Economic Modelling 121 (2023) 106231
A ailable online 7 Feb ua y 2023
0264-9993/© 2023 The Au ho . Published by Else ie B.V. This is an open access a icle unde he CC BY license (h p://c ea i ecommons.o g/licenses/by/4.0/).
Con en s lis s a ailable a ScienceDi ec
Economic Modelling
jou nal homepage: www.jou nals.else ie .com/economic-modelling
Can con lic s uni e a na ion?
Da yna G echyna1
Depa amen o de Teo ía e His o ia Económica, Uni e si y o G anada, Facul ad de Ciencias Económicas y Emp esa iales, Campus Uni e si a io de Ca uja, P.C
18071, G anada, Spain
ARTICLE INFO
Keywo ds:
Violen con lic
Poli ical pola iza ion
Panel da a
ABSTRACT
Despi e he global commi men o os e ing peace, he wo ld su e s om iolen con lic s. Rela ed li e a u e
connec s in as a e e hnic con lic s o pola iza ion, bu he ela ionship be ween he o he ypes o con lic s
and pola iza ion is unclea . I build a simple model showing ha con lic s ini ia ed by an ex e nal agg esso
can educe he poli ical pola iza ion o a coun y. Fu he mo e, using egional panel da a om Geo gia and
Uk aine, I assess how iolen con lic s in he o m o o eign s a e suppo ed e i o ial dispu es a e ela ed o
he egion-speci ic poli ical suppo o he winne in p esiden ial and pa liamen a y elec ions. The analysis
sugges s ha di e ences in poli ical p e e ences ac oss egions dec ease a e a con lic . Finally, I con i m
a nega i e associa ion be ween con lic s and poli ical pola iza ion using coun y-le el da a om a ound he
wo ld. A shi in poli ical p e e ences away om le -wing public policies is an impo an channel o a decline
in poli ical pola iza ion a e a iolen con lic .
1. In oduc ion
Despi e he global commi men o os e ing peace and sus ainable
de elopmen decla ed in he Uni ed Na ions 2030 Agenda and accep ed
in 2015 by all 193 membe coun ies, he wo ld su e s om iolen
con lic s. The Uppsala Con lic Da a P og am (UCDP) da ase epo s
52, 55, and 56 con lic s in 2018, 2019, and 2020, espec i ely (Ha -
bom e al.,2008;Pe e sson e al.,2021).2The Cen e o P e en i e
Ac ion (CPA)’s Global Con lic T acke epo s 27 con lic s ha a e o
pa icula impo ance o he U.S. in 2022.3
The con lic s can be ca ego ized as ex asys emic, be ween a s a e
and a non-s a e g oup ou side i s own e i o y; in e s a e; in as a e,
be ween a go e nmen and a ebel g oup; and in e na ionalized in-
as a e, be ween a go e nmen and a ebel g oup suppo ed by o eign
go e nmen s wi h oops (see Pe e sson e al. (2021)). By con lic o m,
he CPA dis inguishes among ci il wa s, e i o ial dispu es, in e s a e
con lic s, sec a ian con lic s, poli ical ins abili y, and ansna ional e -
o ism. The mos p e alen con lic s a e in as a e and in e na ionalized
in as a e in he o m o ci il wa s and e i o ial dispu es (Ha bom
E-mail add esses: [email p o ec ed],[email p o ec ed].
1I am e y g a e ul o Edi o , Gues Co-Edi o , and wo anonymous Re iewe s whose commen s helped o imp o e he manusc ip conside ably. I acknowledge
he inancial suppo o he Minis y o Science and Inno a ion o Spain, p ojec PID2019-111708GA-I00 inanced by MCIN/AEI/10.13039/501100011033. Funding
o open access cha ge: Uni e sidad de G anada / CBUA.
2UCDP da ase epo s yea s o con lic s ha ha e c ossed he 25 ba le- ela ed dea hs h eshold.
3The con lic s a e epo ed in he in e ac i e map: h ps://www.c .o g/global-con lic - acke See also h ps://www. isualcapi alis .com/mapped-whe e-a e-
he-wo lds-ongoing-con lic s- oday/.
4Fo example, he ‘‘A ab Sp ing’’, d i en by ci izens’ desi e o imp o e he poli ical and economic condi ions o hei coun ies (see Ansani and Daniele (2012)),
caused a signi ican GDP loss in he yea s ollowing he con lic s (see Eche a ía and Ga cía-En íquez (2019a,b)). Tunisia became he only coun y in he Middle
Eas and No h A ica whe e he ‘‘A ab Sp ing’’ esul ed in a eal ansi ion o democ acy (Ne e s øm,2015).
e al.,2008;Pe e sson e al.,2021). These ypes o con lic s ha e
been linked o social pola iza ion and ac ionaliza ion in he coun y
(Es eban and Ray,2008,2011a,b;Es eban and Schneide ,2008). In
pa icula , e hnic di isions ha e esul ed in e i o ial dispu es, wi h
one side o en being suppo ed by a o eign s a e. Mos o he iolen
con lic s ha e de as a ing consequences o he economy (Le e al.,
2022), bu hey can igge posi i e social and economic changes i he
esul o a con lic is he all o dic a o ship and success ul democ a ic
e o ms.4
The ques ion a ises as o whe he a iolen con lic can educe social
di isions, hus educing he p obabili y o u he con lic s. The answe
is ambiguous in gene al. On he one hand, he iolence and economic
des uc ion can u he exace ba e social ensions and pola iza ion. On
he o he hand, he incen i e o educe he sha ed high economic
cos o he con lic can b ing di e en social g oups close o he
same poli ical and economic objec i es. Gi en ha social pola iza ion
plays an impo an ole in economic g ow h and de elopmen (see,
o example, Azzimon i (2011), Azzimon i and Talbe (2014), F ye
h ps://doi.o g/10.1016/j.econmod.2023.106231
Recei ed 30 July 2021; Recei ed in e ised o m 4 Feb ua y 2023; Accep ed 4 Feb ua y 2023
Economic Modelling 121 (2023) 106231
2
D. G echyna
(2002)), a be e unde s anding o he ela ionship be ween iolen con-
lic and social di isions could acili a e policymaking and p o-g ow h
e o ms ollowing a con lic . In pa icula , i socie y’s p e e ences align
in ce ain economic and poli ical domains, pu suing goals in hose
domains could igge economic de elopmen and imp o e wel a e.
This pape e alua es he ela ionship be ween iolen con lic s and
he alignmen o poli ical p e e ences in socie y. Fi s , I de elop a
heo e ical model ha ela es a cos ly con lic o poli ical pola iza ion,
based on he s anda d poli ical economy models o op imal public
policies. In he model, he e a e wo ypes o agen s who ha e di e en
p e e ences ega ding public spending. The agen s can change hei
p e e ences a a cos , i such a change inc eases hei u ili y. A con lic
educes he expec ed u ili y o bo h ypes and makes he change in
poli ical p e e ences less cos ly, i public policy a e such a change
inc eases he p obabili y o success ul con lic e mina ion.
Second, I explo e he a ia ion in poli ical p e e ences ac oss di -
e en egions in Geo gia and Uk aine ha expe ienced simila iolen
con lic s in he las decade. I dis inguish be ween he egions con igu-
ous and non-con iguous o con lic zones and e alua e how a iolen
con lic in he o m o a e i o ial dispu e suppo ed by a o eign
s a e changes he egion-speci ic poli ical suppo o he winne in
p esiden ial and pa liamen a y elec ions. The analysis sugges s ha
he di e ences in poli ical p e e ences ac oss egions a e nega i ely
associa ed wi h con lic s.
