Does the tax relief for homeownership have effect on household mortgage leverage?
Abstract
This article presents results of the analysis of the relationship between the tax relief for the homeownership and the household mortgage debt. The advantageous treatment of housing is provided especially by a personal income tax if owner-occupiers do not report imputed rents as income but can deduct mortgage interest costs. This preferred tax status is justified by the existence of positive externalities and a desire to enhance housing opportunities available to citizens. However, evidence that the housing policy via the taxation achieves its objectives is still weak. Moreover the tax provisions for the homeownership benefit rather higher-income households. Furthermore there are indications that the housing taxation encourages levered property purchases and thus contributes to the household debt growth. Since the household indebtedness can have adverse effects on households and macroeconomic performance we focused on the issue whether the income tax relief for homeowners that finance their dwellings via a mortgage does affect the household leverage. We constructed the variable capturing especially the mortgage interest payments deductibility. We employed the multiple regression and data for the former 15 EU member countries (except Greece) for the period 2004-2013. We estimated two models for two representative taxpayers who vary in a family status using the panel data analysis with fixed effects. From our results we inferred that the income tax relief for the homeownership might not have influenced the mortgage leverage significantly in the selected European countries in the given period. The mortgage debt was affected rather by the economic level, price of own housing, mortgage interest payments or demographic structure.
Full text
52 2018, XXI, 1
Ekonomie
DOI: 10.15240/ ul/001/2018-1-004
In oduc ion
Tax sys ems o de eloped coun ies con ain
p o isions ha gi e a p e e ed s a us o
housing and homeowne ship. A signi i can
elie is p o ided especially by he combina ion
o non- axa ion o impu ed en al income and
mo gage in e es paymen deduc ibili y. Real
use cos s o owne -occupied housing a e
educed (see Po e ba & Sinai, 2008) and hus
a bias in a ou o he homeowne ship is c ea ed
when households a e encou aged o buy a he
han o en hei dwellings. Al hough he e is
a p ope y ax, ha is le ied o ax he impu ed
en , he elie o he mo gage in e es paymen
can be so gene ous ha i mi iga es he e ec
o he p ope y ax (C owe, Dell’A iccia, Igan,
& Rabanal, 2011). The ac is ha he p ope y
ax in many coun ies is no high enough o be
a pe ec subs i u e o he impu ed en axa ion
(Hemmelga n & Nicodéme, 2010). Fu he mo e,
homeowne ship is mo e a ac i e when capi al
gains a e no axed in a neu al manne . And
p o i s om home sales a e no usually axed.
The housing- ela ed ax allowances a e one o
main ax expendi u e i ems in EU coun ies (see
Eu opean Communi ies [EC], 2013).
The p o-homeowne ship ax policy is
jus i i ed by ma ke ailu es and a desi e o
enhance housing oppo uni ies a ailable
o ci izens (And ews, Calde a Sánchez, &
Johansson, 2011). Howe e i seems ha o
he ime being economis s a he conclude ha
he ax ad an ages p o ided o homeowne s
a e no e y e i cien a achie ing gi en
objec i es. A gene al goal in p ac ice is o boos
he homeowne ship o o inc ease housing
demand and consump ion. Cecche i, Mohan y,
and Zampolli (2011) claim ha he gene ous ax
elie o mo gage in e es paymen could ha e
played a ole in expanding he homeowne ship
in some coun ies. Howe e acco ding o
And ews, Calde a Sánchez, and Johansson
(2011) he e is no clea c oss-coun y
ela ionship be ween he ex en o he mo gage
in e es deduc ibili y and he homeowne ship
a es. Ne e heless hey admi ha households’
p e e ence o he homeowne ship can be
in l uenced also by ax policy. Bou assa and
G igsby (2000) o Glaese and Shapi o (2002),
based on empi ical e idence on U.S. si ua ion,
concluded ha impac o he mo gage in e es
deduc ibili y on he homeowne ship a e was
minimal. And C owe e al. (2011) e en asse
ha he homeowne ship a es a e nega i ely
ela ed o he ex en o he ad an ageous ax
ea men o homeowne ship.
An a gumen in a ou o a highe a e o
he homeowne ship is exis ence o posi i e
ex e nali ies, e. g. enjoymen o neighbou s
and passe sby gene a ing h ough home
main enance and ga dening, be e ou comes
o child en, and long- e m p ospec s o
a communi y (Bou assa & G igsby, 2000 o
Glaese & Shapi o, 2002). Glaese and Shapi o
(2002) hink ha e idence on ex e nali ies is
weak bu sugges i e. Bu Hilbe and Tu ne
(2013) belie e ha he homeowne ship
gene a e ew o no posi i e ex e nali ies.
Mo eo e , he e could be nega i e ex e nali ies
ela ed o housing consump ion, e. g. lea ing
small ci y apa men s o la ge places on
he inge o a ci y, inc easing seg ega ion by
income o en y inci ed by ancy homes (Glaese
& Shapi o, 2002). A lowe labou mobili y
and highe unemploymen among owne -
occupan s in compa ison wi h he mobili y
and unemploymen o en e s can be ha m ul
e ec o he homeowne ship as well (And ews,
Calde a Sánchez, & Johansson, 2011). Finally,
he homeowne s could u ilize hei poli ical
powe o cu o new house cons uc ion in
o de o aise hei house p ices (Glaese &
Shapi o, 2002). As O’Sulli an and Gibb (2012)
summed i up, he homeowne ship does no
DOES THE TAX RELIEF FOR
HOMEOWNERSHIP HAVE EFFECT
ON HOUSEHOLD MORTGAGE LEVERAGE?
Ba bo a Slin áko á, S anisla Klaza
EM_1_2018.indd 52EM_1_2018.indd 52 21.3.2018 12:01:3521.3.2018 12:01:35
53
1, XXI, 2018
Economics
gene a e economic bene i s o he mac o-
economy o speci i c households which would
jus i y a gene al p og amme o ax concessions
o homeowne s.
