Financial Systems, Structural National-wide Differences in Housing Finance

(1)

Financial Systems, Structural National-wide Differences in Housing Finance

University of Economics and Business Administration, Vienna

Economic Analysis Division, Oesterreichische Nationalbank Vienna, 29 September 2008

Workshop Housing Market Challenges in Europe and the United States – any solutions available?“

(2)

Outline

Stylized Facts

Determinants of House Prices: Attempts of explanation Housing finance and the role of the state

Empirical Investigation: The Model used Empirical results

Conclusions

(3)

1. Stylized Facts

Increasing in house prices in the US over the last years => burst of bubble led to severe problems for

households

Foreclosures

the housing finance system

Semi-public agencies; Developments of the last years: opening financing possibilities for non-prime debtors (mortgage backed securities)

US: developments vary significantly across states European Economies:

House price development vary

Residential mortgage debt ratio increases

(4)

1. Stylized Facts

(5)

1. Stylized Facts

(6)

2. Determinants of House Prices: Attempts of explanation

Neoclassical models:

Demand: user costs of housing

positive: household income, financial wealth and the expected rate of return on housing (Mishkin, 2007)

negative: change in house prices, the real rate on housing loans (Girouard, 2006; Egert / Mihaljek, 2007);

Supply: highly inelastic

Additional interrelations:

Households formation (Pozdena, 1988) Demographic changes

innovations and institutional factors mostly neglected (Miles, 1994;

Pozdena, 1988) or integrated as vector (Egert / Mihaljek 2007)

(7)

2. Determinants of House Prices: Attempts of explanation

Transmission mechanism of the housing sector:

• Restructuring of the mode of mortgages

Competitive finance sector; interrelation of house prices, interest rates and consumer spending

Characteristics of housing finance: (Goodhart / Hofmann, 2008;

Tsatsaronis / Zhu, 2004; Girouard, 2006; Warnock / Warnock, 2007;

Zhu, 2006; Committee on the Global Financial System, 2006; Egert / Mihaljek 2007)

• Exogenous changes in credit supply

• Innovations that enable and ease access to the financial market increase the demand for housing

Additionally structural and institutional features are important (Hoeller /

Rae, 2007; Catte, 2004; ECB, 2003) – role of the state => no empirical

studies.

(8)

3. Housing Finance and the Role of the State

4 different indirect financing modes (Green / Wachter 2007; Lea, 2001;

Mooslechner, 1990)

simple deposit system; Contract savings systems, mortgage banks and secondary market systems

Financial structures are a mixture :

European Mortgage Credit Integration Sub Prime lending via MBS

Housing finance revolution (Shiller, 2007): in the US and UK.

Increase in the demand for loans – decrease in interest rates (costs of financing) – consumer spending – impact for house prices

Model: representative interest rates on new mortgages, number of permissions

(9)

3. Housing Finance and the Role of the State

Country Introduction of securitization - mortgage backed

securities (MBS)

Usage of mortgage backed securities

Austria -- no

Belgium yes limited

Denmark yes limited

Germany unclear ?*

Greece yes limited

Spain 1992 extensive

France 1999 limited

Ireland second half 1990s yes

Italy yes extensive

Luxembourg yes yes

Netherlands yes extensive

Portugal yes limited

Finland 1989 ?*

Sweden yes limited

UK 1987 extensive

USA yes extensive

Source: Springler, 2008, Table 15; Tsatsaronis / Zhu, 2004, Table 1.

Introduction and Usage of Securitization for housing finance in

European Economies and the USA.

(10)

3. Housing Finance and the Role of the State

3 basic forms : (Amann, 2000; Czerny, 2001)

Supply side allowances Demand side allowances

=> Data not comparable due to differences in definitions. Additionally national

differences in the existence and perception of the rental sector have to be taken into account.

Tax allowances

=> Structural features of housing allowance schemes can be captured indirectly (see the causal relation between total building permissions and subsidized housing units)

National tax system

Industrial view – taxation of capital income: (Goulder, 1989 and Anas / Arnott, 1997) Residential view – taxation of income:

Depending on historical background, also indirect interrelation (Green / Vandell, 1999; Poterba 1984, 1991)

Cost of homeownership: (Wolswijk, 2005; Van den Noord, 2005): interest

deductibility from mortgage rate, taxes on imputed rent, taxation of capital

gains, property taxation; empirical studies of European Economies

(11)

3. Housing Finance and the Role of the State

Selected Economies: Supply side allowances to GDP ratio in % in 2001

Source: Stagel (2004).