Finally, I e alua e he ela ionship be ween iolen con lic s and he
a ia ion in poli ical p e e ences in socie y, based on he pola iza ion
o a i udes da a cons uc ed om he Wo ld Values Su ey (WVS)
combined wi h he coun y-le el iolen con lic da a om he UCDP
da ase . The analysis sugges s ha iolen con lic s a e nega i ely asso-
cia ed wi h he di e gence o indi idual poli ical o ien a ions measu ed
as le – igh posi ioning on he poli ical scale. Fu he explo a ion o he
da a sugges s ha in e s a e con lic s a e he mos signi ican co ela es
o he educ ion in pola iza ion, compa ed o he o he ypes o con lic ,
such as in as a e o in e na ionalized in as a e. The WVS da a sugges s
ha he sha e o esponden s who posi ion hemsel es as le -wing is
nega i ely associa ed wi h con lic s, in line wi h he p edic ion o he
heo e ical model ha poli ical p e e ences can shi ollowing con lic s.
A well-es ablished li e a u e on he causes o con lic poin s o
economic condi ions, such as income inequali y and po e y, as he
main d i e s o iolen con lic s (see, o example, Collie and Hoe le
(1998,2004), Miguel e al. (2004), Jus ino (2009), Weede (1981)).5
Democ acy and ins i u ions a e o he signi ican de e minan s o con-
lic s (Mousseau,2001;Jha e al.,2022). While be e ins i u ions
educe he p obabili y o iolen con lic s, he impac o democ acy
and economic de elopmen on con lic s is nonlinea , wi h he ea ly
s ages o democ a iza ion associa ed wi h highe equency o con lic s
(Heg e,2001). This s udy conside s a sample o ela i ely homogeneous
egions om wo coun ies o he o me USSR o he analysis o
he ela ionship be ween con lic s and poli ical uni y wi hin a coun y,
con olling o egion and yea ixed e ec s, hus implici ly cap u ing
o some ex en he le els o democ acy, economic condi ions, and
ins i u ions. Fo he analysis based on he coun y-le el da a a ound
he wo ld, I include he eal g oss domes ic p oduc (GDP) pe capi a,
public spending, and he index o democ acy by G ündle and K iege
(2021) as con ol a iables, in addi ion o coun y and ime ixed
e ec s.
E hnic di isions and social pola iza ion ha e been conside ed as
con ibu o y ac o s o poli ical iolence, hough he e is some e idence
ha e hnic he e ogenei y is no associa ed wi h highe le els o io-
lence wi hin na ions (see, o example, Mousseau (2001), Collie and
5Sambanis (2002) p o ides a de ailed e iew o he li e a u e on economic
causes o con lic and B ück e al. (2017) and Ray and Es eban (2017) su ey
he ecen ad ances on he analysis o iolen con lic and i s impac on he
economy.
Hoe le (1998,2004), B own and Boswell (1997), Fea on and Lai in
(2003)). Es eban and Ray (2008,2011a,b) ha e concep ualized he
heo e ical link be ween social di isions and con lic , while Es eban
e al. (2012) ha e demons a ed empi ically ha e hnic di isions in
he o m o social pola iza ion and ac ionaliza ion posi i ely in luence
he p obabili y o con lic occu ence. In his pape , social di isions
cha ac e izing a coun y be o e he con lic a e aken as gi en. I ocus
ins ead on whe he socie y can become less di ided ollowing a con lic .
To his end, his s udy aims o igge a discussion on he na u e
o social changes a e he iolen con lic s. Such a discussion could
po en ially help p o ide an unde s anding o he d i e s o economic
p og ess o eg ess ollowing economic des uc ion.
The emainde o he pape is o ganized as ollows. Sec ion 2
p o ides a simple model connec ing pola iza ion o p e e ences o
cos ly con lic s, based on he s anda d poli ical economy models o
op imal public policy. Sec ion 3desc ibes he da a and conduc s an
empi ical explo a ion o he ela ionship be ween iolen con lic and
he alignmen o poli ical p e e ences based on egional da a om
Geo gia and Uk aine. Sec ion 4p o ides he obus ness check using he
coun y-le el da a om a ound he wo ld. Sec ion 5concludes.
2. The model
Conside a wo-pe iod economy model wi h wo ypes o agen s,
𝑖and 𝑗, in equal measu e, wi h o al popula ion no malized o one.
The wo ypes o agen s ha e equal endowmen no malized o one
bu di e en p e e ences ega ding he size o he public sec o . The
ins an aneous u ili y 𝑢o an agen is addi i ely sepa able, inc easing,
and conca e in p i a e and public consump ion, wi h ype-speci ic
weigh 𝜆ℎ>0,ℎ∈ {𝑖, 𝑗}, on public consump ion.
The go e nmen collec s axes o p o ide public goods. Due o
he disag eemen be ween he wo ypes o agen s o e he size o
public sec o , gi en by he di e ence in ype-speci ic weigh s on public
consump ion, poli ical pa ies will endogenously a ise in a democ a ic
en i onmen (Azzimon i,2015). The e a e wo pa ies, 𝑖and 𝑗, ep e-
sen ing he wo ypes o agen s and compe ing o o ice e e y pe iod.
They al e na e in powe wi h p obabili y 𝑝∈ (0,1). Gi en equal sizes
o he g oups ep esen ing he wo di e en ypes o agen s, 𝑝= 0.5.
E e y pe iod, he pa y ha wins he elec ions o ms he go -
e nmen and decides on iscal policy o maximize he u ili y o i s
elec o a e. Fo simplici y, he go e nmen is pe ec ly impa ien .6Since
ax e enues a e used o inance he p o ision o public goods and
p i a e consump ion is equal o he endowmen ne o axes, public and
p i a e consump ion a e unc ions o he ax a e chosen by he pa y
in powe . Thus, he iscal policy chosen by he pa y in powe can be
summa ized by he ax a e. Gi en ha he pa y in powe maximizes
he u ili y o i s elec o a e, ha is, o i s ype, and gi en he conca i y
o he u ili y unc ion, he u ili y o he ype which is no in powe is
lowe han he u ili y o he ype which is in powe . Deno e by 𝑢ℎ𝑘 he
u ili y o ype ℎwhen poli ical pa y o ype 𝑘is in powe , ℎ, 𝑘 ∈ {𝑖, 𝑗}.
Then, 𝑢𝑖𝑖 > 𝑢𝑖𝑗 and 𝑢𝑗𝑗 > 𝑢𝑗𝑖. Mo eo e , i mus be he case ha ei he
𝑢𝑖𝑖 ≥𝑢𝑗𝑗 o 𝑢𝑗𝑗 ≥𝑢𝑖𝑖. Suppose he o me inequali y holds, wi hou loss
o gene ali y.
The agen s can decide o adjus hei ype a a cos (as in Caselli
and Coleman (2013)). Speci ically, i agen s o ype ℎdecide o change
hei ype o ¬ℎ,ℎ∈ {𝑖, 𝑗}, hei li e ime u ili y declines by ac ion
1 − 𝜙, which e lec s he u ili y loss due o he cul u al, emo ional,
o mone a y adjus men necessa y o he change o pe sonal poli ical
iden i y e lec ed in poli ical p e e ences. The agen s can make a choice
abou a possible adjus men o hei poli ical iden i y each pe iod
be o e he elec ions. An agen o ype ℎwill change o ype ¬ℎi he
expec ed li e ime u ili y o such a change is g ea e han he expec ed
u ili y wi hou such a change. I he change is wo hwhile o one agen
6I s discoun ac o is ze o.