As an a gumen suppo ing he owne -
occupa ion we could p esen ha accumula ed
housing weal h could se e as means o p i a e
insu ance (Ansell, 2013) o ha home equi y
could p o ide an addi ional sou ce o e i emen
income beyond pensions, i.e. e e se
mo gages (Toussain , 2013). Howe e we a e
no amilia so a wi h any e idence ha he
owne -occupa ion se es well as an ins umen
o pension secu i y. On he o he hand, wha is
al eady a subjec o examina ion is co ela ion
be ween he mo gage in e es deduc ibili y
and g ow h o house p ices o inc eased house
p ices ola ili y. O ganisa ion o Economic Co-
ope a ion and De elopmen [OECD] (2009) o
And ews, Calde a Sánchez, and Johansson
(2011) o e some e idence o a posi i e
co ela ion be ween he ax elie on mo gage
in e es and a iabili y in house p ices. Keen,
Klemm, and Pe y (2010), And é (2010),
Su he land, Hoelle , Me ola and Ziemann
(2012) o u he au ho s poin ou ha he ax
subsidy is likely, depending on a p ice elas ici y
o housing supply, o be capi alised in o house
p ices. In addi ion he ax incen i e o deb -
i nancing o e o he sou ces o i nancing o
own dwellings could esul in o e in es men
in housing and misalloca ion o capi al s ock
wi h nega i e e ec s on a long- e m economic
g ow h (see e.g. Saa imaa, 2009; And é, 2010;
Hemmelga n & Nicodéme, 2010; Ven y, 2010).
Finally, he e is ag eemen among au ho s
ha he housing ax elie is eg essi e, i.e.
i a ou s ich households. Highe -income
households bene i om he ad an ageous ax
ea men o housing, namely om he mo gage
in e es deduc ibili y, mo e han lowe -income
households because hey ha e highe a es o
he homeowne ship, hey buy mo e expensi e
houses, and hei ma ginal ax a es a e highe
(see Bou assa & G igsby, 2000; And é, 2010;
Keen, Klemm, & Pe y, 2010; Ma saganis, 2014;
Ven y, 2010; And ews, Calde a Sánchez, &
Johansson, 2011).
The e is ano he impo an eason why
he housing axa ion, especially he mo gage
in e es deduc ibili y, a ac a en ion o
economis s. I is because i migh suppo
deb c ea ion. Jo da, Schula ick, and Taylo
(2014) ound ha household le e age a ios
ha e inc eased subs an ially in many coun ies
o e he 20 h cen u y: abou wo hi ds o
bank lending oday consis s o he loans o
he household sec o o he pu chase o eal
es a e, acco ding o Eu opean Cen al Bank
[ECB] (2014) he majo i y o he Eu opean
household bo owing is comp ised by loans o
house pu chase. Fu he mo e hey showed ha
con empo a y business cycles a e inc easingly
shaped by he dynamics o he mo gage
c edi and ha he mo gage c edi became
a speci i c sou ce o i nancial ins abili y in
ad anced economies a e he Second Wo ld
Wa . And Jo da, Schula ick, and Taylo (2015)
demons a ed ha he c edi - i nanced housing
p ice bubbles a e mo e dange ous o i nancial
sec o and eal economy han he unle e aged
bubbles. How indeb edness can a ec
mac oeconomic pe o mance and households,
see Su he land e al. (2012) o McGowan
(2013). The household indeb edness impac on
economic g ow h ha e been analysed by Izák
(2012).
A ques ion aises whe he he housing
axa ion, and he p esupposed deb bias
embedded in i , does a ec he household
indeb edness, and whe he he nega i e
consequences o high le els o deb could
be, pa ly, ex a cos o he axa ion a ou ing
housing. Hemmelga n and Nicodéme (2010),
Keen, Klemm, and Pe y (2010), Be na di
(2011), Cecche i, Mohan y, and Zampolli
(2011), Hemmelga n, Nicodéme, and Zanga i
(2011), Su he land e al. (2012) o Ga nie e al.
(2013) sugges ha he ax ea men o owne -
occupied housing may ha e played a ole.
Households ha e been p obably encou aged
by he mo gage in e es elie o p e e
bo owing. As a esul he household deb may
be highe han i would be o he wise. Wolswijk’s
(2005) simple g aphical analysis shows ha he
lowe he i nancing cos s, dec eased by he ax
subsidy, he highe he deb -GDP a io. Keen,
Klemm, and Pe y (2010) a i m ha coun ies
o e ing mo e a ou able ax ea men o he
homeowne ship ha e highe a ios o mo gage
deb . On he o he hand C owe e al. (2011)
do no see a signi i can ela ion be ween he
ax ea men o housing and he a io o he
mo gage deb o GDP. I is necessa y o be
ca e ul, e.g. Ellis (2006, p. 11) wa ns ha i is
no easy o p o e he co ela ion be ween he
ax ad an ages o he homeowne ship and he
household deb since “ he ax egime in e ac s
EM_1_2018.indd 53EM_1_2018.indd 53 21.3.2018 12:01:3521.3.2018 12:01:35
54 2018, XXI, 1
Ekonomie
wi h o he aspec s o he housing- i nance
sys em in some imes complex ways”. OECD
(2009) poin s ou ha home-equi y loans and
eal house p ices ha e isen in many coun ies
despi e he ac ha ax incen i es a ied
conside ably. I sugges s ha o he ac o s
a ec hese phenomenons. Howe e OECD
(2009) also admi s ha he high ax elie on
mo gage in e es co ela es wi h he high
a iabili y in house p ices which can lead o
se ious household c edi p oblems.