(12)

3. Housing Finance and the Role of the State

Source: Amann (2005).

Development of Subsidized housing supply and Housing Permissions

and Completions in Austria 1996-2003

(13)

4. The model behind…

Several ways how tax measures can affect housing cost for private households:

• by deductibility of interest payments on mortgages from taxable income (with a deduction limit or not),

• by the possibility to get a tax credit and the taxation of imputed income from owner-occupied housing.

We expanded model by Fukao and Hanazaki (1986), Paul van de Noord (2005).

typical price of one unit of housing P

Set P equal to 6* APW (APW….disposable income of an average production

worker).

(14)

4. The model behind…

MTR…marginal tax rates i…. Pre-tax nominal interest rate (long-term government bond yield)

…after tax nominal interest rates (limited duration of tax relief ->

time-dependent

Relationship

….inflation rate …..nominal cost of financing

i MTR

i

i _a = − *

∞ ∞

−

− =

0 0

) (

)

( dt i ( t ) Pe dt

Pe

i _f ⁱ ^π ^t _a ⁱ ^π ^t

π i _f

)

(t

i _a

(15)

4. The model behind…

Transformation - 3 relevant cases:

• constant regime over time:

• one change over time:

i

_a1

and i

_a2

….the financing cost before and after t1.

• two changes within the regime :

i

_a3

is the financing cost after t2.

a

f i

i =

)

1

( 1

2 1 ( _a _a ) ⁱ ^t

a

f i i i e

i = + − ⁻ ⁻ ^π

2

1

( )

2 3

) ( 1

2 1 ( _a _a ) ⁱ ^t ( _a _a ) ⁱ ^t

a

f i i i e i i e

i = + − ⁻ ⁻ ^π + − ⁻ ⁻ ^π

(16)

4. Data used for calculation of the cost of financing i _f

information on countries’ tax regulation and tax deductibility:

European Tax Handbooks (International Bureau of Fiscal Documentation - IBFD, 1996 to 2006).

MTR: marginal rates of income tax plus employees’ social security contributions and personal income tax – OECD.

P unit of housing: data on the disposable income – OECD.

-> take-home pay of a married couple with 2 children and one earner

(i.e. gross wage earnings - total payments to general government +

cash transfer from general government for two children)

(17)

4. Tax models used for the calculation of cost of financing…

• deduction with a ceiling but no time limit

e.g. Austria 1999: Interest payments … are deductible as special expenses up to ATS 20,000.

e.g. Ireland, 1997: deductibilty of interest paid on a loan applied for the purchase, improvement or repair of only or main residence, up to

…the lower of 80% (100% for a first-time buyer) of the interest paid or IEP 5,000 for a married couple, the first IEP 200 (married ) is not deductible, except for first-time buyers.

) , /

000 ,

20 min(

* P i

MTR i

i _a = −

) , / 200 min(

*

* 8 . 0 )

, / 5000 min(

*

* 8 .

0 MTR P i MTR P i

i

i _a = − +

(18)

4. Tax models used for the calculation of cost of financing…

• deduction with or without a ceiling, taxation of imputed rent e.g. Belgium, till 2004:

interest on mortgages may be deducted from taxable income up to the total amount of income from immovable property.

Imputed rental income from the taxpayer’s main dwelling-house included in the taxable income. The basic rate of the levy is 1.25% for the Brussels and the Walloon regions.

e.g. Belgium, since 2005: Additionally, interest on a mortgage contracted on or after 1 January 2005 may be deducted up to EUR 2,000 for the first 10 years and EUR 1,500 thereafter.

>

−

≤

−

= −

10 ),

0125 .

0 ,

/ 500 ,

1 min(

*

10 ),

0125 .

0 ,

/ 000 ,

2 min(

*

t i

P MTR

i

t i

P MTR

i _a i

) 0 , 0125 .

0 min(

* −

−

= i MTR i

i _a

(19)

4. Tax models used for the calculation of cost of financing…

• deduction of a fixed fraction of the acquisition value

e.g. Germany, 1996: a taxpayer acquiring or constructing a new owner-occupied dwelling receives a cash grant up to 5% of the construction or acquisition costs in the year of completion or

acquisition and in the following 7 years, with a ceiling of DEM 5,000 per annum. Acquisition or construction cost is the cost of the dwelling and the cost of the land.

