Jou nal o Geog aphic In o ma ion Sys em, 2015, 7, 29-42
Published Online Feb ua y 2015 in SciRes. h p://www.sci p.o g/jou nal/jgis
h p://dx.doi.o g/10.4236/jgis.2015.71003
How o ci e his pape : Leh onen, O., Wuo i, O. and Muilu, T. (2015) Compa ing he Ex end o he Sp ead E ec s: Ru al-
U ban Commu ing in Finnish Wo king Regions. Jou nal o Geog aphic In o ma ion Sys em, 7, 29-42.
h p://dx.doi.o g/10.4236/jgis.2015.71003
Compa ing he Ex end o he Sp ead E ec s:
Ru al-U ban Commu ing in Finnish Wo king
Regions
Olli Leh onen, Olli Wuo i, Toi o Muilu
Ru al Policy and Resea ch, Na u al Resou ces Ins i u e Finland, Helsinki, Finland
Email: olli.leh onen@luke. i
Recei ed 6 Janua y 2015; accep ed 29 Janua y 2015; published 2 Feb ua y 2015
Copy igh © 2015 by au ho s and Scien i ic Resea ch Publishing Inc.
This wo k is licensed unde he C ea i e Commons A ibu ion In e na ional License (CC BY).
h p://c ea i ecommons.o g/licenses/by/4.0/
Abs ac
I has long been posi ed ha u al a eas can bene i om nea by u ban agglome a ions h ough
commu ing, and ha hei de elopmen could be a side p oduc o u ban g ow h, bu he esul s
om he spa sely popula ed egions ha e no been p omising. In his s udy, we in es iga e he
possibili ies o u al-u ban commu ing o suppo u al de elopmen , and explo e he spa ial
ex- en o he wo king egions i jobs a e concen a ed in he egional cen e as indica ed by e-
cen ends. Sp ead e ec s a e simply ope a ed by an economic possibili y o u al-u ban com-
mu ing de e mined by a sum o housing and commu ing cos s om disposable income. The esul s
show ha he popula ion g ow h in ci y egions does no ex end o dis an u al a eas. I , ins ead,
leads o popula ion losses in emo e a eas due o backwash e ec s as he low disposable incomes
encou age especially low-income households o mig a e om emo e loca ions close o he cen-
e. The sp ead e ec s seem o wo k only in he limi ed u al a eas loca ed nex o he u ban co e
which encou ages he suppo o emo e u al a eas h ough place-based policy.
Keywo ds
U ban-Ru al Commu ing, Po en ial Sp ead E ec , Housing and Commu ing Cos s, Finland
1. In oduc ion
I has long been posi ed ha one u al de elopmen s a egy may be u ban de elopmen as u al a eas can bene-
i om nea by u ban agglome a ions h ough commu ing [1]-[5]. Howe e , esul s om he spa sely popula ed
egions sugges ha he ex en o his s a egy migh be limi ed in spa sely-popula ed egions [6]. The ex en
(including geog aphic) o which u al communi ies can ely on nea by u ban cen es o employmen and g ow h
O. Leh onen e al.
30
a ies g ea ly depending on he u ban cen e’s size, anspo a ion ne wo k, opog aphy, clima e and dis ances
om he se o compe ing u ban cen es [7]. Howe e , p e ious s udies ha e ocused on commu e s’ cha ac e-
is ics, commu ing dis ances and ela ed ac o s [2] [7]-[9], bu empi ical s udies ocusing on he ex en o u ban
sp ead e ec s on u al a eas a e lacking.
In heo y, he expansion o he u ban s uc u e hinges on he adeo be ween land en and commu ing cos s.
Inc easing commu ing oppo uni ies equi e lowe housing cos s a he u ban inge o o se he highe com-
mu ing cos o he u ban co e. Compensa ion o he inc easing commu ing cos s is impo an because he house-
hold selec ions be ween commu ing and mig a ion om u al iew ha e e y di e en impac s on popula ion
and di e en implica ions o egional policy and planning. Fo example, a signi ican dec ease in local labou
demand may induce mobili y o o he egions. The impac s o his mobili y on he municipali y’s ax base and
housing ma ke depend on he indi idual’s possibili ies and choice be ween mig a ion and commu ing. The li e-
a u e has e ealed ha he likelihood o commu ing inc eases signi ican ly wi h he in e egional accessibili y
o job oppo uni ies, and is also mo e likely o people li ing in me opoli an a eas han mig a ion [8]. Also,
lowe u al o exu ban housing p ices ha e a ac ed new esiden s om he u ban cen e, who e ain hei u ban
jobs and commu e o wo k in he cen e [10] [11].