Deepe insigh in o he ela ionship be ween
he ax incen i e and household deb can be
p o ided by s udies conce ning indi idual
coun ies. Dunsky and Follain (2000) o Mun oe
(2014) analysed he Uni ed S a es da a and
ound ou ha he demand o home mo gage
deb esponded o he mo gage in e es
deduc ion. Hende sho , P yce, and Whi e
(2002) e ealed ha he homeowne le e age
in he Uni ed Kingdom was sensi i e o he
deduc ibili y o mo gage in e es . Alan and Le h-
Pe e sen (2006) examined he 1987 ax e o m
in Denma k which made he holding o deb less
a ac i e. Rouwendal (2007) iden i i ed he ax
incen i e o i nancing he homeowne ship wi h
a mo gage as impo an d i ing o ce behind
he inc ease in equency o mo gage use in
he Ne he lands. Acco ding o Somme oll
(2007) No wegian households educed hei
deb as a esponse o less gene ous in e es
deduc ions a e he ax e o m. Finally,
Saa imaa’s (2009) esul s om he s udy o he
impac o he ax e o m in Finland indica e ha
high income households wi h high ma ginal ax
a es esponded o he ax incen i e educ ion
by clea ly dec easing hei mo gage bo owing.
On he o he hand Jappelli and Pis a e i (2004)
ound no e idence ha he ax ea men
shaped he demand o mo gage deb in I aly.
I is possible o sum up ha ax e o ms, ha
educed alue o he mo gage in e es elie
(which is a ec ed also by a ma ginal ax a e)
and hus weakened he incen i e o bo ow
in o de o own home, mos ly led o lowe
household le e age. This could p o e he
ole o he ax incen i e a ou ing he deb -
i nanced homeowne ship (see he Eu opean
Commission axa ion pape s abou ax e o ms
published in 2011-2014). A e all, he Eu opean
Commission ecommends o he EU Membe
S a es o educe he deb bias in hei housing
axa ion and subsequen ly maps changes in
axa ion ules in he membe coun ies, see e.g.
(EC, 2014). Ga nie e al. (2013) con i m ha
majo changes in he housing axa ion in many
coun ies conce ned he deb bias and ocused
on limi ing he deduc ibili y o mo gage in e es .
Ne e heless Somme oll (2007) poin s ou
ha e ec s o ax e o ms on housing ma ke s
can be o se by o he ac o s in he economy.
O cou se, he housing axa ion is no
he only ac o in l uencing he household
demand o mo gage. In e es a es, i nancing
condi ions (i.e. i nancial de egula ion and
inno a ion on i nancial ma ke ), income and
demog aphy ha e been iden i i ed in bo h
heo e ical and empi ical s udies as impo an
de e minan s o he household indeb edness.
In addi ion, ac o s cha ac e ising housing
ma ke , especially house p ices, should be
aken in o conside a ion, oo – see e.g. Debelle
(2004), Jacobsen and Naug (2004), Gi oua d,
Kennedy, and And é (2006), Dynan and Kohn
(2007), ECB (2009), And é (2010) o Bokha i,
To ous, and Whea on (2013).
Aim o ou esea ch was o explo e whe he
he e is he ela ion be ween he income ax
incen i e o homeowne s, i.e. he mo gage
in e es deduc ibili y and exemp ed impu ed
en income, and he household indeb edness.
To he con a y o s udies men ioned abo e,
which analysed mic oda a, we conduc ed
a c oss-coun y s udy like Wolswijk (2005)
who analysed impac o he housing axa ion
on he mo gage deb g ow h in EU coun ies
on assump ion ha he axa ion may ha e
a po en ially la ge ole in a ec ing he household
le e age. The emainde o he pape is di ided
in o h ee sec ions and a conclusion. In he nex
sec ion we ocus on he cons uc ion o ou
key explana o y a iable ep esen ing he ax
ea men o he owne -occupied housing which
is supposed o be deb encou aging. Then we
desc ibe b ie l y he o he explana o y a iables
we wo ked wi h. The econome ic me hods we
used a e discussed in sec ion 2 and sec ion 3
epo s esul s.
1. Va iables Used in he Household
Mo gage Le e age Reg ession
Model
We ollowed Wolswijk’s (2005) wo k in ou
esea ch. Wolswijk applied a mul iple (panel)
eg ession analysis in o de o measu e he
e ec o he i scal ins umen s on he mo gage
deb g ow h in he EU coun ies. We also
employed he eg ession analysis whe e
EM_1_2018.indd 54EM_1_2018.indd 54 21.3.2018 12:01:3521.3.2018 12:01:35
55
1, XXI, 2018
Economics
he dependen a iable was he household
mo gage le e age. The mo gage le e age
was measu ed by he a io o o al ou s anding
esiden ial loans (on lende ’s books a he
end o he yea , in EUR million) o household
g oss disposable income (in EUR million). The
da a on bo h loans and income we e p o ided
by he HYPOSTAT 2014 published by he
Eu opean Mo gage Fede a ion – Eu opean
Co e ed Bond Council (EMF-ECBC) – see he
webpage h p://www.hypo.o g/Con en /de aul .
asp?PageID=524. The e ec o he income ax
incen i e o homeowne s was measu ed by
a special a iable – see sec ion 1.1. Conside ing
li e a u e and da a a ailabili y we p oposed
a numbe o he o he a iables which migh
explain he household deb – see sec ion 1.2.
1.1 Cons uc ion o he Homeowne
Tax Relie Va iable
Wolswijk (2005) cons uc ed he i scal
ins umen s a iable as he a e - ax capi al
cos s which cap u ed he deduc ibili y o
mo gage in e es paymen . The capi al cos s
we e de i ed om a nominal mo gage in e es
a e and amoun o mo gage. Then hey we e
adjus ed by he ele an ax a e which e l ec ed
whe he he mo gage in e es was deduc ible
om he income ax and whe he i was
deduc ible ully o wi h a limi . The cos s we e
exp essed as he pe cen age o a house p ice.