Germany, since 2000 no tax relief: The tax relief for owner-occupied dwellings abolished.

Instead of tax relief, the taxpayer may currently be entitled to a tax-free cash grant for acquiring or constructing a new dwelling to be used by himself.

>

≤

= −

8 ,

8 ),

/ 5000 ,

05 . 0 min(

* t

i

t P MTR

i

_a

i

i _a =

(20)

4. Tax models used for the calculation of cost of financing…

• tax credit with indefinite duration

e.g. Italy, 2003: A tax credit equal to 19% of certain personal expenses is granted.

The expenses which qualify for credit are among others:

interest paid on mortgage loans contracted to finance the purchase of an owner-occupied dwelling-house, up to a maximum credit of EUR 686.89.

) 19 . 0 , / 89 . 686

min( P i

i

i _a = −

(21)

4. Determinants of ownership rate – the model

Panel data

Depending on model 15-17 countries (AT, BE, DK, FI, FR, DE, GR, IT, IR, PT, ES, SWE, LUX, NL, NORW, UK, US), 10 years (1997 to 2006)

owner …ownership rate GDPcap… GDP per capita

Populationgrowth… populationgrowth (year-on-year) permit… number of permissions

mortgdebtratio… Residential mortgage debt to GDP ratio costfin… nominal cost of financing i

_f

interestnew… representative interest rate on new mortgage loans

it i it i

it i

t i i

it i

t

v tfin

ap mortgdebtc

permit growth

population GDPcap

GDPcap owner

+ +

⋅ +

⋅

+

⋅ +

⋅

=

₋ ₋ ₋

µ β

β

β β

) ln(cos

) ln(

6 5

1 , 4

1 , 3

1 , 2

1

(22)

5. Empirical results: fixed effects-model

. xtreg owner gdpcap l.gdpcap l.populationgrowth l.permit mortgdebtratio costfin ,fe vce(robust)

Fixed-effects (within) regression Number of obs = 102 Group variable: varland Number of groups = 16 R-sq: within = 0.4288 Obs per group: min = 2 between = 0.0172 avg = 6.4 overall = 0.0176 max = 9 F(6,15) = 29.32 corr(u_i, Xb) = -0.4042 Prob > F = 0.0000 (Std. Err. adjusted for 16 clusters in varland) --- | Robust

owner | Coef. Std. Err. t P>|t| [95% Conf. Interval]

---+--- gdpcap |

--. | .1001613 .0521042 1.92 0.074 -.010896 .2112187 L1. | -.0521102 .0560829 -0.93 0.368 -.171648 .0674276 population~h |

L1. | .0103811 .0039163 2.65 0.018 .0020337 .0187285 permit |

L1. | -.0304139 .0121474 -2.50 0.024 -.0563055 -.0045223 mortgdebtr~o | .0279287 .0192121 1.45 0.167 -.0130209 .0688783 costfin | -.0245943 .0074942 -3.28 0.005 -.0405678 -.0086209 _cons | 3.811632 .2912128 13.09 0.000 3.190926 4.432337 ---+--- sigma_u | .20886496

sigma_e | .0153427

rho | .99463296 (fraction of variance due to u_i)

--- !"

(23)

5. Empirical results: fixed effects-model

. xtreg owner gdpcap l.gdpcap l.populationgrowth l.permit mortgdebtratio costfin interestnew ,fe vce(robust)

Fixed-effects (within) regression Number of obs = 96 Group variable: varland Number of groups = 15 R-sq: within = 0.4630 Obs per group: min = 2 between = 0.0120 avg = 6.4 overall = 0.0124 max = 9 F(7,14) = 86.38 corr(u_i, Xb) = -0.4356 Prob > F = 0.0000 (Std. Err. adjusted for 15 clusters in varland) --- | Robust

---+--- gdpcap |

--. | .1517806 .0643399 2.36 0.033 .0137852 .2897759 L1. | -.0960605 .0672633 -1.43 0.175 -.240326 .0482051 population~h |

L1. | .0133821 .0039155 3.42 0.004 .0049842 .02178 permit |

L1. | -.0368985 .0117886 -3.13 0.007 -.0621825 -.0116145 mortgdebtr~o | .0357541 .0237002 1.51 0.154 -.0150778 .0865861 costfin | -.041917 .0132649 -3.16 0.007 -.0703674 -.0134667 interestnew | .0273823 .0202113 1.35 0.197 -.0159667 .0707312 _cons | 3.670055 .2766834 13.26 0.000 3.076628 4.263482 ---+--- sigma_u | .21984747

sigma_e | .01532199

#

!"