Economic Modelling 121 (2023) 106231
3
D. G echyna
o ype ℎ, i is also wo hwhile o all agen s o ype ℎ, he e o e, a e
such a change occu s, he economy consis s only o ype ¬ℎand he
pa y ep esen ing his ype wins wi h p obabili y one.7
Conside i s he economy ha is a peace in bo h pe iods, ha is,
i does no expe ience any con lic s. In he i s pe iod, agen s o ype
𝑗decide o adjus hei p e e ences o hose o ype 𝑖, in o he wo ds,
change hei poli ical iden i y om 𝑗 o 𝑖, i he ollowing inequali y
holds:
0.5(𝑢𝑗𝑗 +𝑢𝑗𝑖) + 𝛽0.5(𝑢𝑗𝑗 +𝑢𝑗𝑖)< 𝑢𝑖𝑖(1 + 𝛽)(1 − 𝜙),(1)
whe e 𝛽∈ (0,1) is he discoun ac o and he le - and igh -hand
sides desc ibe he expec ed li e ime u ili y i he agen s ays in ype
𝑗and changes o ype 𝑖, espec i ely. I an agen emains in ype 𝑗, he
li e ime u ili y consis s o he sum o expec ed u ili ies i ype 𝑗wins he
elec ions and implemen s policy 𝑗,0.5𝑢𝑗𝑗 , in he i s pe iod, and 𝛽0.5𝑢𝑗𝑗
in he second pe iod, and i ype 𝑖wins he elec ions and implemen s
policy 𝑖,0.5𝑢𝑗𝑖, in he i s pe iod, and 𝛽0.5𝑢𝑗𝑖 in he second pe iod. I
he agen s o ype 𝑗change o ype 𝑖, hey lose a ac ion 𝜙o li e ime
u ili y, bu hei expec ed li e ime u ili y ne o his loss i gi en by
𝑢𝑖𝑖 +𝛽𝑢𝑖𝑖, because ype 𝑖wins he elec ions in bo h pe iods gi en ha
he majo i y o he popula ion becomes ype 𝑖.8
In he second pe iod, he agen s o ype 𝑗decide o change hei
poli ical iden i y o ha o ype 𝑖i
0.5(𝑢𝑗𝑗 +𝑢𝑗𝑖)< 𝑢𝑖𝑖(1 − 𝜙).(2)
In bo h cases, in he i s and in he second pe iod, he change occu s
only i
𝜙 < 1 − 0.5(𝑢𝑗𝑗 +𝑢𝑗𝑖)
𝑢𝑖𝑖
=𝜙⋆
𝑝,(3)
whe e 𝜙⋆
𝑝is he minimum alue o which he change occu s in peace-
ime. Fo he alues o cos 𝜙highe han 𝜙⋆
𝑝, he change o poli ical
iden i y does no occu . In pa icula , o 𝜙 > 𝜙⋆
𝑝, he cos o change
om 𝑗 o 𝑖is so high ha no 𝑗- ype agen changes o 𝑖- ype; and gi en
he assump ion ha 𝑢𝑖𝑖 ≥𝑢𝑗𝑗 , he minimum cos o change om 𝑖 o 𝑗
mus be smalle han 𝜙⋆
𝑝, so ha o 𝜙>𝜙⋆
𝑝no 𝑖- ype agen changes o
𝑗- ype.
Conside now he economy whe e an unexpec ed con lic , ini ia ed
by an agg esso om ou side he economy, occu s a he end o he
i s pe iod. The con lic means ha i he coun y loses, mos o i s
esou ces will be des oyed o app op ia ed by he con lic ini ia o , so
ha bo h ypes’ u ili y will d op o ze o, unless he coun y wins he
con lic . The coun y wins he con lic wi h ype-speci ic p obabili y 𝑃ℎ.
The agen s o ype 𝑗will decide o change hei poli ical iden i y i
0.5(𝑃𝑗𝑢𝑗𝑗 +𝑃𝑖𝑢𝑗𝑖)< 𝑃𝑖𝑢𝑖𝑖(1 − 𝜙).(4)
The change occu s i
𝜙 < 1 − 0.5(𝑃𝑗𝑢𝑗𝑗 +𝑃𝑖𝑢𝑗𝑖)
𝑃𝑖𝑢𝑖𝑖
=𝜙⋆
𝑐,(5)
whe e 𝜙⋆
𝑐is he minimum alue o he cos o which he change occu s
in he economy in con lic . Conside he di e ence:
𝜙⋆
𝑐−𝜙⋆
𝑝=0.5(𝑢𝑗𝑗 +𝑢𝑗𝑖)
𝑢𝑖𝑖
−0.5(𝑃𝑗𝑢𝑗𝑗 +𝑃𝑖𝑢𝑗𝑖)
𝑃𝑖𝑢𝑖𝑖
>0.5𝑢𝑗𝑗
𝑢𝑖𝑖
(1 − 𝑃𝑗∕𝑃𝑖)>0,i 𝑃𝑗< 𝑃𝑖.(6)
Tha is, in he economy in con lic he change o poli ical p e e ences
occu s o a wide ange o cos s 𝜙compa ed o peace ime. I he cos
7This is he ou come o he assump ion ha ype ℎcan change di ec ly o
ype ¬ℎ; i he agen s who change hei ype do so by ‘‘ge ing close ’’ in hei
p e e ences o ype ¬ℎ, he analysis is mo e compu a ionally cumbe some bu
he esul s a e simila , as shown nume ically a he end o his sec ion.
8The p obabili y o eelec ion is independen ac oss pe iods, 0.5, i no ype
changes occu .
𝜙is such ha 𝜙𝑝< 𝜙 < 𝜙𝑐, he peace ime economy is cha ac e ized by
no changes in agen ypes, bu an ini ia ion o a con lic ha h ea ens
he coun y-wide esou ces causes all he agen s o swi ch o he ype
ha inc eases he p obabili y o winning he con lic , educing (in his
case, elimina ing) poli ical pola iza ion.
Conside an example o a model economy wi h a pa icula unc-
ional o m imposed on 𝑢, so ha he impac o con lic on agen
decisions and poli ical pola iza ion can be compu ed quan i a i ely and
illus a ed. Suppose ha u ili y depends on p i a e consump ion, 𝑐, and
public consump ion, 𝑔, as ollows:
𝑢ℎ(𝑐, 𝑔) = 𝑐1−𝜎
1 − 𝜎+𝜆ℎ
𝑔1−𝜎
1 − 𝜎, ℎ ∈ {𝑖, 𝑖}.(7)
Poli ical pola iza ion, a measu e o he di e gence o indi idual
poli ical a i udes, is hen ansla ed in o he absolu e alue o he di -
e ence in weigh s on public consump ion, |𝜆𝑖−𝜆𝑗|; a la ge di e ence
co esponds o g ea e poli ical pola iza ion.
P i a e consump ion is he endowmen ne o axes, 𝑐= 1−𝜏, and ax
e enues a e used o inance he p o ision o public goods, so ha he
go e nmen budge cons ain is 𝑔=𝜏. The p oblem o he go e nmen
is s a ic and can be summa ized by an op imal choice o 𝜏in a gi en
pe iod. Speci ically, subs i u ing 𝑐and 𝑔as unc ions o 𝜏in o he u ili y
unc ion, he p oblem o he pa y o ype ℎwhen in powe can be
w i en as ollows:
max
𝜏
(1 − 𝜏)1−𝜎
1 − 𝜎+𝜆ℎ
𝜏1−𝜎
1 − 𝜎, ℎ ∈ {𝑖, 𝑖}.
