Matti Viren Department of Economics and Public Choice Research Centre, University of Turku, and Monetary policy and Research Department, Bank of Finland.

Matti Viren Department of Economics and Public Choice Research Centre, University

of Turku, and Monetary policy and Research Department, Bank of Finland

Housing allowance: subsidy to landlords?

Presentation at the RPTU workshop in Bank of Finland, November 26, 2009

Research problem

• Incidence of housing assistance (subsidy), who will finally get the subsidy, who will pay?

• Next to pensions, biggest income transfer in Finland

• Literature: Ditch, J. Lewis, A. and Wilcox, S. (2001) Fack (2006) Gibbons, S., Manning, A. (2003) Kangasharju, A. (2003, 2008) Olsen, E. (2001) Poterba, J. (1984) Rosen, H. (1985) Slesnick, D.T. (1996) Susin, S. (2002)

Policy problem

• Allowances (subsidy to housing) create deadweight losses

• Allowances increase the general rent level; also the rents of those who do not receive allowances will increase

• Income dependence of allowances creates poverty traps

• Large fiscal burden; inflationary bias

Government

Poor households

Other households

Rents increase

House prices increase due to capitalization of subsidies

Rent income increases

Landlords

Tax revenues increase

Transfer of income

What are the demand and supply elasticities?

• D SD’P

Q

Elasticities are probably not very high; no “corner” solutions

Estimating equation

• (v/p)it = a0i + a1(MC/p)t + a2N t + a3Space t + a4Year t

+ a5(Y/p) t + a6(Max/p) t + a7D t + uit,•  where v denotes the rent level, MC the cost of

renting, N the size of household, Year the age of house, Y income, p the price level and Max the maximum amount of subsidy. D is a dummy for not changing the apartment.

• Alternative: “Event study” framework  

The data

• Two set of data are used: (1) Large panel data with 50 000 households for 2000-2008 from the Finnish National Pension Fund which consist of households that have obtained housing allowance in 2008, and (2) Finnish income distribution data that consist of about 26500 households (half of the annual sample is overlapping)

.018

.020

.022

.024

.026

.028

00 01 02 03 04 05 06 07 08

Mean rent/p Mean MAX/p

-.001

.000

.001

.002

.003

.004

00 01 02 03 04 05 06 07 08

Mean D(rent/p)Mean D(MAX/p)

1.5

1.6

1.7

1.8

1.9

2.0

2.1

00 01 02 03 04 05 06 07 08

Mean of rent/MC

% -muutos % -osuusx < - 15 0.31

-15 ≤ x < -10 0.14-10 ≤ x < -5 0.91-5 ≤ x < 0 7.49

x = 0 26.830 < x ≤ 5 50.525 < x ≤ 10 9.9710 < x ≤ 15 2.2615 < x ≤ 20 0.7725 < x ≤ 25 0.22

25 < x 0.58

Annual changes in the rent level, n = 4163

Frequencies of annual changes in rents, %, n=152623

0

200

400

600

800

1,000

0 200 400 600 800 1,000

rent 2001

ren

t 2

00

2

Change of rent for those who receive allowance

0

200

400

600

800

1,000

1,200

0 200 400 600 800 1,000 1,200

rent 2005

ren

t 2

00

6

Change of rent for those who receive allowance

Estimation

• Results in the subsequent tables represents panel data estimates for 2000-2008 with or without fixed effects.

• Table 3 represents cross-estimates for the 2007 Finnish income distribution data

• The displayed results are OLS estimates, here the estimator (e.g. using the GMM) does not seem to make much difference.

(1)level

(2) level

(3)Level

(4)log

(5)Log

(6)Level/m2

Max/P .435(42.02)

.429(26.65)

.269(14.67)

.345(36.36)

.204(20.35)

.589(40.17)

MC/P .352(51.00)

.389(47.66)

.342(43.92)

.235(49.78)

.219(48.03)

4.800(31.75)

Space .185(37.06)

.194(24.69)

.226(23.95)

.361(45.58)

.434(53.72)

-.002(27.40)

N -.185(9.53)

-.310(8.25)

-.006(1.67)

-.022(4.96)

.018(4.02)

-.006(7.62)

Y/P 3.862(30.74)

4.219(8.56)

.003(5.42)

.009(8.96)

.005(5.25)

.014(1.43)

Age of the house

-.013(15.46)

-.020(8.88)

-.027(12.24)

-.527(6.06)

-.008(9.43)

.037(6.77)

No change -.057(16.61)

-.045(14.02)

-.020(16.49)

-.017(14.23)

-.008(10.81)

R2 0.935 0.952 0.956 0.952 0.956 0.895

DW 1.28 1.48 1.49 1.45 1.46 1.47

Fixed E CS CS CS +local CS CS + local CS + local

Rent equation estimates

(1)level

(2) log

(3)log

(4)log

(5)Log

Assistance/p

.503(23.53)

.186(70.09)

.215(84.39)

.068(49.69)

.078(52.72)

Y/P .101(18.01)

.139(48.26)

.172(50.75)

.046(34.78)

.057(36.89)

N .008(61.16)

.328(123.7)

.295(99.89)

.088(56.24)

.080(50.52)

Rentm2/p -4.024(122.1)

-.421(121.5)

-.441(125.9)

-.155(61.58)

-.172(63.40)

No change -.002(21.57

-.019(12.73)

-.020(13.62)

-.058(31.04)

-.060(32.47)

Space-1 .721(213.4)

.711(205.9)

Panel No FE No FE Period FE No FE Local

R2 0.704 0.716 0.723 0.905 0.906

DW 0.37 0.32 0.33 1.56 1.54

Demand equation

(1)level

(2)level

(3)level

(4) level

(5)level

(6)log

Dependent variable

Rent/m2 Rent/m2 Rent m2 m2 Log(m2)

Assistance .169(2.51)

1.515(4.06)

1.776(4.62)

AssistanceDummy

.127(0.84)

28.656(3.47)

.088(6.16)

Y (W/Tr) .024(6.07)

.022(5.57)

.025(6.42

.253(8.57)

.296 / .542(9.05/6.01)

.167(13.81)

N .401(5.59)

.444(6.40)

45.896(7.91)

10.030(18.95)

9.270(15.72)

.366(27.09)

m2 -.079(15.85)

-.079(15.88)

1.799(5.65)

Rent/m2 -2.799(15.74)

-2.773(15.60)

-.355(16.77)

R2 0.242 0.240 0.398 0.566 0.577 0.616

SEE 2.914 2.918 155.0 17.30 17.19 0.273

Results from 2007 cross-section data

1 2 3 4 5 6constant .054

(1.09/1.59).008(0.95)

.018(1.25/1.42)

.032(0.47/0.94)

.011(2.06/)

.016(1.33(1.52)

∆(Space) -.588(8.34/14.65)

-.451(5.17/12.32)

A2001>0 -.033(0.36/0.97)

A2002>0 .072(1.04/2.30)

A2005>0 .041(0.74/1.64)

S2005>0 .069(0.98/2.53)

SEE 0.516 0.520 0.401 0.529 0.530 0.342Method OLS LAD OLS OLS LAD OLSData A2001>0 A2001>0 all 2001

& 2002A2005>0 A2005>0 all 2005

&2006

Results from 2001/2002 and 2005/2005 panels

Conclusions

• More than 20 % of housing subsidy goes to rents

• Subsidy increases housing demand more than other transfers or income in general

• The rent level (rent/m2) does not seem to depend on the allowance (The law of one price.)

• Rents are relatively persistent