The con ibu ion o his pape is he use o an u ban hie a chy con ex o disen angle he local job g ow h
e sus popula ion g ow h e ec s on u al ou -commu ing in Finland and o conside based on ou indings u al
de elopmen policy implica ions. We in es iga e he possibili ies o u al-u ban commu ing o suppo u al de-
elopmen based on housing and commu ing cos s, and explo e he spa ial ex en o he wo king egions i jobs
a e concen a ed in he egional cen e as in ecen ends. In ou de ini ion, he spa ial ex en o he wo king e-
gion de ines he egion whe e po en ial sp ead e ec s o posi i e spillo e s om u ban g ow h can be di ec ed o
u al a eas. Po en ial sp ead e ec s a e simply ope a ionalized by an economic possibili y o u al-u ban com-
mu ing de e mined by a sum o housing and commu ing cos s om disposable income. The spa ially une en
employmen g ow h in Finland, as in he shi o knowledge-based economy in he 1990s, is likely o p omo e
s ong mig a o y and commu ing esponses, as ela i ely ad an aged wo ke s seek ou employmen oppo uni-
ies. I also unde lines he impo ance o commu ing oppo uni ies o u al a eas. Based on he obse ed de el-
opmen ends i has long been posi ed ha one u al de elopmen s a egy may be u ban de elopmen as u al
a eas can bene i om nea by u ban agglome a ions h ough commu ing [1]-[5].
In he ollowing chap e s, we p esen he heo e ical backg ound o he s udy. Nex we p esen a desc ip ion o
s udy egions and u ban- u al classi ica ion. Then we p esen he esul s o he analysis. In he ending sec ion,
we discuss he esul s o he s udy and poin a hei heo e ical and p ac ical implica ions.
2. Theo e ical Backg ounds
A p esen , mos local labou ma ke s in Finland a e expe iencing a down u n in he supply o compe en labou
as well as a ising unemploymen le el [12]. Du ing he pas yea s, only knowledge-based economy cen es ha e
had compe i i e ad an ages in e ms o he a ailabili y o labou , economies o scale and indus ial s uc u e,
and, he e o e, mos new jobs a e loca ed in hese cen es [13]. As employmen oppo uni ies in u al a eas a e
declining, inc eased in e - egional commu ing could be i al o imp o ing he job-ma ching p ocess and helping
o c ea e mo e compe i i e and la ge egions. Fo ins ance, in disad an aged a eas o No he n I eland, e-
sea che s ound signi ican pa icipa ion by u al dwelle s in he nea by u ban labou ma ke , and concluded ha
acili a ing access o own should be a clea policy p io i y o u al de elopmen [1]. Wi hou access o su i-
cien quali y oads, go e nmen s and esiden s o u al a eas will ei he need o de elop hei local economies, o
o look owa ds la ge u ban cen es in o de o su i e. Clea ly, many emo e u al a eas ha e insu icien c i i-
cal mass o sus ain hemsel es, indica ing ha access o commu ing may be he only iable ou come o a sus-
ainable li elihood [2]. Ano he impo an issue is ha possibili ies o commu e long dis ances om spa sely
popula ed a eas a e mo e limi ed o easons o dis ance and access o public anspo a ion compa ed wi h me -
opoli an and o he mo e densely popula ed a eas [14].
A common explana ion o ecen u ban- u al popula ion de elopmen and loca ion o new indus ies is
based on agglome a ion economies and knowledge spillo e s in u ban cen es [15]. The agglome a ion econo-
mies, u ban ameni ies and economic s uc u e ha a e ound in la ge ci ies p omo e local economic success.
Economies o scale and scope in economic ac i i y and in he p o ision o ameni ies gene a e u ban ad an ages
ha pe pe ua e he concen a ion o economic ac i i y and popula ion in, and in close p oximi y o, hem [4].
The g ow h o he ci y will bene i b oade egions wi h access o hese ad an ages. Thus, he “engine o g ow h”
O. Leh onen e al.
31
s a emen is linked o he changing ole and cos o dis ance in he in o ma ion economy, and ela ed issues such
as cus ome accessibili y and access o highe -o de p oduce se ices [4] [16] [17]. The cumula i e concen a-
ion o jobs has inc eased he hopes o u al a eas o bene i om he u ban agglome a ions. Howe e , in dis an
u al a eas he ocus is on agglome a ion economics independen economic ac i i ies o al e na i ely on declin-
ing he ic ion o dis ance. In Finland, only he na u al esou ce-based g ow h and ou ism ha e had simila
p obabili ies o he inc ease o jobs in he cen e and u al a eas, bu hese new jobs ha e no been able o co e
he job losses in ag icul u e, o es y and indus y [13].