The solu ion is he op imal ax a e:
𝜏⋆
ℎ=𝜆1∕𝜎
ℎ∕(1 + 𝜆1∕𝜎
ℎ),(8)
which is an inc easing unc ion o he weigh on public spending 𝜆ℎ.
Gi en ha 𝑐and 𝑔a e unc ions o he ax a e, he u ili y o an agen
o ype ℎcan be compu ed as a unc ion o 𝜏⋆
ℎ o he cases when
he pa y in powe is o he agen ’s ype, 𝑢ℎℎ, and o he o he ype,
𝑢ℎ¬ℎ,ℎ∈ {𝑖, 𝑗}. Fu he suppose ha he p obabili y o winning he
con lic , 𝑃ℎ, is a conca e unc ion o he public sec o size, in pa icula ,
𝑃ℎ=𝑔(𝜏ℎ)𝛾, 𝛾 ∈ (0,1).
The le panel o Fig. 1 epo s he alues o he cos 𝜙 o which
he change o ype occu s in peace ime, in ligh g ay, and he addi ional
anges o which he change occu s in imes o con lic , in da k g ay,
o di e en pola iza ion le els and he ollowing pa ame e alues:
𝜎= 0.95 and 𝛾= 0.7. In his case, 𝑢𝑖𝑖 > 𝑢𝑗𝑗 , so ha o ype 𝑖i is no
wo hwhile o change he ype o any le el o posi i e cos , bu ype
𝑗changes o ype 𝑖i he cos o change is wi hin he co esponding
shaded a ea.
The model economy conside ed abo e includes only he ex eme
e sion o poli ical iden i y changes, om ℎ o ¬ℎ. The ype adjus men
whe e he p e e ences o public spending change pa ially om 𝑗 o
𝑗+𝛥𝑗, whe e 𝛥𝑗 e lec s he deg ee o adjus men in he di ec ion o
ype 𝑖p e e ences, can be conside ed mo e ealis ic. In his case, he
change occu s, i
𝑢𝑗𝑗 +𝑢𝑗𝑖 <(𝑢𝑗′𝑗′+𝑢𝑗′𝑖)(1 − 𝜙𝑝),(9)
𝑃𝑗𝑢𝑗𝑗 +𝑃𝑖𝑢𝑗𝑖 <(𝑃𝑗′𝑢𝑗′𝑗′+𝑃𝑖𝑢𝑗′𝑖)(1 − 𝜙𝑐),(10)
o he economy in peace and con lic , espec i ely, whe e 𝑗′is agen 𝑗’s
new ype, a e a change.9The change esul s in pola iza ion declining
by 𝛥𝑗.
The igh panel o Fig. 1 depic s he alues o he cos 𝜙 o which
he change o ype occu s in peace ime, in ligh g ay, and he addi ional
anges o which he change occu s in imes o con lic , in da k g ay,
whe e 𝑗adjus s close o 𝑖, as a unc ion o he magni ude o adjus men ,
om 0 o 100 pe cen , o he ollowing se o pa ame e s: 𝛥𝜆 = 0.4,
9In he case o pa ial adjus men , he p obabili y o eelec ion o each
ype emains 0.5, because he e a e s ill wo di e en ypes in equal measu e.
Economic Modelling 121 (2023) 106231
4
D. G echyna
Fig. 1. The pe missible cos s o poli ical p e e ences adjus men in peace ime and du ing con lic s.
No e:The igu e depic s he cos s o p e e ences adjus men o which he adjus men o p e e ences occu s in peace ime, in ligh g ay, and addi ional anges o cos s o which
he adjus men occu s in imes o con lic , in da k g ay, as a unc ion o ini ial pola iza ion, in panel (a), and as a unc ion o he deg ee o poli ical p e e ences adjus men , in
panel (b); in panel (b), ini ial pola iza ion is 0.4 and pola iza ion a e he adjus men is plo ed as a dashed line.
𝜎= 0.95 and 𝛾= 0.7. The igh panel o Fig. 1 also shows he poli ical
pola iza ion le el (dashed line) as a esul o he ype adjus men o
pe missible ange o cos s. Tha is, i he adjus men om 𝑗 o 𝑖is 100%,
he esul ing pola iza ion is ze o, bu i he adjus men om 𝑗 o 𝑖is
posi i e bu close o 0%, he esul ing pola iza ion is lowe han bu
close o i s ini ial alue, 0.4.
As a inal ema k, no e ha he p obabili y o winning he con lic
𝑃ℎcan be modeled as a unc ion o o al public spending, a ac ion o
public spending de o ed o mili a y se ices and de ense, o any o he
unc ion o public policy. An in-dep h analysis o possible s uc u es o
𝑃ℎas d i e s o poli ical iden i y changes could p o ide mo e insigh
in o he dynamics o poli ical p e e ences a e a con lic a he na ional
le el, leading o es able hypo heses ha a e beyond he scope o his
pape .
The ocus o his pape is on he es able hypo hesis esul ing
om he model p esen ed abo e: poli ical pola iza ion can dec ease
ollowing a coun y-wide con lic . The da a and empi ical me hodology
used o es his hypo hesis a e p esen ed in he nex sec ions.
3. Con lic s and pola iza ion: E idence om Geo gia and Uk aine
This sec ion analyzes he ela ionship be ween iolen con lic s and
poli ical pola iza ion, measu ed as he a ia ion in he egional sha e o
o es cas o he winne in he na ional pa liamen a y and p esiden ial
elec ions. The analysis is based on egional panel da a om Geo gia
and Uk aine. The simila i y o hese coun ies in e ms o his o ical
backg ound and economic and democ a ic s age o de elopmen , as
well as in e ms o he ype and o m o iolen con lic s, allows an
explo a ion o egional di e ences based on he compa ison o he
egions ha a e non-con iguous and con iguous o a con lic zone, in
peace ime and du ing con lic s.
3.1. His o ical backg ound
Geo gia and Uk aine o med pa o he So ie Union (USSR). Since
he collapse o he So ie Union in 1991, he economic and poli ical
de elopmen o he 15 newly independen coun ies has been e y
di e se. The pos -communis ans o ma ion complica ed by a poli ical
s uggle be ween ex-communis and an i-communis ac ions caused
poli ical ins abili y and pola iza ion, which had a de imen al e ec on
economic g ow h (F ye,2002).
Economic and poli ical ins abili y impeded he democ a iza ion p o-
cess in pos -So ie coun ies (see, o example, Hale (2005), Gel’Man
(2003)). Fig. 2 shows he democ acy index by G ündle and K iege
(2021) o pos -So ie coun ies, whe e alues be ween 0.5 and 1
co espond o democ acies and alues below 0.5 co espond o au oc a-
cies. Excep o Es onia, La ia and Li huania, all pos -So ie coun ies
ha e been cha ac e ized by a highly ola ile democ acy index, wi h
only A menia, Geo gia, Moldo a and Uk aine app oaching democ a ic
egimes o e ime and he emaining coun ies apped in au oc acies.
Addi ionally, a numbe o iolen con lic s ueled by sepa a is
mo emen s and suppo ed by neighbo ing s a es o geopoli ical ea-
sons (see Sussex (2012)) u he des abilized he egion. In pa icula ,
all ou ela i ely democ a ic coun ies, A menia, Geo gia, Moldo a
and Uk aine, expe ienced iolen con lic s ei he in ol ing local sepa-
a is s g oups o he neighbo ing s a es, wi h Russia ac i ely suppo ing
he dispu es, exace ba ing poli ical and economic ins abili y. These
con lic s ha e been cha ac e ized by he CPA Global Con lic T acke
as e i o ial dispu es and by he UCDP as in e na ionalized in as a e
con lic s.