(24)

If we check for endogeneity of GDP per capita and of interest rate…

Fixed-effects (within) IV regression Number of obs = 102

Group variable: varland Number of groups = 15 R-sq: within = 0.3749 Obs per group: min = 1 between = 0.1765 avg = 6.8 overall = 0.1507 max = 9 Wald chi2(6) = 6.80e+06 corr(u_i, Xb) = -0.5152 Prob > chi2 = 0.0000 --- owner | Coef. Std. Err. z P>|z| [95% Conf. Interval]

---+--- gdpcap |

--. | .0909136 .0769655 1.18 0.238 -.059936 .2417632 L1. | -.043141 .0724821 -0.60 0.552 -.1852032 .0989213 interestnew | .0180565 .0136501 1.32 0.186 -.0086972 .0448101 population~h |

L1. | .0085948 .0047537 1.81 0.071 -.0007223 .0179118 mortgdebtr~o |

L1. | .0262298 .0163827 1.60 0.109 -.0058797 .0583392 costfin | -.0290365 .0137941 -2.10 0.035 -.0560724 -.0020006 _cons | 3.434435 .2639153 13.01 0.000 2.91717 3.951699 ---+--- sigma_u | .21459137

sigma_e | .01597688

--- F test that all u_i=0: F(14,81) = 481.92 Prob > F = 0.0000 --- Instrumented: gdpcap L.gdpcap interestnew

Instruments: L.populationgrowth L.mortgdebtratio costfin interestnew gdpcap L.gdpcap

---

(25)

5. Dynamic GMM model

$ # % !

. xtabond2 owner l.shipneu l2.shipneu gdpcap l.gdpcap l.populationgrowth l.permit

l.mortgdebtratio costfin, gmm (l.owner, lag(3 4)) iv(l.interestnew) nolevel robust small Favoring space over speed. To switch, type or click on mata: mata set matafavor speed, perm.

Warning: Two-step estimated covariance matrix of moments is singular.

Using a generalized inverse to calculate robust weighting matrix for Hansen test.

Difference-in-Sargan statistics may be negative.

Dynamic panel-data estimation, one-step difference GMM

--- Group variable: varland Number of obs = 60 Time variable : vartime Number of groups = 14 Number of instruments = 12 Obs per group: min = 0 F(8, 14) = 5.33 avg = 4.29 Prob > F = 0.003 max = 7 --- | Robust

---+--- owner |

L1. | 3.863666 1.385978 2.79 0.015 .891038 6.836294 L2. | -.0958903 .6783046 -0.14 0.890 -1.550709 1.358928 gdpcap |

--. | -.6043866 .450992 -1.34 0.202 -1.571668 .3628951 L1. | .3800539 .6712064 0.57 0.580 -1.059541 1.819648 population~h |

L1. | -.0357837 .0456982 -0.78 0.447 -.1337967 .0622292 permit |

L1. | .1983531 .1858121 1.07 0.304 -.2001743 .5968805 mortgdebtr~o |

L1. | -.0699612 .1920465 -0.36 0.721 -.4818599 .3419375 costfin | .0775615 .080057 0.97 0.349 -.0941436 .2492666 --- Instruments for first differences equation

Standard

D.(L.interestnew)

GMM-type (missing=0, separate instruments for each period unless collapsed) L(3/4).L.shipneu

--- Arellano-Bond test for AR(1) in first differences: z = -0.53 Pr > z = 0.594 Arellano-Bond test for AR(2) in first differences: z = -0.21 Pr > z = 0.835 --- Sargan test of overid. restrictions: chi2(4) = 0.67 Prob > chi2 = 0.955 (Not robust, but not weakened by many instruments.)

Hansen test of overid. restrictions: chi2(4) = 1.56 Prob > chi2 = 0.815 (Robust, but can be weakened by many instruments.)

Difference-in-Hansen tests of exogeneity of instrument subsets:

iv(L.interestnew)

Hansen test excluding group: chi2(3) = 0.83 Prob > chi2 = 0.843

(26)