F om a u al de elopmen pe spec i e, he in e es in commu ing is ypically ela ed o he ex en o which
u al households can bene i om u ban agglome a ion economies h ough access o employmen oppo uni ies,
while main aining a u al esidence [7]. This opic is discussed in neo-classical spa ial-economic u ban models,
whe e he si ua ion o households’ esiden ial loca ion is a esul o he ade-o be ween commu ing cos s and
p e e ed housing cos s [18]-[20]. A cen al assump ion o hese adi ional models is ha wages, housing p ices
and popula ion densi ies all ou side a cen al business dis ic . Acco dingly, commu ing is de e mined by he
equilib ium s a e be ween housing and labou ma ke , whe e he mone a y and men al cos s o commu ing a e
compensa ed h ough he housing ma ke . Households may, he e o e, choose o loca e u he om wo k and
ha e g ea e commu ing cos s in exchange o lowe housing cos s. Se e al s udies ha ha e ocused on he e-
la ionship be ween commu ing and esiden ial loca ion show ha longe commu es a e also chosen as pa o
housing p e e ences. Fo example, a s udy on commu ing decisions abou he dual-ea ne households shows ha
spouses join ly chose longe commu ing dis ances as pa o p e e ed housing and neighbou hood cha ac e is-
ics [21]. In Finland, simila indings ha e also been obse ed [22]. Fu he mo e, lowe u al o exu ban housing
p ices a ac new esiden s om he u ban cen e, who e ain hei u ban jobs and commu e [10] [11].
Ex ended, o u al- o-u ban, commu ing is o en ci ed as an example o he spillo e s o u ban g ow h [7].
Sp ead and backwash ha e been used espec i ely o desc ibe he posi i e and nega i e e ec s o u ban g ow h
on he u al a eas [23]-[25]. In his iew, u al- o-u ban commu ing decisions a e closely ela ed o local em-
ploymen p ospec s, access o u ban employmen and he o he loca ion a ibu es o bo h u al and accessible
u ban a eas. O cou se, he esiden ial loca ion decision is also closely ied o he commu ing- o-wo k possibili-
ies [8]. Posi i e spillo e s (sp ead) occu when u al popula ion and employmen inc ease as a esul o com-
mu ing, popula ion mig a ion and i ms and households leeing u ban conges ion and high cos s. Re e ence [26]
ound ha in he U.S. s a e o No h Ca olina, he mo emen o u ban popula ion o u al a eas (deconcen a ion)
inc eased he geog aphic ex en o u ban commu ing sheds. E idence om Canada sugges s ha a la ge sha e o
u al a eas ha expe ience job and popula ion g ow h a e nea u ban cen es (say wi hin 120 kms), especially
la ge cen es, and ha commu ing om u al places-o - esidence o u ban places-o -wo k can “decouple” local
job and popula ion g ow h [27]. Thus, local popula ion g ow h may no longe be so dependen on local job
g ow h, bu ins ead on job g ow h in u ban places wi hin commu ing dis ance, leading o geog aphically la ge
egional labou ma ke s [5] [11] [28]. Fo u al a eas nea u ban cen es, he means o popula ion g ow h may be
hei bes u al de elopmen s a egy.
3. Wo king Regions and De ini ion o Ru al A eas in Finland
Finland is an in e es ing case o he s udy a hand. I is a classic example o a wes e n coun y which has s ong
p esump ions o he sp ead e ec s c ea ed by egional cen es suppo ed by egional policy. The Na ional Inno-
a ion Sys em was based on es ablishing he Cen es o Excellence in a ew, la ge u ban a eas and a ou ing
expe imen al esea ch and de elopmen in companies. The egional policy pa adigm supposes ha a ious e-
gional cen es sp ead success in hei su ounding a eas. This should ake place di ec ly o ia knowledge spil-
lo e s o companies, hus inally imp o ing household incomes and a he end quali y o li e. Since 1994, eigh
u ban-based cen es o expe ise ha e been c ea ed in Finland o he incuba o s o egional de elopmen , bu
he majo i y o unding was alloca ed o he la ges u banised nodal a eas o Helsinki, Tu ku, Tampe e and Oulu.
The policy o he cen es o expe ise expanded o 14 cen es in 1998 and o 22 in 2003 as he aim o he policy
was o co e geog aphically almos he en i e coun y aiming a a oiding he o ma ion o idle human esou ces.
Howe e , he analysis o he policy has shown ha in ecen yea s emphasis has been ins ead o e i o ial cohe-
sion in he compe i i eness o egional cen es and inno a ion policy [29].