The o he pos -So ie coun ies, including Aze baijan, Russia, Tajik-
is an, and Uzbekis an, also expe ienced iolen con lic s (see UCDP
da ase , Ha bom e al. (2008), Pe e sson e al. (2021)). None he-
less, he p e alence o non-democ a ic egimes in hese coun ies im-
plies ha hei da a canno be used o analyze he ela ionship be-
ween con lic and poli ical p e e ences, because people’s ue poli ical
p e e ences a e no e ealed in au oc acies.
O he ou pos -So ie democ acies ha expe ienced iolen con-
lic s, A menia and Moldo a a e excluded due o insu icien da a.
In pa icula , A menia has been in a e i o ial dispu e wi h Aze bai-
jan (wi h escala ions in 1988–1994, 2014, 2016 and 2020), al hough
i s own e i o y has no been a ec ed. Moldo a’s e i o ial uni o
T ansnis ia has been unde mili a y occupa ion by Russia since 1990.
The pe sis ence o con lic s in Moldo a and A menia ( om he 1990s
o he p esen day) does no allow o an explo a ion o he a ia ion
in poli ical p e e ences as a esul o he con lic .
Geo gia and Uk aine ha e been cha ac e ized by he swi ches be-
ween peace ime and con lic . In Geo gia, dispu es aging in wo e-
gions o Geo gia, Abkhazia and Sou h Osse ia, be ween local sepa a is s
suppo ed by Russia and he Geo gian-majo i y popula ion, s a ed
in 1990s and in ensi ied in 2008, esul ing in he de ac o indepen-
dence o he dispu ed egions. In Uk aine, dispu es in wo egions o
Economic Modelling 121 (2023) 106231
5
D. G echyna
Fig. 2. Democ acy in pos -So ie coun ies.
No e: The igu e depic s he democ acy index by G ündle and K iege (2021) o pos -So ie coun ies; alues be ween 0.5 and 1 co espond o democ acies and alues below 0.5
co espond o au oc acies.
Uk aine, Done sk and Lugansk, be ween local sepa a is s suppo ed by
Russia and he Uk aine-majo i y popula ion, begun in 2014 ollowing
he Uk ainian Re olu ion o Digni y. The i s eigh yea s o con lic
included he Russian annexa ion o C imea (in 2014) and he wa
in Donbas (2014–p esen ), as well as na al inciden s, cybe wa a e,
and poli ical ensions. The loca ion o Geo gia and Uk aine be ween
Wes e n Eu ope on one side and Russia and Asia on he o he , combined
wi h hei So ie pas , con ibu ed o se e e pola iza ion o social
a i udes, ueled by p o-Eu opean and p o-Russian poli ical mo emen s.
E en as he majo i y suppo ed he democ a ic de elopmen pa h, pub-
lic co up ion and a lack o ins i u ions led o se e al igged elec ions
and slow democ a ic e o ms. A numbe o peace ul e olu ions in
bo h Geo gia and Uk aine eme ged as a emp s o mo e owa d mo e
democ a ic egimes. The iolen con lic s, suppo ed by Russia, ol-
lowed hose e olu ions, imposing signi ican economic cos s. Di icul
ansi ion o democ acy has hus been complica ed by se e e economic
bu dens and impo e ishmen . Acco ding o heo ies on e hnic con lic s
(see, o example, Es eban and Ray (2011b)), pola iza ion in hese
coun ies should ha e u he inc eased ollowing he con lic s. Ne -
e heless, as a simple model p esen ed in he p e ious sec ion sugges s,
he ela ionship be ween con lic s and poli ical pola iza ion can be
nega i e i he whole socie y iews he con lic as a h ea , o in o he
wo ds, i social g oups cha ac e ized by di e en poli ical p e e ences
ha e simila p e e ences ega ding he con lic e mina ion.
The a ia ion in he con lic da es and egions in con lic , combined
wi h he egional da a on poli ical elec ions in Geo gia and Uk aine,
is used o explo e he ela ionship be ween con lic s and poli ical
alignmen ac oss di e en egions. The da a a e desc ibed in mo e
de ail below.
3.2. The da a
I collec egional da a on he esul s o poli ical elec ions, including
he p esiden ial and pa liamen a y elec ions, o Geo gia and Uk aine.
The da a sou ces a e he o icial websi es o he elec ion commissions,
he news, and Cons i uency-Le el Elec ions A chi e (CLEA) da ase
(Kollman e al.,2019).10 Table 4 in he Appendix epo s he egions
10 Fo Geo gia, he da a ha e been collec ed om www.elec o algeog aphy
.com and CLEA da ase ; o Uk aine, om h ps://www.c k.go .ua/.
and he elec ions included in he analysis ( o hose elec ions o
which he egional da a a e a ailable), he da es o iolen con lic s
as epo ed by he UCDP, and he classi ica ion o egions as being
noncon iguous o a con lic zone, con iguous o a con lic zone, and
in a con lic zone.11
Fo each o he elec ions, I conside he egional sha e o o es cas
o he coun y-wide elec ion winne (whe e he elec ion winne e e s
o he poli ical pa y o he candida e o p esiden ). Fo pa liamen a y
elec ions, he winning pa y is he pa y ha secu es he la ges sha e
o o es ac oss he coun y, and his numbe can be below 50%. The e
is signi ican he e ogenei y in he numbe o compe ing pa ies by
elec ion yea and by coun y, wi h a ound eigh pa ies compe ing pe
elec ions, on a e age. Fo p esiden ial elec ions, he winning candida e
mus ob ain mo e han 50% o he o es. Unless he leading candida e
wins in he i s ound, I use he da a on second ounds, which include
only wo candida es.
The e is a signi ican a ia ion in he egional sha e o o es cas
o he pa y ha won he elec ions by coun y and o e ime. In
pa icula , he egional suppo o he winning pa y in Geo gia a ies
om 16.81% o 94.21% and in Uk aine om 1.94% o 96.0%. Highe
a ia ion in he winne ’s suppo ac oss egions e lec s g ea e poli ical
pola iza ion ac oss egions.
Fo mally, I compu e egional poli ical pola iza ion as he squa e o
he di e ence be ween he egional and he coun y-a e age sha e o
o es cas o he elec ion winne , no malized by he coun y-a e age
sha e. Speci ically, 𝑃 𝑜𝑙𝑗𝑖𝑡, pola iza ion in coun y 𝑗, egion 𝑖, and yea
𝑡is gi en by:
𝑃 𝑜𝑙𝑗𝑖𝑡 =(𝑉 𝑆𝑗𝑖𝑡 −𝑉 𝑆𝑗𝑡
𝑉 𝑆𝑗𝑡 )2
,(11)
whe e 𝑉 𝑆𝑗𝑖𝑡 deno es he o e sha e ob ained by he winne in egion
𝑖coun y 𝑗a ime 𝑡and 𝑉 𝑆𝑗𝑡 deno es he o e sha e ob ained by he
winne h oughou coun y 𝑗.
Fig. 3 epo s he compu ed pola iza ion measu es, as a e ages
ac oss egions, o he coun ies unde conside a ion. The compu ed
11 The classi ica ion o egions by con igui y is based on go e nmen-
al da a (h p://shidaka li.go .ge/ge/pages/index/47 and h ps://m a.go .ge/
Occupied-Te i o ies/Law.aspx o Geo gia).

Economic Modelling 121 (2023) 106231
6
D. G echyna
Fig. 3. Poli ical pola iza ion in suppo o winne , based on elec ions da a om
Geo gia and Uk aine.
