row / corner house versus apartments)
Local indices available from 1993
ortec-finance.com
House price indices - III
Using indices from 1993 underestimates house price risk
To extend the time series to the early 1970s we can:
Figure 2.2: Change in sales prices within the Netherlands 40 35 30 25
1. Estimate the missing data (backcasting) Regress on other time series without missing data 2. Use derived time series:
C h a n g e (% )
20 15 10 5 0
Derive scenarios from scenarios of other variables
Page 8
We use a combination of these two approaches
-5
1973
1976
1979
1982
1985
1988
1991
-10 -15 Year
1994
1997
2000
2003
2006
ortec-finance.com
Case description
Impact analyzed on 5 housing corporations from different COROP regions
1. Leeuwarden (COROP region 4: Northern Friesland); 2. Almelo (COROP region 12: Twente); 3. Amsterdam (COROP region 23: greater Amsterdam); 4. Alphen aan de Rijn (COROP region 28: Eastern South Holland); 5. Goes (COROP region 32: other Zeeland).
Page 9
Focus on risk only, expected local house price increases assumed equal.
All distributions based on 500 economic scenarios.
Analysis 1 (“asset only”): impact on probability distribution of sales revenue in 2018 of €100K house (price level 31 December 2008).
Analysis 2 (ALM): impact on solvency levels using the actual multi-year plans.
ortec-finance.com
Local house price increases row / corner houses 0,250
0,200
Leeuwarden
0,150
Almelo Amsterdam
0,100
Alphen Goes
0,050
National
0,000 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 -0,050 Year Page 10
ortec-finance.com
Local house price increases apartments 0,250
0,200
Leeuwarden
0,150
Almelo Amsterdam
0,100
Alphen Goes
0,050
National
0,000 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 -0,050 Year Page 11
ortec-finance.com
Impact on sales revenues - I Figure D.2: Distribution of values in 2018 - Alphen
Market values (x1000)
Page 12
40 0 M or e
36 0 38 0
34 0
30 0 32 0
26 0 28 0
22 0 24 0
20 0
16 0 18 0
12 0 14 0
Weighted
80 10 0
Per type
Weighted
60
National
Per type
40
National
20
40 0 m or e
38 0
34 0 36 0
32 0
28 0 30 0
26 0
22 0 24 0
20 0
16 0 18 0
12 0 14 0
80 10 0
40
60
20
Figure D.3: Distribution of values in 2018 - Amsterdam
Market values (x1000)
Amsterdam and Alphen show largest differences
Amsterdam: standard deviation increases by more than 50%, VaR’s 25% lower
Alphen: Standard deviation +33%; VaR’s -16% to -24%
Differences in variances are significant at 1% level
No significant differences between separate indices per type and a weighted index
ortec-finance.com
Impact on sales revenues - II Figure D.5: Distribution of values in 2018 - Leeuwarden
Figure D.4: Distribution of values in 2018 - Goes
20
40
60
80
100 120 140 160 180 200 220 240 260 280 300 320 340 360 Market value (x1000)
Page 13
National
National
Per type
Per type
Weighted
Weighted
20
40
60
80
100 120 140 160 180 200 220 240 260 280 300 320 340 360 Market value (x1000)
For Goes and Leeuwarden, differences are not significant
For Leeuwarden risks seem slightly lower with detailed index
Almelo shows significant differences: standard deviation +12%, VaR’s -10%
Again, no significant differences between separate indices per type and a weighted index
ortec-finance.com
Impact on solvency - I Figure E.3: Distribution of solvency ratio Amsterdam in 2023
Figure E.2: Solvency ratio distribution Alphen in 2017
weighted weighted
national
7, 5% 15 ,0 % 22 ,5 % 30 ,0 % 37 ,5 % 45 ,0 % 52 ,5 % 60 ,0 % 67 ,5 % 75 ,0 % 82 ,5 % 90 ,0 %
0, 0%
-7 ,5 %
-1 5, 0%
national
-5%
0%
5%
10%
15%
Solvency ratio
Page 14
20%
25%
30%
35%
40%
45%
Solvency ratio
•
Amsterdam: 25% of current stock sold in coming 15 years
•
Using a national index leads to significant underestimation of actual risk profile;
•
1% VaR drops below zero when using a local index
•
Alphen: risk slightly lower with local index (in contrast to asset only)!
50%
ortec-finance.com
Impact on solvency - II Figure E.1: Solvency ratio distribution Almelo in 2013
12,5% 15,0% 17,5% 20,0% 22,5% 25,0% 27,5% 30,0% 32,5% 35,0% 37,5% 40,0%
Figure E.4: Distribution of solvency ratio Leeuwarden in 2024
weighted
weighted
national
national
-60% -50% -40% -30% -20% -10%
Solvency ratio
Page 15
0%
10% 20% 30% 40% 50% 60% 70% 80%
Solvency ratio
For Almelo, like Alphen, risk decreases with a local index (in contrast to asset only)
Impact small due to shorter horizon and relatively small sales program (5.4% in 5 yr)
Mixed results for Leeuwarden, differences insignificant.
ortec-finance.com
Conclusions
Impact of local versus national index differs strongly depending on region and house type.
Especially for the greater Amsterdam area the risk is underestimated when the national index is used.
ALM results sometimes opposite to asset only results.
Housing corporations should investigate the need to switch to local indices.
Topic for further research: extend local house price indices to the early 1970s
Page 16