The a eal uni o ou analysis is pos code a ea, which cons i u es he smalles egional s a is ical en i y based
on unc ionali y in he dis ibu ion o a se ice, he e he pos al se ice o Finland. The mos commonly used
geog aphical uni in Finnish egional analyses is he municipali y, bu because each municipali y is a he he e-
O. Leh onen e al.
32
ogeneous in e ms o housing cos s , i is jus i ied o use smalle and mo e homogeneous a eal uni s in o de o
obse e he spa ial pa e ns o li ing cos . The da a on he a e ages o household disposable incomes, commu -
ing dis ances and housing sizes we e de i ed om he SuomiCD da abase consis ing o 2669 pos code a eas.
The pos code a ea was he smalles spa ial uni which con ained in o ma ion abou housing p ices and had
enough la ge housing ma ke s, and so also included u al a eas. The g ea e p oblem wi h smalle spa ial uni s
han pos code a eas, like s a is ical g id cells, is ha due o small housing ma ke s, he co e age o he da a
would be spa se and p obably highly andom in u al a eas. Housing p ices in 2010 and 2011, which a e used in
he calcula ion o housing cos , we e based on he ma ke p ices om he housing ma ke s.
3.1. Wo king Regions
In his pape , we will ocus on h ee wo king egions—Helsinki, Tu ku and Joensuu—which all ep esen he
di e en cen es o expe ise and ci y sizes (Figu e 1). Wo king egions a e based on he de ini ion o S a is ics
Finland o he yea 2013 and hey a e modelled sepa a ely because in e ms o housing cos s, he e has been
Figu e 1. Wo king egions and hei loca ion in Finland.
O. Leh onen e al.
33
ound g ea a ia ion wi hin wo king egions in Finland [30]. The analysis ocus on wo king egions as i allows
deepe unde s anding abou he po en ial sp ead e ec s in mic o-scale. The selec ion c i e ion o egions was e-
la ed o he assump ion ha he size o he cen e will ha e an impac on he ex en o sp ead e ec s. P e iously,
e e ence [7] ound ha he la ge u ban cen es ha e la ge ‘ oo p in s’ in u al space and ha all u ban places
see hei a ac ion as commu ing des ina ions dissipa e wi h dis ance. S a is ics abou he used wo king egions
a e collec ed in Table 1. Also he di icul ies acqui ing a decen da abase o analysis o ced us o concen a e
on hese selec ed wo king egions.
3.2. De ini ion o Ru al A eas in Wo king Regions
De ining u al a eas has o a long ime been a subjec o s udy and no common de ini ion sui able o all coun-
ies and di e en e i o ial con ex s exis [31]. In his pape , we de ine u al a eas as being ou side o u ban
a eas which include he cen al u ban co e, subu bs and exu ban inge. This means ha u al a eas a e de ined
as being ou side o he a ea ha mos likely bene i om u ban sp awl, o , in o he wo ds, om whe e people
cu en ly commu e o u ban a eas. Pos code a ea ypology is done wi h clus e analysis being commonly em-
ployed in geog aphy o educe a la ge numbe o a eas o clus e s ha a e in e nally as homogeneous as possible
[32]. In ou case, he clus e analysis was based on wo a iables: dis ance o he egional cen e and a e age
dis ance o jobs (Table 2). These a iables we e used as hey i he abo e classi ica ion o pos code a eas. In
heo y, he cen e and subu bs should be a eas whose dis ance o he cen e and wo kplace is ela i ely small.
The exu ban inge should consis o a eas ha loca e u he away om subu bs and whose ips o wo k a e
a he long due o g ea e commu ing dis ance om he cen e. The u al a eas a e desc ibed as being loca ed a
om he cen e and ha ing low commu ing dis ance o he cen e because o he declining equency o u-
al-u ban commu ing.
Fo he pu pose o clus e analysis, all he alues o he a iables ha a e included ha e o be s anda dized o
z-sco es wi h a mean o 0 and a s anda d de ia ion o 1. O he wise, he dis ance o he egional cen e wi h he
la ges alues would domina e he clus e ing solu ion. Since he me hod employed he e does no make any dis-
ibu ional modi ica ions, no o he ans o ma ions o he da a we e pe o med. To o de he pos code a eas in o
he abo e classi ica ion, a K-means clus e analysis was pe o med. K-means clus e ing aims o pa i ion he
Table 1. S a is ics abou he wo king egions.
Region Popula ion
2011 Popula ion change
2000-2011 (%) A e age squa e
p ice 2010 (€) A e age income o
household 2011 (€) Lowe qua ile o
a e age incomes 2011 (€)
Longes dis ance
om cen e
Helsinki (n = 201) 1,290,970 11.7 2801 43,098 32,869 83.5
Tu ku (n = 46) 288,157 7.1 1645 34,891 26,720 62.8
Joensuu (n = 21) 90,213 1.8 1428 28,977 24,597 74.3
Table 2. Desc ip i e s a is ics o o iginal a iables by clus e s.