No e: The igu e depic s he compu ed pola iza ion measu es based on he a e ages
ac oss egions o Geo gia and Uk aine, in squa e- and ci cle-ma ked lines, espec i ely;
g ay-shaded a eas ep esen iolen con lic s.
pola iza ion a ies signi ican ly by coun y and wi hin coun y o e
ime, e lec ing he poli ical ins abili y ha cha ac e ized Geo gia and
Uk aine du ing he pos -communis ansi ion.
Fig. 4 summa izes he a e age pola iza ion le els o he egions
con iguous and non-con iguous o con lic zones, as well as egions
in a con lic zone, du ing peace ime and du ing imes cha ac e ized
by iolen con lic in he las ou yea s.12 Fo egions in a con lic
zone, da a a ailabili y is limi ed and based on pa s o egions which
a e no comple ely occupied and whe e elec ions could be held. The
egions con iguous o a con lic zone a e mo e pola ized, in he sense
ha poli ical suppo o he coun y-wide suppo ed poli ical pa ies
in hese egions is u he away om he a e age compa ed o he
egions non-con iguous o a con lic zone. In Uk aine, he pola iza ion
in all ypes o egions declines du ing con lic s compa ed o peace ime,
whe eas in Geo gia i inc eases. This co ela ion is u he explo ed
below in he empi ical model ha con ols o ime and egion ixed
e ec s.
12 Since elec ion yea s o which he pola iza ion could be compu ed and he
yea s o con lic s do no coincide in gene al, imes o con lic a e conside ed
as elec ion yea s o which some con lic occu ed du ing he ou yea s p io
o elec ions.
3.3. Empi ical analysis and esul s
I e alua e he ela ionship be ween iolen con lic s and poli ical
pola iza ion in egions o Geo gia and Uk aine using he ollowing
model:
𝑃 𝑜𝑙𝑗𝑖𝑡 =𝑎+𝛽𝐶𝑗𝑡 +𝜇𝑖+𝑇𝑡+𝜖𝑗𝑖𝑡,(12)
whe e 𝑃 𝑜𝑙𝑖𝑗𝑡 is pola iza ion in coun y 𝑗, egion 𝑖, and yea 𝑡;𝐶𝑗𝑡 is a
measu e o con lic in coun y 𝑗yea 𝑡;𝜇𝑖is egion ixed e ec ; 𝑇𝑡is
ime (yea and mon h) ixed e ec s; and 𝜖𝑗𝑖𝑡 is he e o e m. Model
(12) is es ima ed by OLS wi h s anda d e o s clus e ed by egion.
A simple bina y indica o o yea s o con lic is no use ul in his
se up, because gaps be ween he elec ion yea s and be ween con lic
yea s a e unequal. The e o e, as a i s measu e o con lic I use a bina y
indica o aking a alue o one i he coun y expe ienced a iolen
con lic epo ed by he UCDP du ing (any o ) he ou yea s p io o
elec ions. The ou -yea in e al is used because elec ions ook place
e e y ou yea s, on a e age.
A limi a ion o he bina y indica o o con lic is ha i does no
ake in o accoun he du a ion o he con lic . The e o e, he second
measu e o con lic , he numbe o con lic yea s, is compu ed as he
o al numbe o con lic yea s since he collapse o he So ie Union.
The summa y s a is ics a e epo ed in Table 3 in Appendix.
Columns (1) and (3) o Table 1 epo he baseline es ima ion
esul s, con olling o ime and egion ixed e ec s.13 The coe icien
on he bina y indica o o iolen con lic is nega i e and signi ican ,
sugges ing ha a iolen con lic in he ou yea s p io o elec ions
is ollowed by a dec ease o app oxima ely one-qua e o a s anda d
de ia ion in egional pola iza ion in he elec ion winne ’s suppo .
Simila ly, he coe icien on he numbe o con lic yea s is nega i e
and signi ican , indica ing ha e e y ou yea s o con lic a e associ-
a ed wi h a educ ion in egional pola iza ion in he elec ion winne ’s
suppo o app oxima ely one- hi d o a s anda d de ia ion.
The egions con iguous o a con lic zone a e mo e pola ized (see
Fig. 4) and can be di e en ly a ec ed by a con lic compa ed o he
o he egions. The e o e, I u he e alua e he impac o con lic on
pola iza ion condi ional on he ype o egions, dis inguishing be ween
he egions con iguous o a con lic zone and egions in a con lic zone,
13 No e ha he es ima ion esul s based on a bina y indica o o con lic
in he p eceding ou yea s (Columns (1)–(2) o Table 1) ha e almos wice as
many obse a ions as he es ima ion esul s based on he numbe o con lic
yea s (Columns (3)–(4) o Table 1). This is because he la e es ima ion
includes only he elec ions ha occu ed a e a posi i e numbe o con lic s
while he o me es ima ion includes all he yea s, be o e and a e he
con lic s.
Fig. 4. The a e age poli ical pola iza ion by con igui y o con lic zone, based on elec ions da a om Geo gia and Uk aine.
No e: The igu e depic s he a e age pola iza ion le els o he egions non-con iguous o a con lic zone, in ligh g ay, con iguous o a con lic zone, in g ay, and egions in a
con lic zone, in black, du ing peace ime and a e iolen con lic s, o Geo gia and Uk aine in he le and igh panel, espec i ely.
Economic Modelling 121 (2023) 106231
7
D. G echyna
Table 1
Con lic s and he di e gence o poli ical p e e ences: Es ima es based on egional da a o Geo gia and Uk aine.
Va iables (1) (2) (3) (4)
Di e gence in egional suppo o elec ion winne
Con lic in he las ou yea s −0.0839*** −0.0557*
(0.0313) (0.0294)
Numbe o con lic yea s −0.0282*** −0.00987***
(0.00662) (0.00307)
Con lic in he las ou yea s ×0.0809**
egions con iguous o con lic zone (0.0350)
Con lic in he las ou yea s ×0.00802
egions in con lic zone (0.0544)
Numbe o con lic yea s × −0.00415
egions con iguous o con lic zone (0.00787)
Numbe o con lic yea s × −0.0410***
egions in con lic zone (0.00717)
Regions con iguous o con lic zone −0.101** 0.0186
(0.0509) (0.0629)
Regions in con lic zone −0.217 0.595***
(0.187) (0.0616)
Pa liamen a y elec ions 0.0381 0.0824*** −0.0935*** −0.0241***
(0.0395) (0.0204) (0.0175) (0.00727)
Cons an 0.0395 0.158*** 0.214*** 0.0927***
(0.0335) (0.00735) (0.0276) (0.0189)
Obse a ions 791 791 494 494
R-squa ed 0.292 0.042 0.241 0.119
Numbe o egions 105 105 104 104
No e: This able epo s he OLS es ima es o Model (12) con olling o ime and egion ixed e ec s; obus s anda d e o s
clus e ed by egion.
*Deno e signi icance a 10% signi icance le el.
**Deno e signi icance a 5% signi icance le el.
***Deno e signi icance a 1% signi icance le el.
Table 2
Con lic s and pola iza ion: es ima es based on coun y-le el da a.