Since we assumed ha indi iduals conside
he mo gage in e es and he elie p o ided
by he income ax law a he sepa a ely when
hey make decision on bo owing we decided
o include wo explana o y a iables in o
ou eg ession model. We used he a iable
ep esen ing he impac o he income ax
ea men o housing on he homeowne capi al
cos s besides he mo gage in e es paymen
as ano he a iable.
Ou key explana o y a iable ( he
homeowne ax elie ) was de i ned as he a io
o he p e- ax capi al cos s (CC0) o he a e - ax
capi al cos s (CC1):
(1)
whe e is nominal mo gage in e es a e; M is
mo gage alue; TW is ax wedge.
I he a io is highe han 1, housing and
homeowne s a e subsidised by he income ax
sys em. The a io smalle han 1 means axa ion
o housing, i.e. impu ed en is axed since i
exceeds he mo gage in e es paymen . The
di e ence be ween he p e- ax capi al cos s
and he a e - ax capi al cos s is called he
ax wedge and i is used as an indica o o he
ex en o he ax elie on deb i nancing o he
owne -occupied housing (see e.g. And ews,
2010). The ax wedge has been calcula ed also
by an den Noo d (2003) when he es ima ed
he eal i nancing cos s o housing.
To es ima e he capi al cos s be o e and
a e axa ion and he ax wedge i is necessa y
o model a axpaye who migh ep esen he
popula ion aking ou mo gages. The e we e
wo inspi a ions a ailable o be used. Fi s ,
we ollowed Keen, Klemm, and Pe y (2010),
who used he In e na ional Mone a y Fund
me hodology, see Hemmelga n, Nicodéme and
Zanga i (2011) o In e na ional Mone a y Fund
(2009), espec i ely, in o de o calcula e e ec i e
ax a es on he owne -occupied housing o
he pu pose o in e na ional compa ison. We
made he ollowing assump ions abou ou
model axpaye : he is an unma ied pe son
wi h no child in he op income ax b acke
who pu chased his p ima y dwelling o a p ice
i nanced 80% wi h a mo gage a a ce ain
in e es a e. The assump ion abou he op
income axpaye was suppo ed by Gi oua d,
Kennedy, and And é (2006) acco ding o whom
he mos indeb ed households a e hose wi h
he highes incomes. The op income was
de i ned as double he a e age wage which is
he h eshold o he op income ax b acke in
he mos EU-15 coun ies. The a e age wages
da a we e d awn om he OECD Taxing Wages
publica ions. The p ice o dwelling was es ima ed
as 5-mul iple o he op income. The mo gage
in e es a e was app oxima ed by he annual
a e age o ep esen a i e in e es a es on new
esiden ial loans p o ided by he HYPOSTAT
2014. Mo eo e , we used he nominal a e like
Wolswijk (2005) who a gued ha i is mo e
ele an o indi idual’s decision-making on he
mo gage. Second, since an den Noo d (2003)
calcula ed his ax wedge o a single ea ne
couple wi h wo child en and Wolswijk (2005)
es ima ed he ele an ax a e o an a e age
amily, we c ea ed an al e na i e model axpaye .
We modi i ed he model axpaye ’s amily s a us:
he is ma ied wi h wo child en. We p o ide
esul s o eg ession analysis o bo h single and
amily axpaye s in he sec ion 3.
EM_1_2018.indd 55EM_1_2018.indd 55 21.3.2018 12:01:3621.3.2018 12:01:36
56 2018, XXI, 1
Ekonomie
Like Wolswijk (2005) we wan ed o include
he o me 15 EU membe s a es in o ou
analysis bu due o a lack o da a ou coun y
sample does no con ain G eece. We co e ed
he pe iod om 2004 o 2013, i.e. we s a ed
when Wolswijk ended. To calcula e he a e -
ax capi al cos s we had o es ima e he ax
wedge o e e y coun y and each yea using
e ec i e income ax law p o isions. We ook
in o accoun he mo gage in e es deduc ibili y
as well as axa ion o impu ed en jus as an
den Noo d (2003). Fu he mo e we explici ly
dis inguished di e en ax elemen s used in
income ax codes o a ou he owne -occupied
housing and deal wi h limi s on he deduc ible
amoun mo e exac ly han Wolswijk (2005).
The e we e di e en o ms o he ax elie
in he EU coun ies in he gi en pe iod. Aus ia,
Belgium, Denma k, Luxembou g and he
Ne he lands applied a deduc ion om income
o a ax base. When he deduc ion is used he
amoun , which a axpaye could sa e, depends
no only on he mo gage in e es paymen bu
also on a ma ginal ax a e ( he ma ginal ax
a es we e gained om he OECD Taxing Wages
publica ions). Mo eo e , Belgium, Luxembou g
and he Ne he lands axed he impu ed en – hen
he in e es is deduc ible i s agains his income.
The impu ed en was se ei he as a ac ion
o he house p ice (i.e. o he Ne he lands,
acco ding o he Eu opean Commission axa ion
pape s) o as a pe cen age o household income
(i.e. o Belgium and Luxembou g, acco ding o
Eu os a (2013)). A c edi lowe ing a ax liabili y
was applied in Finland, F ance, I eland, I aly,
Po ugal, Spain and Sweden. The ax wedge
in his case is easie o es ima e because i is
equal o he in e es . The e was no ax elie in
Ge many and he Uni ed Kingdom in he pe iod
2004-2013.