Region Va iable Cen e Subu bs Exu ban inge Ru al
Helsinki
A e age dis ance o egional cen e 8.87 8.95 34.52 66.76
A e age dis ance o wo k 6.91 5.19 10.17 7.69
n 77 60 45 19
Tu ku
A e age dis ance o egional cen e 7.88 8.45 32.05 53.36
A e age dis ance o wo k 6.20 8.37 8.96 7.26
n 19 14 8 5
Joensuu
A e age dis ance o egional cen e 4.66 6.55 28.55 47.12
A e age dis ance o wo k 7.20 6.37 8.85 8.92
n 6 7 4 4

O. Leh onen e al.
34
poin s in o k g oups such ha he sum o squa es om poin s o he assigned clus e cen es is minimized. He e,
he k was se o 4 based on ypology discussed abo e. Clus e ing was based on Ha igan-Wong algo i hm wi h
20 andom s a s and 50 i e a ions o ind he bes s a o clus e ing [33], and s abili y o clus e ing. The em-
pi ically g ounded ou -clus e solu ion consis s o clus e s named cen e, subu bs, exu ban inge and u al
a eas (Table 2).
The a eal classi ica ion gi es a mo e s uc u al iew o he cu en egional di e en ia ion han single s a is-
ics om he pos code a eas. The cen e a eas a e he mos populous clus e whe e he housing p ices a e also
high (Table 3). A leas pa ly due o conges ion, he popula ion g ow h has been lowe compa ed wi h subu bs
o exu ban inge (Table 2). Howe e , he popula ion has s ill g own in Helsinki and Tu ku egions (Table 3).
Compa ed wi h he subu bs and inge a eas he popula ion in he cen e is also olde and he a e age size o he
households is smalle . Due o age s uc u e, he disposable incomes a e also lowe han in subu bs and inge
a eas. The clus e o subu bs is he only one whe e he o al popula ion and numbe o jobs ha e inc eased in
e e y wo king egion (Table 3). Young people wi h amilies ha e mo ed o he subu bs, as he low a e age age
indica es. These a eas a e loca ed in he a el- o-wo k a ea o cen es, bu ou side he co e o he ci y i sel as
a e age dis ances o wo k and he egional cen e a e demons a ing (Table 2).
Pos code a eas in exu ban inge a eas a e loca ed in he inge o he a el- o-wo k a eas. These pos code
a eas a e in Helsinki and Tu ku egions cha ac e ised by he highes disposable income o households, la ges
amily and house sizes (Table 3). The high income le el indica es be e economic esou ces o cope wi h
commu ing cos s. The size o he cen e municipali y seems o ha e an e ec on popula ion g ow h as in Helsin-
ki and Tu ku he g ow h has been he highes in exu ban inge, while in Joensuu he popula ion g ow h has
Table 3. Desc ip i e backg ound o he clus e s.
Region Va iable Cen e Subu bs Exu ban inge Ru al
Helsinki
A e age popula ion o pos code a eas 7213.89 6933.98 5675.26 4434.89
Popula ion change (2008-2010) 0.52 1.74 1.35 0.21
A e age age 39.71 36.56 36.48 42.05
Family size 2.09 2.17 2.44 2.05
Mean disposable income o households (€) 41523.92 44790.90 47490.84 33725.11
House p ice (€/m2) 3214.24 3374.26 2214.57 1690.78
A e age house size (m2) 73.89 76.45 92.4 80.00
Tu ku
A e age popula ion o pos code a eas 6609.89 8511.85 3840.00 3130.20
Popula ion change (2008-2010) 0.25 1.26 1.32 −2.79
A e age age 41.63 38.64 40.00 42.80
Family size 2.03 2.07 2.32 2.16
Mean disposable income o households (€) 33132.95 35444.86 40669.50 30779.20
House p ice (€/m2) 1655.36 2066.43 1454.00 997.20
A e age house size (m2) 78.63 79.64 97.75 92.20
Joensuu
A e age popula ion o pos code a eas 4966.33 5191.14 2191.75 3999.50
Popula ion change (2008-2010) −2.05 2.51 −2.22 −2.83
A e age age 33.33 37.42 43.00 46.75
Family size 2.10 2.10 2.17 1.97
Mean disposable income o households (€) 30881.17 30941.71 27625.25 24036.50
House p ice (€/m2) 1739.16 1981.14 991.75 926.50
A e age house size (m2) 78.50 78.85 82.00 77.50
O. Leh onen e al.
35
been nega i e in his clus e . This seems o indica e ha he sp ead has no ex ended o he exu ban inge in
Joensuu wo king egion. In Helsinki and Tu ku egions he exu ban inge seems o be g owing and clea ly be-
ne i ing om he sp ead e ec s gene a ed by cen es.