Va iables (1) (2) (3) (4) (5) (6)
Poli . pola iza ion P o-le P o- igh
Con lic −0.0310** −0.0293** 0.0158
(0.0140) (0.0128) (0.0302)
Real GDP pe capi a −0.0161 −0.0242 −0.0140 −0.0172 0.0292 −0.0136
(0.0223) (0.0200) (0.0213) (0.0206) (0.0220) (0.0329)
Public expendi u es 0.00194 0.00240 0.00209 0.00213 0.00328*0.00223
(0.00199) (0.00204) (0.00202) (0.00206) (0.00192) (0.00257)
Democ acy index 0.0208 0.0292 0.0219 0.0245 0.0154 0.0456
(0.0231) (0.0227) (0.0239) (0.0232) (0.0153) (0.0587)
In e s a e con lic −0.0949***
(0.0139)
In as a e con lic −0.0215
(0.0249)
In e na ionalized −0.00577
in as a e con lic (0.0226)
Cons an 0.454** 0.509*** 0.432** 0.454** −0.220 0.340
(0.193) (0.175) (0.185) (0.183) (0.192) (0.276)
Obse a ions 236 236 236 236 236 236
R-squa ed 0.243 0.249 0.228 0.220 0.368 0.288
N coun ies 89 89 89 89 89 89
No e: This able epo s he OLS es ima es o Model (13) wi h yea and coun y ixed e ec s, o he dependen a iable pola iza ion, in Columns
(1)–(4), and he ac ion o p o-le and p o- igh indi iduals, in Columns (5) and (6), espec i ely; obus s anda d e o s clus e ed by coun y.
*Deno e signi icance a 10% signi icance le el.
**Deno e signi icance a 5% signi icance le el.
***Deno e signi icance a 1% signi icance le el.
Economic Modelling 121 (2023) 106231
8
D. G echyna
wi h he o he egions being a baseline ca ego y.14 The es ima ion o
Model (12) wi h addi ional bina y indica o s o egions con iguous o
a con lic zone o in a con lic zone, and hei in e ac ion wi h he
indica o o con lic , is epo ed in Columns (2) and (4) o Table 1.
The es ima ion esul s sugges ha pola iza ion in egions ha become
con iguous o a con lic zone is no s a is ically di e en om ha
in he o he egions, bu i inc eases when he con lic akes place.
Conside ing only con lic yea s, wi h co esponding es ima ion esul s
epo ed in Column (4) o Table 2, egions in a con lic zone a e
signi ican ly mo e poli ically pola ized compa ed o o he egions, bu
as con lic du a ion inc eases, pola iza ion in hese egions declines.
These esul s should be in e p e ed wi h cau ion because he da a on
egions in a con lic zone a e incomple e, and a signi ican ac ion o
he popula ion could be displaced du ing he pe iod conside ed in his
s udy.
The indings o his sec ion p o ide empi ical e idence in suppo o
he heo e ical a gumen p esen ed in Sec ion 2. Violen con lic s in he
o m o e i o ial dispu es suppo ed by o eign s a es a e associa ed
wi h a educ ion in mass pola iza ion.
4. Con lic s and poli ical pola iza ion a ound he wo ld
This sec ion con i ms he esul s ob ained in he p e ious sec-
ion in a mo e gene al se up, based on a panel o coun ies a ound
he wo ld and pola iza ion measu es compu ed om indi idual sel -
epo ed poli ical p e e ences. While such da a a e much mo e he e o-
geneous, explo ing hem se es as a obus ness check and pe mi s some
gene aliza ion o he esul s ob ained in he p e ious sec ion.
4.1. The da a
I use he coun y-speci ic da a on iolen con lic s om he UCDP
da ase (Ha bom e al.,2008;Pe e sson e al.,2021), he wo ld’s main
p o ide o da a on o ganized iolence. The UCDP de ines con lic yea s
in a gi en coun y as yea s ha ha e c ossed he 25 ba le- ela ed
dea hs h eshold. Fo he measu e o p e e ences pola iza ion, I use
he indi idual esponses om he di e en wa es o WVS (Ingleha
e al.,2020), co e ing he pe iod 1981–2020. The indi idual esponses
con ain hei sel -iden i ied posi ion on he poli ical scale, based on he
ollowing s a emen :
(S) ‘‘Sel posi ioning in poli ical scale’’ In poli ical ma e s, people
alk o ‘‘ he le ’’ and ‘‘ he igh ’’. How would you place you iews
on his scale, gene ally speaking (choose a numbe be ween 1 and
10)? Le (1) 2 3 4 5 6 7 8 9 Righ (10).
Fo each coun y-yea a ailable in he su ey, I compu e he co-
e icien o a ia ion o indi idual esponses o his s a emen ( he
s anda d de ia ion o indi idual esponses di ided by he mean alue
o esponses), ollowing he me hodology used o measu e p e e ences
pola iza ion by Lindq is and Ös ling (2010). The la ge he a ia ion,
he g ea e he disag eemen in socie y.
These da a, combined wi h he da a on iolen con lic s, yield
an unbalanced panel o 89 coun ies o e he pe iod 1981–2020. As
con ol ac o s, I conside he da a on he eal GDP pe capi a (in
loga i hms) and he go e nmen expendi u es om he Wo ld Bank,
because hose a iables a e obus de e minan s o pola iza ion (see
G echyna (2016)), as well as a ime- a ying index o democ acy by
G ündle and K iege (2021). Table 3 p o ides he summa y s a is ics
o all he a iables.
14 The con igui y o a con lic zone is a ime- a ying indica o o some
egions; see Table 4 o da es when a pa icula egion becomes con iguous
o a con lic zone o loca ed in a con lic zone.
4.2. Empi ical analysis and esul s
I e alua e he ela ionship be ween pola iza ion and con lic using
a model simila he one used in he p e ious sec ion bu based on
coun y-le el a he han egional da a. I is speci ied as ollows:
𝑃 𝑜𝑙𝑗𝑡 =𝑎+𝛽𝐶𝑗𝑡 +𝛾𝐗𝑗𝑡 +𝜇𝑗+𝑇𝑡+𝜖𝑗𝑡,(13)
whe e 𝑃 𝑜𝑙𝑗𝑡 is he pola iza ion in poli ical a i udes in coun y 𝑗yea 𝑡;
𝐶𝑗𝑡 is a bina y indica o aking a alue o one o he yea s o con lic
in coun y 𝑗yea 𝑡, and ze o o he wise; 𝐗𝑗𝑡 a e he con ols, he eal
GDP pe capi a (in loga i hms), he go e nmen spending as a sha e o
GDP, and he democ acy index; 𝜇𝑗deno es coun y ixed e ec s and 𝑇𝑡
deno es yea ixed e ec s; 𝜖𝑗𝑡 is he e o e m. Model (13) is es ima ed
by OLS wi h s anda d e o s clus e ed by coun y.
Columns (1)–(4) o Table 2 epo he esul s. Violen con lic s
a e nega i ely associa ed wi h poli ical pola iza ion measu ed as he
a ia ion in indi idual sel -posi ioning on he poli ical scale, con olling
o coun y and ime ixed e ec s. Speci ically, a iolen con lic is as-
socia ed wi h a dec ease o app oxima ely 40% o a s anda d de ia ion
in pola iza ion, all else being equal.
The coun ies and con lic ypes included in he sample a e e y
he e ogeneous. The e o e, I e-es ima e Model (13) including bina y
indica o s o di e en con lic ypes, dis inguishing among he in e -
s a e, in as a e, and in e na ionalized in as a e con lic s.15 The esul s
indica e ha in e s a e con lic s a e signi ican ly nega i ely co ela ed
wi h poli ical pola iza ion. An in e s a e con lic is associa ed wi h
a ound a one-s anda d-de ia ion dec ease in pola iza ion. A he same
ime, he in as a e con lic s do no exhibi any signi ican co ela ion
wi h pola iza ion, con olling o coun y and ime ixed e ec s. These
esul s a e in ui i e. The in as a e con lic s a e mo e likely o be d i en
by e hnic di e ences and exis ing social di isions and, he e o e, a e
no likely o esul in he educ ion o pola iza ion. The in e s a e
con lic s impose a common cos on all social g oups in he coun y and
he e o e, can ‘‘uni e’’ he na ion, simila o he esul s ob ained in he
p e ious sec ion based on he egional da a.