In ac we applied complica ed o mulas
o he ax wedge es ima ions because he e
we e a ious limi s on he amoun s o he
deduc ions o c edi s. The amoun s we e
usually de i ed om he mo gage in e es
paymen as a ce ain pe cen age and hey we e
mos ly capped wi h a ceiling which somewhe e
depended on axpaye ’s cha ac e is ics, e.g.
income. Especially, he p esence o wi e and
child en inc eased he ceiling on he c edi in
Finland, F ance and I eland.
The ax wedge should be equal o he sum
o annual amoun s sa ed du ing he pe iod o
epaying a mo gage. Howe e since ou model
axpaye s in all he coun ies we e assumed
o epay a mo gage o he same ime we
could simpli y he calcula ion. The sa ings on
he ax liabili y we e es ima ed as he amoun
sa ed in he i s yea o he epaymen pe iod.
Mo eo e we assumed ha he ax ea men o
he mo gage in e es and impu ed en would
be s able o e he whole epaymen pe iod. In
ac , he ax ea men o he owne -occupied
housing signi i can ly changed in Belgium,
F ance, I eland, Po ugal o Spain du ing he
pe iod 2004-2013.
P ecision o ou ax wedges es ima es
was in l uenced by quali y o a ailable da a.
We had o ely on in o ma ion abou axa ion
p o ided by seconda y sou ces. We compiled
he in o ma ion om he Taxes in Eu ope
da abase (see he webpage h p://ec.eu opa.
eu/ axa ion_cus oms/ axa ion/gen_in o/in o_
docs/ ax_in en o y/index_en.h m), he OECD
Taxing Wages publica ions and he Eu opean
Commission axa ion pape s published in
2011-2014. Quali y o he da a di e ed ac oss
coun ies: some coun ies did no p o ide so
much de ails and some imes di e en sou ces
did no p o ide he same da a abou he same
coun ies.
Nei he an den No d (2003) no Wolswijk
(2005) included he p ope y ax o he
homeowne in o hei calcula ions o he ax
wedge o he a e - ax capi al cos s, espec i ely.
We also igno ed his ax on he assump ion ha
he e is no di e ence in he use cos s be ween
own and en ed housing since he p ope y ax
bu den is assumed o be ully bo ne by he
indi idual using he dwelling (i.e. an owne o
a enan ). Fu he mo e, Wolswijk (2005) used
a modi i ed i scal ins umen s a iable, which
included axa ion o expec ed capi al gains
om he house sale, in ano he speci i ca ion
o his eg ession model bu wi h no sa is ac o y
esul s. We igno ed he capi al gain axa ion
( he capi al gain is usually ax-exemp ) as well
as ansac ion axes on assump ion ha he
model axpaye s do no conside selling hei
dwelling in u u e. A e all he mo gage in e es
deduc ibili y combined wi h he non- axa ion
o impu ed en a e he p o isions which a e
assumed o il households o bo owing he
mos .
1.2 The O he Explana o y Va iables
D awing om li e a u e and checking a ailabili y
o s a is ical da a we iden i i ed he ollowing
EM_1_2018.indd 56EM_1_2018.indd 56 21.3.2018 12:01:3621.3.2018 12:01:36
57
1, XXI, 2018
Economics
a iables which migh impac on he household
indeb edness and which would be sui able o
he eg ession analysis. On assump ion ha
weal hie households a e able o bo ow mo e
money in o de o buy houses we used he
GDP and i nancial wo h pe capi a a iables
ep esen ing weal h. Financial wo h pe capi a
was de i ned as he household i nancial ne
wo h pe capi a o annual household income
pe capi a a io. Supposing ha he unemployed
people ha e limi ed oppo uni ies o bo ow he
unemploymen ( a e) a iable was added, oo.
Mo gage in e es paymen as a po ion
o household income cons i u ed he a iable
ep esen ing cos s o bo owing in ou analysis.
The in e es paymen was de i ed om he
annual a e age o ep esen a i e in e es a es
on new esiden ial loans.
Fu he unde lying d i e o he housing
deb a e house p ices. We supposed ha he
p ices could impac on he mo gage demand
bo h nega i ely and posi i ely. I he p ices
inc ease households ha e o ake ou la ge
mo gages. On he o he hand inc eased p ices
can educe demand on he housing ma ke and
hus a lowe amoun o mo gage is necessa y
o acqui e dwellings. The house p ices
de elopmen was app oxima ed by he nominal
house p ices index om he HYPOSTAT 2014
which e l ec s he changes in house p ices
obse ed o e he gi en pe iod. In addi ion o
his “p ice” a iable we included he indica o o
he di e ence be ween he HYPOSTAT house
p ices index and he ha monised index o ac ual
en s om he EUROSTAT s a is ics, he ea e
he home- en wedge a iable. The e ec o he
home- en wedge on he mo gage le e age
was supposed o be nega i e, i. e. i he own
housing p ice inc eases in ela ion o he en al
housing p ice he amoun o mo gage deb
aken up is supposed o dec ease because
households p e e he en al housing.
To cap u e he in l uence o popula ion
s uc u e which migh de e mine he demand o
housing we wo ked wi h bo h he young and old
dependency a ios. Since senio s ha e al eady
secu ed hei housing he le el o he household
mo gage le e age should dec ease wi h
a highe po ion o olde people in popula ion.
On he o he hand we assumed a posi i e e ec
o he young dependency a io because young
people should ha e a highe need o secu e
hei shel e . Mo eo e hei i nancial si ua ion
p esumably allows hem o bo ow.
Finally, since he e we e changes in
measu ing he da a necessa y o ou dependen
a iable in di e en yea s o di e en coun ies
we used a dummy a iable o con ol o he
ime se ies b eaks.
Da a o he independen a iables we e
ob ained om a ious sou ces, i.e. om
he HYPOSTAT 2014, he OECD s a is ics,
EUROSTAT s a is ics and he ILO s a is ics.