In e ms o ypology, he u al a eas a e loca ed a long dis ance om he cen e o he wo king egion, and
ha e been se e ely es uc u ed by he ou -mig a ion and ageing o he popula ion (Table 3). The popula ion
change has been mos nega i e o smalles (Helsinki egion), and on a e age he popula ion is oldes compa ed
o o he clus e s. The economic and socie al s uc u es a e de e io a ing, and hey ha e also aced subs an ial e-
s uc u ing, which will p obably lead o decline in he u u e. Pos code a eas belonging o his clus e seem o be
ma ginal a eas om he iew poin o commu ing o he cen e, and he ic ion o dis ance seems o ha e he
highes impac on de elopmen .
4. Empi ical Implemen a ion
4.1. Calcula ion o he Housing and Commu ing Cos s
In ea lie s udies, he po en ial sp ead e ec s a e mos ly ope a ionalized by he g ow h o popula ion o jobs [7]
[23]-[26]. In his pape , we c ea ed a new p ac ical way o ope a ionaliza ion he po en ial sp ead e ec s based
on he sum o housing and commu ing cos , which canno be mo e han 80 pe cen o he household incomes as
households a e spending 20 pe cen o he incomes o basic li ing including ood, heal h and communica ion
[34]. Howe e , mos p obably po en ial sp ead e ec s weaken ea lie as he household canno long e m use 80
pe cen om i s incomes o housing and commu ing because o he wise i has no le enough esou ces o ba-
sic li ing. The e o e, he po en ial sp ead e ec s weaken, o ins ance, i he housing and commu ing p opo ion
is o e 60 pe cen . Acco ding o s a is ics, abou 43 pe cen om incomes a e spen on housing and commu ing
[34]. The p opo ion o housing and commu ing cos s om household income
( )
i
D
is calcula ed by sub ac -
ing s a e axable incomes
( )
i
I
, housing cos s
( )
i
L
, commu ing cos s
( )
i
C
and municipali y axes
( )
i
T
, and
di iding his by incomes mul iplied by 100. By equa ion, his can be w i en as
100
n
niii i
i
i
ILC T
DI
−− −
= ×
(1)
whe e
n
i
C
deno es a e age commu ing cos calcula ed based on a e age commu ing leng h o pos code a ea.
Respec i ely, he disposable income o a e age household in pos code a ea in si ua ion when loca ion o jobs is
changed o highes job concen a ion pos code a ea is calcula ed by subs i u ing
n
i
C
wi h
k
i
C
. By compa ing
hese al e na i e commu ing cos s, we can es ima e he po en ial sp ead e ec s wi h u al-u ban commu ing op-
po uni ies. This es ima e will also indi ec ly desc ibe he impac ha concen a ion o jobs migh ha e on
household’s disposable income.
Housing cos s
( )
i
L
a e calcula ed as he sum o capi al expendi u e
( )
i
P
and ea men cos s
()
i
K
[35]
ii i
L PK= +
(2)
Capi al expendi u e consis s o mo gage epaymen s as well as mo gage in e es a es. Ca e expenses a e in-
cluded in he calcula ion o ene gy and o he ene gy cos s, p ope y axes and wa e and was e cha ges. The la -
e cos s a e di icul o es ima e eliably by pos al code o by municipali y, so we use he gene al housing cos s
de ined by Social Insu ance Ins i u ion o Finland based on he calcula ion o he housing allowance [36]. The
ca e expenses o he esidence a e calcula ed pe m2, which means ha di e ences in he a e age sizes o houses
ha e impac on hese expenses. The use o gene al cos s is jus i ied by he ac ha hei sha e o he o al cos o
housing is, in spi e o municipali y-speci ic a ia ions, small. Real es a e axes a e calcula ed by municipali ies
om alid axes o 2011. Calcula ion uses he s a is ics o a e age building lo s and si e cos om he na ional
land su ey o Finland. Calcula ion o he housing cos is based on owne -occupied housing, which was, in he
yea s 2010 and 2011, excep ionally cheap in Finland compa ed wi h en ed housing [30].