The WVS da a used o cons uc he dependen a iable, poli ical po-
la iza ion, can gi e some addi ional insigh s in o he shi s in poli ical
p e e ences ollowing a con lic . I hus compu e he sha e o indi iduals
who sel -posi ion hemsel es as p o-le ( hose who answe 1, 2, o 3
o ques ion (S)) o p o- igh ( hose who answe 8, 9, o 10 o ques ion
(S)). The esul s o Model (13) es ima ed wi h he sha e o p o-le
o p o- igh as he dependen a iable a e p esen ed in Columns (5)
and (6) o Table 2, espec i ely. The sha e o p o-le esponden s is
signi ican ly nega i ely associa ed wi h iolen con lic ; he coe icien
on he sha e o p o- igh is posi i e bu no s a is ically signi ican .
While a comp ehensi e analysis o he shi s in poli ical p e e ences is
beyond he scope o his pape , hese shi s could be ela ed o public
policies du ing he con lic s and o he impac o public policies on he
p obabili y o success ul con lic e mina ion, as desc ibed by he model
p esen ed in Sec ion 2.
5. Conclusions
Violen con lic s ha e a de as a ing impac on o dina y people’s
li es. This pape p o ides heo e ical and empi ical a gumen s suppo -
ing he claim ha people may ‘‘uni e’’ in hei poli ical p e e ences
du ing iolen con lic s. Fi s , he model de eloped in his pape shows
ha a con lic which is cos ly o all social g oups can educe social
pola iza ion by aligning poli ical p e e ences in socie y. Socie al p e -
e ences shi owa d he public policies ha inc ease he p obabili y
o success ul con lic e mina ion. Second, he empi ical analysis o
15 The e a e no da a o he las ype, ex asys emic con lic s, o he pe iod
conside ed in he analysis.
Economic Modelling 121 (2023) 106231
9
D. G echyna
Table 3
Summa y s a is ics.
Va iables (1) (2) (3) (4) (5)
N Mean SD Min Max
Summa y s a is ics o Geo gia and Uk aine egional da a
Poli ical winne ’s suppo pola iza ion 791 0.170 0.340 2.99e−07 4.689
Winne ’s egional sha e o o es 791 50.32 22.62 1.940 96.20
Winne ’s coun y-wide sha e o o es 791 51.06 13.89 22.14 73
Con lic in he las ou yea s 791 0.372 0.484 0 1
Numbe o con lic yea s 494 4.032 1.288 1 6
Pa liamen a y elec ions 791 0.540 0.499 0 1
P esiden ial elec ions 791 0.427 0.495 0 1
Regions no con iguous o con lic zone 791 0.864 0.344 0 1
Regions con iguous o con lic zone 791 0.104 0.305 0 1
Regions in con lic zone 791 0.033 0.178 0 1
Summa y s a is ics o wo ldwide coun y-le el da a
Poli ical a i udes pola iza ion 236 0.401 0.080 0.159 0.900
Con lic 236 0.195 0.397 0 1
Con lic s in ensi y 236 0.848 1.525 0 4
In e s a e con lic 236 0.013 0.112 0 1
In as a e con lic 236 0.153 0.360 0 1
In e na ionalized in as a e con lic 236 0.030 0.170 0 1
Real GDP pe capi a 236 8.888 1.302 6.130 11.406
Public expendi u es 236 15.247 5.092 1.315 27.518
Democ acy index 236 0.793 0.286 0.003 1
P o-le sha e 236 0.137 0.074 0.004 0.665
P o- igh sha e 236 0.189 0.109 0.021 0.819
No e: This able epo s he summa y s a is ics o egional panel da a o Geo gia and Uk aine co e ing yea s 1991–2019 and 105 egions,
27 in Uk aine and 78 in Geo gia, used o he es ima ions epo ed in Table 1, in he op panel; and he summa y s a is ics based on he
coun y-le el da a co e ing yea s 1981–2020 and 89 coun ies, used o he es ima ions epo ed in Table 2, in he bo om panel.
Table 4
Geo gia and Uk aine: Regions and poli ical elec ions analyzed.
Da a sou ces: CLEA elec ions da abase, go e nmen websi es (elec ions and occupied egions); UCDP da abase (con lic s).
Geo gia Uk aine
Elec ions P esiden ial: 5 Jan 2008,
27 Oc 2013
Pa liamen a y: 28 Ma 2004,
21 May 2008, 1 Oc 2012,
8 Oc 2016
P esiden ial: 1 Dec 1991,
26 Jun 1994, 31 Oc 1999,
31 Oc 2004, 26 Dec 2004,
17 Jan 2010, 25 May 2014,
31 Ma 2019
Pa liamen a y: 29 Ma 1998,
31 Ma 2002, 26 Ma 2006,
30 Sep 2007, 28 Oc 2012,
26 Oc 2014
Con lic s Dec 1991–No 1993
Aug 2004–Oc 2004
Aug 2008–No 2008
since Feb 2014
(con inued on nex page)
egional da a om Geo gia and Uk aine, coun ies ha ecen ly expe i-
enced se e al iolen con lic s, sugges s ha he di e ences in poli ical
p e e ences ac oss egions a e nega i ely associa ed wi h iolen con-
lic s suppo ed by a o eign s a e. Finally, he nega i e associa ion
be ween iolen con lic s and poli ical pola iza ion is con i med us-
ing a panel o coun ies a ound he wo ld and pola iza ion measu es
compu ed om indi idual sel - epo ed poli ical p e e ences.
The esul s p esen ed in his pape open up se e al ques ions o
u he esea ch. Fi s , he empi ical es ima es p esen ed p o ide de-
sc ip i e a he han causal e idence on he ela ionship be ween i-
olen con lic and poli ical pola iza ion. The e o e, he analysis o he
causal impac , i any, o con lic s on poli ical pola iza ion is an a enue
o u he esea ch. The heo e ical model de eloped in his pape
can be ex ended o analyze he na u e o poli ical changes ollow-
ing a con lic and he impac o a ious ypes o public policies on
con lic du a ion and poli ical pola iza ion. In pa icula , he model
sugges s ha indi idual poli ical p e e ences shi owa d mo e de ense-
o ien ed, o igh -wing public policies, ollowing a iolen con lic .
The empi ical analysis o wo ldwide da a on pola iza ion o poli ical
a i udes con i ms ha he poli ical p e e ences o he majo i y shi
away om le -wing public policies ollowing a iolen con lic . Fu he
esea ch is needed o e alua e he condi ions unde which he decline in
pola iza ion o p e e ences is pe sis en and leads o posi i e economic
changes.
P e ious li e a u e has widely discussed how iolen con lic s can
lead o social di ides and poli ical ins abili y. The indings o his
pape sugges ha a public policy agenda aimed a s eng hening he
coun y’s esilience o o eign h ea s can e e se he impac s o iolen
con lic s om di iding o uni ing he socie y.
Decla a ion o compe ing in e es
The au ho s decla e ha hey ha e no known compe ing inan-
cial in e es s o pe sonal ela ionships ha could ha e appea ed o
in luence he wo k epo ed in his pape .
Da a a ailabili y
Da a will be made a ailable on eques .