S a iona i y o da a is eques ed o a p ope
calcula ion o p- alues in eg ession models o
panel da a. We used he Le in-Lin-Chu es
(LLC), assuming ha he e was a common uni
oo p ocess iden ical ac oss c oss-sec ions,
as well as Dickey–Fulle es (ADF-GLS),
allowing o indi idual uni oo p ocesses o
a y ac oss c oss-sec ions, in o de o unco e
he p esence o uni oo s in ime se ies o da a
o he a iables desc ibed abo e o all he
coun ies in ou sample. The es designed by
Le in, Lin, and Chu (2002) was applied wi h
he cons an including 0 lags. The ADF es
was done using he GLS p ocedu e sugges ed
by Ellio , Ro henbe g, and S ock (1996) wi h
a g ea e powe han he s anda d Dickey–
Fulle app oach. Mo eo e , he o e all es wi h
null hypo hesis, ha he se ies in ques ion had
a uni oo o all he panel uni s, was calcula ed
using he me hod o Im, Pesa an, and Shin
(2003). Resul s o he in e se chi-squa e,
in e se no mal and logi es s we e agg ega ed
using he Choi me a- es (see Choi, 2001;
Co ell & Lucche i, 2016) – i nal esul s a e
shown in he las column o Tab. 1. The same
null hypo hesis (H0) and al e na i e hypo hesis
(H1) we e o mula ed o bo h es s: H0 was
ha all o he indi idual ime se ies exhibi a uni
oo , and H1 was ha none o he se ies has
a uni oo . Fu he mo e i he p- alue was low
he null hypo hesis was ejec ed which mean
ha a ime se ies was s a iona y. Resul s o he
s a iona i y es s a e in Tab. 1.
The p- alues showed ha he i nancial
wo h pe capi a and nominal house p ices index
se ies we e non-s a iona y. The da a o he
i nancial wo h pe capi a exhibi ed he uni oo s
and he s a iona i y o hese da a was ob ained
by i s di e ences. The p oblem wi h non-
s a iona i y o he nominal house p ices index
se ies was sol ed by using da a on (annual and
ela i e) changes in his index which we e also
a ailable in he HYPOSTAT 2014. The change
in nominal house p ices indices se ies was
al eady s a iona y and a new a iable, labelled
EM_1_2018.indd 57EM_1_2018.indd 57 21.3.2018 12:01:3621.3.2018 12:01:36
58 2018, XXI, 1
Ekonomie
“house p ices”, was se up. See he p- alues o
he adjus ed se ies in he bo om o he Tab. 1.
Values o he LLC es o he GDP pe capi a
a iable we e ambiguous, ne e heless we
decided o conside he se ies as s a iona y. All
he o he se ies we e s a iona y.
Tab. 2 p esen s desc ip i e s a is ics o all
he a iables used in a eg ession analysis,
a iable
LLC ADF-GLS
coe i cien - a io z-sco e [p- alue] Me a- es s p- alue
household mo gage le e age -0.303110 -8.524 -6.40937 [0.0000] < 0.00
single homeowne ax elie -0.607120 -6.050 -2.98838 [0.0014] NA
amily homeowne ax elie -1.012400 -10.698 -6.9322 [0.0000] NA
GDP pe capi a -0.223060 -3.771 -1.18036 [0.1189] < 0.00
i nancial wo h pe capi a -0.021903 -0.567 1.58079 [0.9430] 0.735
unemploymen -0.606710 -9.791 -5.65002 [0.0000] < 0.00
mo gage in e es paymen -0.231590 -4.070 -1.48657 [0.0686] < 0.01
nominal house p ices index -0.092375 -2.230 0.24923 [0.5984] < 0.00
home- en wedge -0.679620 -8.305 -5.3787 [0.0000] < 0.00
young dependency a io -0.115020 -8.512 -7.70736 [0.0000] < 0.00
old dependency a io -0.116960 -7.551 -6.9874 [0.0000] < 0.00
b eak-poin dummy -0.154430 -4.144 -2.74214 [0.0031] < 0.00
adjus ed se ies
i nancial wo h pe capi a (1s -di .) -0.661910 -8.286 -4.09091 [0.0000] < 0.00
house p ices (change in %) -0.495910 -7.848 -5.21249 [0.0000] < 0.00
Sou ce: own
Tab. 1: S a iona i y es s esul s
Va iable Mean Median Minimum Maximum S d. De .
household mo gage le e age 0.990606 0.886642 0.190009 2.26104 0.519354
single homeowne ax elie 1.1738 1.1269 0.98624 2.6178 0.23242
amily homeowne ax elie 1.2650 1.1401 0.98624 2.6178 0.34274
GDP pe capi a 35,193 33,462 14,534 87,231 13,302
i nancial wo h pe capi a (1s -di .) 1.1167 1.0687 0.38264 2.2013 0.40244
unemploymen (%) 7.81508 7.53600 2.75100 26.0920 3.79661
mo gage in e es paymen 0.30541 0.30469 0.11224 0.84372 0.10831
house p ices (change in %) 2.5528 2.5337 -18.776 30.769 6.6330
home- en wedge -0.26071 0.0000 -18.700 20.000 3.3553
young dependency a io (%) 22.685 23.123 18.138 28.236 2.3741
old dependency a io (%) 4.4620 4.5180 2.6265 6.1160 0.79377
Sou ce: own
Tab. 2: Desc ip i e s a is ics o non-bina y a iables, obse a ions o 14 EU coun ies,
2004-2013
EM_1_2018.indd 58EM_1_2018.indd 58 21.3.2018 12:01:3621.3.2018 12:01:36
59
1, XXI, 2018
Economics
i.e. he dependen a iable as well as he
independen a iables.
2. Discussion o Econome ic
Me hods
Ou analysis was based on panels which can be
desc ibed as balanced. The e was an obse a ion
o e e y yea (2004-2013, i.e. 10 yea s) and o
e e y uni (i.e. 14 coun ies).