Acco ding o he Equa ion (1), he li ing cos will also depend on he cos o commu ing o wo k. These cos s
a e calcula ed in wo al e na i es, one o which desc ibes he cu en si ua ion wi h a e age commu ing dis-
ances and he second which desc ibes he si ua ion when he household’s jobs a e loca ed in he pos code a ea
wi h he highes job concen a ion in he wo king egion. Thus, he calcula ion eminds a heo e ical model o
he cen al business dis ic (CBD) whe e jobs end o concen a e on ce ain a eas. Commu ing cos s a e calcu-
O. Leh onen e al.
36
la ed as a p oduc o cu en a e age dis ance o job
( )
n
i
d
and uni cos o dis ance
( )
d
c
sub ac ed by ax
deduc ion om a elling expenses
( )
i
m
n nd
ii i
C dcm=×−
(3)
The a e age commu ing dis ance by pos code a eas was ga he ed om he SuomiCD da abase. The uni cos
o he dis ance was based on he es ima e o he Au omobile and Tou ing Club o Finland as ega ds he cos pe
kilome e, which in 2010 amoun ed o 0.42 €/km. Commu ing cos s a e calcula ed by ca because u al a eas, in
pa icula , o en lack egula public anspo , so he use o a ca is he only op ion, and, he e o e, p i a e ca s
a e he mos commonly used o a elling. The p i a e ca is also he mos common means o anspo a ion in
Finland, as he days o a el dis ance by ca is 72 pe cen [37]. This is also he eason why ax deduc ion is
calcula ed based on he use o a p i a e ca o commu ing. In 2010, compensa ion o he use o a ca was 0.25
€/km. Howe e , ax deduc ion based on he ax a e is limi ed so ha compensa ion is paid when he a el ex-
penses exceed he 600 €/yea , and he uppe limi o compensa ion is se a 7000 € yea .
Simila ly, he commu ing cos o he cen e was calcula ed by mul iplying he dis ance be ween he pos code
a ea and he cen e
( )
k
i
d
and uni cos
( )
d
c
, and hen sub ac ing ax deduc ion om a elling expences
( )
i
m
k kd
ii i
C dcm
=×−
(4)
The dis ance o he cen e was calcula ed along he oad ne wo k based on he Digi oad da abase wi h A c-
Map 10.1. The calcula ion o he model was c ea ed wi h s a is ical p og am R. The esul s o he calcula ions
a e app oxima e because due o da a sou ces he analysis ocuses on he a e age household´s li ing pa ame e s
as ega ds he house size, house p ices, household disposable incomes and commu ing dis ances. Howe e , he
me hodology s ill p o ides es ima es o he in e - egional di e ences in housing and commu ing cos s, as well
as hei impac on he household’s disposable income, and, ul ima ely, conc e ely in o ma ion abou he po en ial
sp ead e ec s o he cen e om he u al iew.
4.2. Es ima ion o he Housing and Commu ing Cos s wi h Non-Pa ame ic Local
Polynomial Reg ession Model
The sum o he housing and commu ing cos s in wo king egions is analyzed using he local polynomial models.
These models a e used o es ima e he cu es ha desc ibe he dependence o housing and commu ing cos s e-
spec o dis ance om he cen e. By compa ing he es ima ed cu es o hese models, we can es i s and
second hypo heses. The used me hod belongs o he amily o nonpa ame ic me hods, whose objec i e is o e-
lax assump ions on he o m o a eg ession unc ion and o allow da a sea ch o a sui able unc ion ha de-
sc ibes he da a well. Re e ence [38] a gued ha his explana o y cha ac e is he bes bene i o nonpa ame ic
modelling as i does no need any p esump ions abou he unc ion, acco ding o which he model is econciled.
Compa ed wi h linea o exponen ial eg ession me hods, local polynomial models igno e he s ic assump ions
ega ding o s a is ical independence o he e o s, homoscedas ici y and no mali y o he e o dis ibu ion.
Small da a se s, as in his s udy (see Table 1), inc ease he likelihood o hese assump ions o be iola ed and i
hese assump ions a e iola ed, he eg ession esul s abou he housing and commu ing cos s may be se iously
biased o misleading.
In addi ion, he me hod's lexibili y makes i ideal o he modelling o complex phenomena, especially in
cases whe e he heo e ical model o dis ibu ion does no exis like in ou case (ibid.). The aim o he local po-
lynomial models is o app oxima e locally he dependen a iable
( )
( )
D
(housing and commu ing cos ) wi h
di e en polynomials by independen a iable
0
x
(dis ance o he cen e). The esul s a e p esen ed isually
because unlike pa ame ic models he model does no es ima e s a iona y eg ession coe icien s. Thus, he me-
hod can some imes be desc ibed as “non-s a iona y” as he es ima ed cu e can e eal locally changing associa-
ions. The me hod is also sui able o small da a se s [38].
Mo e p ecisely, he used local polynomial eg ession o
y
(housing and commu ing cos s) on
x
(dis ance
om cen e) can be w i en as
( ) ( ) ( )( ) ( )( ) ( )( )
2
01 2
p
p
D x xDx xDx xDx
ββ β β
≈ + −+ − ++ −
(5)
O. Leh onen e al.
37
weigh ing he obse a ions in ela ion o hei p oximi y o he ocal alue
0
x
. He e, we use a common weigh
unc ion named icube unc ion:
( )
( )
3
3
1 , o 1;
0, o 1.
w
−<

=
>

(6)
whe e
( )
0ii
xxh= −
. The
h
is he hal -wid h o a window enclosing he obse a ions in he local eg ession.