In ou pilo wo k (see Slin áko á & Klaza ,
2015a o Slin áko á & Klaza , 2015b) we
used he pooled OLS me hod because o i s
wo ad an ages. Fi s , we gained in e i ciency
because i was no necessa y o allow o
non-exis en wi hin-g oups au oco ela ion.
Second, we ook ad an age o he i ni e-sample
p ope ies ins ead o elying on he asymp o ic
p ope ies o andom e ec s (see Doughe y,
2007). In case he unobse ed e ec is
weak, i. e. he e a e no ele an unobse ed
cha ac e is ics, he pooled eg ession is
he bes me hod o desc ibe a ela ionship
be ween an explained a iable and explana o y
a iables. We belie ed ha we iden i i ed such
explana o y a iables on he basis o li e a u e
s udy and he da a a ailabili y examina ion
ha we could suppose ha he unobse ed
componen would no in l uence ou analysis
signi i can ly. Howe e we could no be su e
ha he e we e any di e ences among he
coun ies in ou sample. Tha is why we decided
o es sui abili y o he panel da a analysis
(see Doughe y, 2007 o Guja a i, 1995). The
es o di e ing g oup in e cep , implemen ed
in he g e l so wa e, was pe o med wi h he
null hypo hesis ha g oups (coun ies) ha e
a common in e cep . The es esul ed wi h
he ollowing es s a is ic: Welch F(13, 48.1)
= 139.847 o he single axpaye model and
288.331 o he amily axpaye model, bo h
wi h p- alue nea ze o. The null hypo hesis
was ejec ed and we concluded ha he panel
da a analysis was mo e sui able ool han he
pooled OLS me hod. The panel da a me hod
would enable o con ol o he di e en le el
o he ela ionship be ween he explained and
explana o y a iables due o di e en coun y
cha ac e is ics.
The e a e wo basic panel da a echniques:
i xed e ec s and andom e ec s. We ollowed
he economic undamen als (a less o mal hin )
as well as mo e o mal ( echnical) econome ic
es s in o de o decide which echnique should
be p e e ed. The less o mal decision was
based on he ecommenda ion o Doughe y
(2007) who, in case ha an obse a ion canno
be desc ibed as being a andom sample om
a gi en popula ion, ecommends using he i xed
e ec s. We supposed ha ou selec ed coun ies
sample could no be a andom sample. We
belie ed ha economic na u e o ou da a was
no compa ible wi h he p esence o andom
e ec s. The mo e o mal es s o B eusch-
Pagan and Hausman we e pa ly in con l ic . The
B eusch-Pagan es (wi h he null hypo hesis
ha he a iance o he uni -speci i c e o is ze o
e sus he al e na i e hypo hesis o exis ence
o andom e ec s) deli e ed asymp o ic es
s a is ic: Chi-squa e(1) = 426.747 o he single
axpaye and 496.52 o he amily axpaye
wi h bo h p- alue nea ze o. The non- andom
e ec s hypo hesis was ejec ed a he 5% le el.
Howe e he Hausman es sugges ed ha he
andom e ec s es ima o s we e inconsis en
and hus he i xed e ec s should be used
ins ead. The Hausman es esul s a e shown
below he Tab. 3. To sum up, econome ic es s
and he na u e o he analysed da a p o ed
ha he i xed e ec s would be mo e sui able
me hod o analyse he ela ionship be ween
he explained and explana o y a iables in ou
case.
We epo esul s om bo h he i xed e ec s
and andom e ec s models in he Tab. 3 in
he sec ion 3. Compa ison showed ha he
coe i cien alues om he i xed e ec s models
we e no , in gene al, signi i can ly di e en han
he coe i cien alues om he andom e ec s
models. The e o e he i xed e ec s models
esul s can be conside ed being obus .
Fu he mo e, since he explana o y
a iables e ec s migh di e no only ac oss
coun ies bu also o e ime he pe iod i xed
e ec s we e es ed by Wald es o join
signi i cance o ime dummies (H0 said ha
he e we e no signi i can pe iod e ec s). On
he basis o he es esul s (see no es below
he Tab. 3) we included ime dummies in ou
eg ession models.
Acco ding o Doughe y (2007)
a eg ession model can be buil ei he om
speci i c o gene al o om gene al o speci i c.
On he basis o his summa y o p os and
cons i seemed ha he om speci i c o
gene al app oach would be mo e sui able
o ou analysis. We de o ed a lo o e o o
s udy ac o s in l uencing he household deb
and s a ing wi h he mos impo an ac o s
EM_1_2018.indd 59EM_1_2018.indd 59 21.3.2018 12:01:3621.3.2018 12:01:36
60 2018, XXI, 1
Ekonomie
men ioned in li e a u e (i.e. GDP, i nancial
wo h and in e es paymen ) and adding o he
po en ially less impo an ac o s seemed o
be mo e app op ia e app oach. Some au ho s
ecommend his app oach in he case ha signs
o coe i cien s a e co ec , see And ews (2010).