Typically, he
h
is adjus ed so ha each local eg ession includes a ixed p opo ion s o he da a. The e o e,
s
is called he span o he local- eg ession smoo he . The la ge he span is, he smoo he he esul , and, con-
e sely, he la ge he o de o he local eg essions
p
is, he mo e lexible he smoo h. In his weigh ing unc-
ion he obse a ions which a e u he loca ed han bandwid h
( )
h
, ge a weigh ing alue 0. This means ha
hey do no ha e an impac on local es ima ion a poin
0
x
. We used he closes neighbo hood es ima ion o
models because i was sui able o spa se da a [39]. The model was es ima ed wi h
R
and i s base s a s lib a y.
5. Resul s
5.1. Sum o Housing and Commu ing Cos s wi h Respec o Dis ance om Cen e
In he cu en si ua ion, when he commu ing o he cen e is no domina ing in he dis an pos code a eas, he
di e ences be ween he households li ing in he cen e and in he dis an a eas a e ela i ely low as he es i-
ma ed cu e o he sum o housing and commu ing cos s wi h espec o dis ance om he cen e is la (Figu e
2). Explana ion o small di e ences is ha he sho a e age commu ing dis ances in he dis an ( u al) pos code
a eas keep he commu ing cos s ela i ely low because he commu ing is s ill concen a ed in smalle local cen-
es loca ed less han 10 kilome es om hei place o esidence. The e o e, commu ing cos s ake only a small
p opo ion om cos s and disposable income is mainly de ined by housing cos s. Thus, due o sho commu ing
ips and lowe housing cos s compa ed wi h u ban a eas he disposable income in dis an a eas is ei he highe
o a leas he same le el wi h he cen es (Figu e 2). A i s lowes , housing and commu ing cos s accoun o 45 -
50 pe cen o a household’s incomes, which a e close o wha ano he Finnish s udy has obse ed in municipals
[40].
The deepe e alua ion o he es ima ed cu es shows ha wo king egions ha e some di e ences in he hous-
ing cos s. Fo ins ance, in Helsinki egion, he housing cos s a e he highes in he pos code a eas loca ed in he
e y cen e and abou 20 kilome es om he cen e. The pos code a eas loca ed wi hin hese dis ances ha e
disposable income anging om 55 - 65 pe cen o he o al incomes (Figu e 2). In Tu ku wo king egion, he
highes cos s a e loca ed 15 km om he cen e, and in Joensuu, abou 10 km om he cen e (Figu e 2), wi h
disposable income anging om abou 40 - 50 pe cen o e enue be ween households. These di e ences a e
likely o indica e socio-economic di e ences be ween housing zones wi hin he egions. In Helsinki egion, he
housing p ices a e conside ably highe in he cen e compa ed wi h o he wo king egions, which is di ec ly e-
lec ed in he household’s disposable incomes (Figu e 2). In o he egions, he highes housing cos s a e loca ed
10 - 15 kilome es om he ci y cen es, whe e he esiden ial buildings a e mos ly de ached houses. Also, he
income le el o he households has di ec impac on he disposable incomes (Figu e 3). The disposable incomes
o he households li ing in he uppe limi o he lowes qua ile a e na u ally conside ably smalle han hose o
he a e age households. Declining disposable incomes is a p oduc o he ela i e decline on incomes, whe e
housing and commu ing cos s ep esen a ela i ely la ge p opo ion o middle-income ea ne s as housing co-
ope a i es.
When he household’s jobs a e loca ed in he cen e like in CBD model, he disposable income is he highes
in he pos code a eas loca ed in he cen e o in he immedia e neighbo hood o i . Excep o he Helsinki wo k-
ing egion, he sum o he housing and commu ing cos is he lowes in he cen e. In Helsinki egion, he lowes
sum o housing and commu ing cos s can be achie ed in pos code a eas loca ed abou 10 km om he cen e
(Figu e 2 and Figu e 3). The di e ences inside he wo king egions a e no such big in he housing cos s ha
hey would de ea he inc easing cos s o commu ing. Howe e , because o he deduc ible eimbu semen o
a el expenses he commu ing cos s inc eases sha ply when he dis ance om he wo k a ea cen e exceeds 30 -
40 kilome e s (Figu e 2 and Figu e 3). A e his zone, he imagina y slopes o he locally es ima ed cu es be-
come la ge which means ha commu ing cos s ha e highe impac on disposable income and he impac o dis-
ance s eng hens. The esul shows ha ax incen i e is no planned o suppo wide sp ead e ec s.