We cons uc ed he sequen ial model and we
pe o med sui able diagnos ic checks in e e y
s ep. We belie e ha ou heo e ical analysis
allowed us o cope wi h he p oblem o model
misspeci i ca ion. We es ima ed he models o he
household mo gage le e age wi h he panel da a
analysis wi h bo h he i xed and andom e ec s. To
ackle he p oblem o he e oskedas ici y we used
obus s anda d e o s, a ian HC1 and A ellano
app oach (Co ell, 2003). The mul icollinea i y
was con olled by he Va iance In l a ion Fac o s
(VIF) me hod (Adkins, 2012). VIF alues lowe
han he bo de alue 10 ( he highes alue 2.2
was de i ed o he old dependency a io a iable)
indica ed he e was no collinea i y p oblem in
ou se o he explana o y a iables. We used
he s anda d signi i cance le els and he es o
no mali y o esiduals based on he Ja que –
Be a p ocedu e. The i nal models, p esen ed in
he Tab. 3 in he nex sec ion, passed all he es s.
single axpaye model amily axpaye model
i xed e ec s andom e ec s i xed e ec s andom e ec s
cons . 0.4824
(1.459)
0.3530
(1.259)
0.1670
(1.400)
0.1276
(1.253)
homeowne ax elie 0.04626
(0.09976)
0.05605
(0.1034)
0.08523
(0.06702)
0.08770
(0.06768)
GDP pe capi a 2.293e-05**
(3.800e-06)
2.206e-05**
(4.187e-06)
2.266e-05**
(3.330e-06)
2.199e-05**
(3.548e-06)
mo gage in e es paymen -0.003623
(0.1509)
-0.006298
(0.1537)
-0.02391
(0.1459)
-0.02160
(0.1463)
house p ices -0.003877**
(0.001668)
-0.003618**
(0.001513)
-0.003686**
(0.001678)
-0.003536**
(0.001561)
home- en wedge -0.005812**
(0.002663)
-0.005692**
(0.002602)
-0.005374*
(0.002694)
-0.005288**
(0.002664)
young dependency a io 0.002328
(0.06857)
-0.004443
(0.05684)
0.009766
(0.06408)
0.01263
(0.05544)
old dependency a io -0.09192
(0.05537)
-0.09518*
(0.04900)
-0.09762*
(0.04768)
-0.1000**
(0.04374)
b eak-poin dummy -0.02961
(0.09297)
-0.03281
(0.09480)
-0.01960
(0.08472)
-0.02110
(0.08553)
n140 140 140 140
Adjus ed R-squa ed 0.7205 NA 0.7323 NA
Sou ce: own
No es: S anda d e o s in pa en heses. * indica es signi i cance a he 10% le el, ** indica es signi i cance a he 5%
le el, based on obus (HAC) s anda d e o s. R-squa ed is a measu e o he p opo ion o he a iance in he household
mo gage le e age ha is p edic able om ou se o he independen a iables. NA means ha his measu e is sui able
only o linea models bu no o andom e ec s models (Adkins, 2012). Wald es esul s o join signi i cance o ime
dummies: chi-squa e(9) = 505.436 o he single axpaye model, 327.4 o he amily axpaye model wi h bo h p- alues
nea ze o. Acco ding o hese esul s we decided o include ime dummies in o he models, ne e heless eg ession
coe i cien s a e supp essed. Hausman es wi h H0 ha GLS es ima es a e consis en : Chi-squa e(17) = 74.0062 wi h
p- alue = 4.35784e-009 o he single axpaye model, 35.8286 wi h p- alue = 0.00483423 o he amily axpaye model.
Robus ness o models was es ed by he F es o he omission o a iables wi h H0: pa ame e s a e ze o o he a ia-
bles. Tes s deli e ed F es s a is ics wi h p- alues nea ze o o all models. Resul s o he diagnos ic esidual es s a e
p esen ed only o ou i xed e ec s models. Tes o no mali y o esidual: chi-squa e(2) = 6.84189 o he single axpaye
model, 5.57706 o he amily axpaye model wi h bo h p- alues > 0.01.
Tab. 3: Reg ession models o he household mo gage le e age
EM_1_2018.indd 60EM_1_2018.indd 60 21.3.2018 12:01:3621.3.2018 12:01:36
67
1, XXI, 2018
Economics
Abs ac
DOES THE TAX RELIEF FOR HOMEOWNERSHIP HAVE EFFECT ON
HOUSEHOLD MORTGAGE LEVERAGE?
Ba bo a Slin áko á, S anisla Klaza
This a icle p esen s esul s o he analysis o he ela ionship be ween he ax elie o he
homeowne ship and he household mo gage deb . The ad an ageous ea men o housing is
p o ided especially by a pe sonal income ax i owne -occupie s do no epo impu ed en s
as income bu can deduc mo gage in e es cos s. This p e e ed ax s a us is jus i i ed by he
exis ence o posi i e ex e nali ies and a desi e o enhance housing oppo uni ies a ailable o
ci izens. Howe e , e idence ha he housing policy ia he axa ion achie es i s objec i es is s ill
weak. Mo eo e he ax p o isions o he homeowne ship bene i a he highe -income households.
Fu he mo e he e a e indica ions ha he housing axa ion encou ages le e ed p ope y pu chases
and hus con ibu es o he household deb g ow h. Since he household indeb edness can ha e
ad e se e ec s on households and mac oeconomic pe o mance we ocused on he issue
whe he he income ax elie o homeowne s ha i nance hei dwellings ia a mo gage does
a ec he household le e age. We cons uc ed he a iable cap u ing especially he mo gage
in e es paymen s deduc ibili y. We employed he mul iple eg ession and da a o he o me 15
EU membe coun ies (excep G eece) o he pe iod 2004-2013. We es ima ed wo models o
wo ep esen a i e axpaye s who a y in a amily s a us using he panel da a analysis wi h i xed
e ec s. F om ou esul s we in e ed ha he income ax elie o he homeowne ship migh no
ha e in l uenced he mo gage le e age signi i can ly in he selec ed Eu opean coun ies in he
gi en pe iod. The mo gage deb was a ec ed a he by he economic le el, p ice o own housing,
mo gage in e es paymen s o demog aphic s uc u e.
Key Wo ds: Household indeb edness, housing axa ion, mo gage in e es deduc ibili y.
JEL Classi i ca ion: G21, H24.
DOI: 10.15240/ ul/001/2018-1-004
EM_1_2018.indd 67EM_1_2018.indd 67 21.3.2018 12:01:3721.3.2018 12:01:37